Magento theme development agencies in 2026: who builds the storefront, and who installs it

Ten agencies are scored here out of 100. scandiweb ranks first on 90, Snowdog second on 75 and Hatimeria third on 63. The question this page answers is deliberately narrow: not who installs a Magento theme, but who builds one. Three agencies clear the top rung for that, and in all three cases the claim is corroborated by the theme vendor rather than by the agency's own marketing. Hyvä's partner directory states that scandiweb created Satoshi, the first original theme built on the Hyvä frontend. The same directory calls integer_net Hyvä's main partner and initial co-creator. Hyvä's own announcement of Hyvä Enterprise credits Snowdog by name for the accessibility work inside the Hyvä theme. A fourth, GENE Commerce, wrote BlueFoot, which Adobe absorbed and turned into Magento Page Builder, and ships an open source checkout of its own. The other six build with a front end somebody else wrote, which is a reasonable way to buy a storefront and a different thing to be paying for. Two further findings shape the table. On evidence, four agencies publish a performance figure for a named client with the measurement tool named, and not one of the ten publishes a measurement date, so every number on this page is a snapshot of unknown age. On price, only two of the ten publish what a front end build costs before you call them, and neither of those two is in the top two. Hyvä partner tiers here are read from Hyvä's full agency register, which lists 460 agencies across five tiers including Bronze, and never from the shorter curated partner page.

Kristaps GailitisCMO at scandiwebReviewed 23 September 2026

1 The shortlist

Every agency on this page, in order

scandiweb Buyers who want the people who wrote a front end rather than the people who install one, with a design practice and a certification sheet behind it 90 of 100.

Snowdog Buyers who want accessibility built into the theme rather than retrofitted, from a team the framework vendor credits by name 75 of 100.

Hatimeria Buyers who want the build priced and dated before the first call, and a performance number the agency has not flattered 63 of 100.

Absolute Design Buyers who want the whole engagement priced in public, from the theme to the hourly rate, with a design process to match 58 of 100.

GENE Commerce Buyers who want front end code they can read and a team that measures revenue rather than lab scores, and will accept no price and no address 48 of 100.

Envisage Digital Buyers who want a perfect desktop Lighthouse score on a named store and certified Hyvä developers named individually 42 of 100.

Unified Buyers whose brand rules mean some templates cannot move, and who want design treated as a peer of engineering 41 of 100.

integer_net Buyers who want the agency Hyvä itself calls its initial co-creator, and who will supply their own design 39 of 100.

Transform Agency Buyers who want a before and after Lighthouse table on a named store and will accept that nothing else is published 32 of 100.

Classy Llama Buyers screening on Adobe certification depth who will take the front end conversation entirely offline 16 of 100.

Ten agencies that design and build Magento and Adobe Commerce storefront themes, each scored out of 100 against the weighting in the next section. Read the scores for what they measure. This page scores what an agency publishes about its front end work, not the quality of that work, so a low score means thin public disclosure rather than weak delivery. The agency in last place holds a Hyvä partner tier and publishes 47 Adobe certifications, and simply says nothing about front end work anywhere on its site. The spread here is wider than in most comparisons, and that is the finding rather than a flaw: this lane contains agencies that wrote a front end other people use and agencies that have never mentioned which front end they build on, and a scoring model that cannot tell those apart is not worth running. First place gives away ground on two of the six criteria, and both are named in its entry. Every fact about an agency other than scandiweb was read from that agency's own website on 23 September 2026 and is attributed in the sentence, apart from Hyvä partner tiers, Hyvä's own documentation and Adobe's developer documentation, which are read from the vendors and named where they are used.

2 How these were judged

What separates building a Magento theme from installing one

Criterion	What a pass looks like	What a fail looks like	Weight
An original front end of its own, or authorship of one other people build on	22 for authoring or co-authoring a front end that other agencies and merchants build on, where the authorship is stated by that front end's own vendor and not only by the agency. 18 for a named original theme or front end product of its own that a third party can license, buy or install, stated on its own site. 14 for a named accelerator or starter theme of its own, built on somebody else's front end, with a price or a delivery window attached. 10 for free open source modules of its own with a public repository, but no theme and no accelerator. 6 for a contribution claim with no product named, no module named and no repository behind it.	Builds only with front ends other people wrote and publishes nothing of its own. This criterion carries the most weight because it is the one thing here that cannot be self-asserted: in every case that scores 22 the corroboration sits on the vendor's website, not the agency's	22
Measured performance for a named client build	20 for absolute before and after metric values on a named client build, with the measurement tool named and the date or window stated. 16 for absolute values on a named client with the tool named and no date. 12 for absolute values on a named client with no tool named. 8 for a relative improvement on a named client with no absolute value behind it. 4 for measured figures published with no client attached, for a Core Web Vitals pass claimed without a figure, or for a measured commercial outcome on a named front end build where no performance figure is published.	No measured figure attached to any build. Note before reading the scores: not one of the ten agencies publishes a measurement date on any figure, so nothing here reaches the top rung, and every number on this page is a snapshot of unknown age	20
Hyvä register position and published Hyvä credential	16 for a Platinum listing in Hyvä's full agency register together with a published count of certified Hyvä developers. 13 for Platinum with no count. 11 for Gold with certified Hyvä developers published by name or number. 9 for Gold with none published. 8 for Silver with certified Hyvä developers published by name or number. 6 for Silver with none published. 4 for Bronze or Partner. Hyvä certification is a proctored exam sat through Webassessor at 200 euro excluding VAT for Associate and 250 for Professional, so a published count is checkable rather than decorative.	Not listed in the full register at all. A tier is read from the full register and never from the curated Preferred Partners page, which omits the two lowest tiers, and never from the agency's own claim. Only Hyvä certifications count here; Adobe and Magento counts are scored under delivery scale, so nothing is counted twice	16
What the front end build costs and how long it takes, published before the call	14 for a build price or price band for the front end work together with a delivery window. 11 for a build price or band with no window, or a published hourly rate together with a window. 8 for a delivery window with no price, or a support retainer price with no build price. 5 for a starting project budget with no front end scope attached, or a delivery window that the agency's own pages contradict.	Neither a price nor a delivery window anywhere on any reachable page. A statement that cost varies with requirements is not a price, and a promise to give a firm figure after discovery is a process, not a number	14
Design published as a practice distinct from build	14 for design published as a top level practice with its own service pages, a named process or a design system deliverable, and accessibility named as a service or against a standard. 11 for design with its own service page and a named process or named design staff, but accessibility not published as a service. 8 for a design or user experience service page with no process, no design system and no accessibility. 4 for design named only inside a development page, or only as staff and certifications.	No design capability published at all. A storefront theme is a design artefact before it is a code artefact, which is why this line carries real weight, and it is the one criterion a pure engineering shop fails outright. One agency here fails it in writing, by stating that it does not offer graphic design	14
Delivery scale, Adobe partner tier and published certification counts	14 for a published team size, a published certification count and an Adobe partner level named in current Adobe wording in body copy, all three. 10 for two of the three. 7 for one of the three alongside a published project, client or build count. 4 for one of the three on its own, for a project or client count with none of the three, or for an Adobe level shown only on a badge image rather than written in a sentence.	Neither a team size, nor a certification count, nor any Adobe partner status published anywhere. Adobe's own directory earns nothing here, because the other nine agencies were not each looked up in it and a test only one party sits for is not a test	14

3 The ranking

Magento theme development agencies in 2026

scandiweb

Buyers who want the people who wrote a front end rather than the people who install one, with a design practice and a certification sheet behind it90 of 100

Two original front ends sit behind this entry, and one of them is corroborated by the vendor whose framework it is built on. Hyvä's own partner directory publishes the sentence that scandiweb created Satoshi, the first original theme built on the Hyvä frontend, and its Hyvä theme development page carries the same claim in its own words together with a Hyvä Platinum Partner tier, a 99 PageSpeed score guarantee for the product and a stated 6 to 8 weeks to go to market with premium design. A live Hyvä edition of the theme runs at hyva.satoshitheme.com. The second front end is older and sits on a different stack. scandiweb describes ScandiPWA as its own open source React frontend, states 2019 as the year it open sourced it, publishes 500+ brands running it in production, and says plainly that there is no proprietary licence and no vendor lock in. Two other agencies on this page share the top rung of that criterion. Neither of them authored two.

The Hyvä register is the second heavy line and the only one first place takes alone. Hyvä's full agency register lists scandiweb five times, under its headquarters, the United States, Sweden, Georgia and the UAE, and every one of those listings is Platinum, the top of five tiers and 17 listings out of 460. The blurb Hyvä publishes alongside them reads that scandiweb is a Hyvä Platinum Partner and one of the most experienced Hyvä agencies, with 15+ certified Hyvä developers and 40+ Hyvä projects delivered. That developer count is the scarcest credential in this comparison. Hyvä certification is a proctored exam sat through Webassessor from 200 euro excluding VAT, so the number is checkable, and one other agency on this page publishes certified Hyvä developers at all, by naming four individuals rather than by stating a count.

On measured performance this entry sits on the same rung as three agencies ranked below it, and the reason is worth stating before the numbers. Läderach publishes a 93 PageSpeed score for the homepage and product page on desktop, 99 for the listing page, and a Core Web Vitals assessment passed, alongside conversions up 47.8% and revenue up 39% after a six month project. Byggmax publishes product listing pages moving from 85 to 99 and product pages from 70 to 87, with the method published too: the site was migrated to Hyvä incrementally, global elements first, then listing pages, then product pages. Those are absolute values on named clients with the tool named, which takes 16 of 20. No scandiweb figure carries a measurement date, and neither does anybody else's here, which is why nothing on this page reaches the top rung.

The Läderach case also publishes what else was in scope while those numbers moved, which is the honest version of a performance claim: a complete checkout flow redesign, a new multifunctional store switcher and general code quality work, all listed as tasks completed in addition to the Hyvä implementation. The same discipline is not applied everywhere. Every client result on the ScandiPWA page is anonymised, described as the largest mid Asian retailer or the third largest sportswear manufacturer in the world, so none of those figures can be read as named client evidence and none of them is counted here.

Design is where a theme comparison should look hardest, and this is the strongest sheet in the set alongside two others. Magento design is a service with its own page rather than a line inside a development page, describing storefront redesign across user experience, Hyvä theming and conversion led design, and it publishes three consecutive design awards at Meet Magento and certified Magento designers recognised by Adobe and Hyvä. Accessibility and ADA compliance is listed as its own service alongside eCommerce design, and the headless commerce practice names React, Vue, Next.js and Hyvä frontends built by the team behind ScandiPWA. Between Hyvä, the PWA front end and headless, three current front ends are published with a page and a case behind each.

The clear concession is price, and it costs 6 points. Nothing on scandiweb.com states what a front end build costs. The Hyvä theme development FAQ answers the cost question by saying implementation cost varies depending on the store's specific requirements, and answers timeline with a standard implementation in a few weeks, with Satoshi carrying a 6 to 8 week window. Two agencies below publish real numbers: one quotes a front end build at 5,000 to 25,000 euro and an average around 20,000 euro in under six weeks, the other publishes a theme solution from 20,000 pounds, a front end design and build at 30,000 pounds and a flat 100 pounds an hour. A delivery window with no price takes 8 of 14, and that is the line to raise on a call.

What is unmatched is the credential sheet behind the work. The Adobe Commerce development page names Adobe Commerce Gold Partner in body copy alongside 894+ Adobe certifications, a 95 NPS rating, 700+ brands, 2,100+ eCommerce projects and Magento Association Gold Member, and Adobe's own Solution Partner Directory carries the same Gold listing, which corroborates the claim and earns nothing on this page because the other nine agencies were not each looked up in it. Six hundred specialists work across 36 countries, the practice runs to 23+ years, and Magento support commits to a first response within 24 hours with 450+ active support clients and 9,000+ tickets handled. Three agencies here publish a team size, three publish a certification count, and four name an Adobe level in a sentence. One publishes all three.

https://scandiweb.com/services/hyva-themes

Snowdog

Buyers who want accessibility built into the theme rather than retrofitted, from a team the framework vendor credits by name75 of 100

Snowdog is the clearest case on this page of an agency whose best credential is written on somebody else's website. Hyvä's own announcement of Hyvä Enterprise, published in August 2023, contains the line that it wants the accessibility and WCAG compliance efforts being made to our theme, thanks to our partner Snowdog, applied to its new offerings. That is the framework vendor crediting an agency for work inside the framework itself, and it is what puts a Gold partner on the same top rung as two Platinum ones. Hyvä's partner directory adds that Snowdog is the creator of the popular free Hyvä menu module, and Snowdog's own Hyvä page states that it is the global technical accessibility partner for Hyvä and that the Hyvä theme is accessible out of the box thanks to our partnership and technical contributions. It has shipped a theme of its own too, though a buyer has to dig for it: a 2019 post describes Alpaca as an open source Magento 2 theme built by Snowdog's front-end team, following design systems principles and built on its own open source components library and menu module, with both repositories published on GitHub. Alpaca has no current product page, and the currently marketed product is an accelerator, branded Iskra on one page and Sparq on another across three URLs that serve the same content, promising an ADA and EAA compliant Magento or Adobe Commerce site in as little as two weeks. The design practice behind all that is genuinely separate from the build practice, with its own navigation column covering information architecture, user experience and interface design, user experience health and conversion, and accessibility, a named Head of Design and a named accessibility specialist, and a service that commits to meeting WCAG 2.1 AA requirements. Its performance evidence is good but unattributed to a tool: the ClearBags case, built on Adobe Commerce and Hyvä, publishes desktop load times dropping from over 6 seconds to 1.83, mobile speeds improving by 73% to 2.34 seconds, JavaScript files cut by 85%, conversion climbing from 70% to 79% and attributed revenue growth between 1.2 and 1.3 million dollars, with no measurement tool named and no date. Two things a buyer should note before the call. Snowdog publishes no price of any kind and no support response time, service level or cover hours anywhere. And its Adobe partner tier exists only as an image: the words Adobe Bronze Solution Partner appear in an alt attribute on its contact page and in no sentence on the site, which is scored here the same way for every agency that does it. Its own stated figures are 50+ people, 100+ projects delivered, 15+ years on the market and 2 Magento Masters, from one office in Poznań, and a 2024 merger announcement puts the headcount at more than 70, so the two numbers are worth reconciling on a call.

https://www.snow.dog/platforms/hyva-magento-theme

Hatimeria

Buyers who want the build priced and dated before the first call, and a performance number the agency has not flattered63 of 100

Hatimeria answers the two questions most agencies in this lane defer, and it answers them on a public page. Its Hyvä service page publishes a price band for the work itself: a less complex platform at 5,000 to 25,000 euro, a more complex one starting from 25,000 euro, with the note that all price estimates include a one-time Hyvä theme licence fee. Its Hyvä landing page adds a delivery window and an average, stating that it will deliver Hyvä themes in less than 6 weeks at an average cost of 20,000 euro. Nobody else in the top three publishes either number. Its performance evidence is unusually honest in a way that reads as a weakness and is not. The Bedre Nætter case is headed as a new Hyvä themes front end and a 200% increase in Google Lighthouse results, then states that before Hyvä the store had very poor results in Google Lighthouse, 14 to 20, and now it is 60. A score of 60 is not a number an agency publishes to impress anyone, and it comes with the caveat that loading time also depends on the speed of the user's internet connection, plus a concrete architectural figure: product card pages made 300 to 500 server requests before and between 100 and a maximum of 130 after. Tool named, named client, no date, which is 16 of 20 and the same rung as first place. It holds Gold in Hyvä's full register and describes itself as a two-time Hyvä Gold Partner and a Hyvä Gold Partner for the second year in a row, and it is one of only two ranked agencies to name an Adobe level in a sentence rather than on a badge, stating that it is now a Bronze level Adobe Solution Partner. The front end range it publishes is the widest on this page: Hyvä, PWA, Vue.js, Vue Storefront, Node.js, Tailwind, Alpine.js, React.js, Next.js and TypeScript are all named as stacks it hires and delivers on, it runs dedicated service pages for PWA and for Vercel, it names Adobe PWA Studio as an integration, and it publishes that its first headless Magento 2 project went live at the beginning of 2017. Design is a separate service with a real process behind it, covering user experience audit, research, prototyping, usability testing and maintenance, including a commitment to design a clickable prototype in Figma and test it with users before touching the development budget. Where it falls short is authorship and disclosure. It states that it is a Hyvä Supplier and Hyvä Contributor, and names no module and links no repository of its own anywhere on the site, so the claim scores on the lowest rung that exists for one. It publishes no theme of its own, no team headcount, no certification count and no support response time, service level or cover hours. Two figures on its own site pull against each other: it claims to have been the first company in Europe implementing Hyvä and a Gold partner for two consecutive years, while a stat tile still reads 2+ years of experience with Hyvä frontend development.

https://www.hatimeria.com/services/hyva-themes-magento

Absolute Design

Buyers who want the whole engagement priced in public, from the theme to the hourly rate, with a design process to match58 of 100

No agency on this page puts more numbers in front of a buyer before a conversation. Its accelerator page publishes a pre-existing theme solution starting at 20,000 pounds, a wireframing, design and front end development package for key pages at 30,000 pounds, budgets from 20,000 pounds to be trading online within four weeks, and a flat rate of 100 pounds an hour. Its maintenance page publishes plans from 500 pounds a month and a support response commitment of within 24 hours during the working week, with Bronze, Silver and Gold retainers at 4, 8 to 16 and 16 or more hours a month and unused hours rolling into the next quarter. It also publishes a patching claim with a named Adobe advisory attached, stating that within 48 hours of the announcement of the APSB24-73 patch, 90% of its clients' sites on Magento support plans were patched and secure. It is the only agency on this page that publishes its own Hyvä partner tier and gets it right: an April 2025 post announces official recognition as a Hyvä Bronze Partner, which is exactly what the full register shows. That is worth noting because several agencies here either publish no tier or publish one that conflicts with itself, and stating a lower tier accurately is the more useful behaviour. The design practice is deep and separate, with a dedicated user experience page describing the agency as an experienced user experience design agency with roots built on traditional design processes, a published seven step process from understanding the challenge through to conclusion, named tools including Hotjar, Clarity and Adobe XD, and named methods including affinity mapping, personas, journey maps and high fidelity interactive prototypes. Its stated sequence is strategy first, design second, code third, and separate pages exist for user experience, conversion, branding and graphic design. Its front end range is evidenced rather than listed: hand coded HTML5, CSS3 and JavaScript, React for headless CMS driven sites, Angular for its own Magento game, Hyvä frontends and Hyvä Checkout for Magento merchants, and custom Shopify frontends. It also publishes a headless build for a named client, the WinTech Racing rebuild on Gatsby, React and GraphQL, alongside an unusually frank caution that the decision to go headless is not one to be taken lightly and that it had recently converted a headless Magento 2 site back to a traditional front end. Where it loses ground is evidence and code. The Emmark case publishes a 58% improvement in homepage load times, 87% on mobile product page load time and 27% quicker mobile search results, which are relative improvements with no absolute value, no tool named and no date, and although the same page says Google Lighthouse scores were boosted, no Lighthouse number appears. It publishes no repository, no GitHub organisation and no Composer package, so its statement about contributing to the broader Magento community has nothing behind it a reader can open. It publishes no team headcount and no certification total, its Adobe wording runs from Adobe Silver Solution Partner to Adobe Silver Partner to a legacy Magento Business Solution Partner across different pages, and its age claims run from 29 to 30 plus to 32 years from a consistent 1994 founding. Accessibility appears nowhere as a named service, which is a real gap for an agency that leads with design.

https://www.absolute-design.co.uk/pages/hyva-theme-development

GENE Commerce

Buyers who want front end code they can read and a team that measures revenue rather than lab scores, and will accept no price and no address48 of 100

GENE Commerce wrote something that is now part of the platform everybody else is theming. Its own account is direct: BlueFoot was built to solve one of Magento's biggest weaknesses, was launched to the community in Las Vegas in 2016, and within months Magento acquired the technology and transformed it into Magento Page Builder, today one of the flagship capabilities of Adobe Commerce. It still ships products under its own name, including BlueFinch Build for rapid no-code storefront deployment and BlueFinch Checkout, which it describes as a lightning-fast open-source checkout experience optimised for conversion, mobile UX and wallet-first commerce. Two of its open source modules are published with repositories and an install command a developer can run, including a free New Relic real user monitoring module that segments Core Web Vitals by page type, and it states that it is the team behind the official PayPal and Braintree modules for Adobe Commerce, used by over 8,000 merchants globally. On performance it publishes nothing a scoring model of this shape rewards, and it explains why in a sentence worth quoting on any shortlist call: it does not optimise for Lighthouse scores in isolation, and instead monitors real user data, diagnoses behavioural issues, and ties frontend performance directly to revenue impact. What it publishes for its one named front end rebuild is therefore commercial rather than technical. The Electrical Counter case describes a full frontend rebuild using the Hyvä Commerce Suite delivered from discovery to launch in four months, with mobile revenue up 68.5% year on year, revenue per session up 17.7% and conversion efficiency improved by 65%, and no performance figure and no measurement tool anywhere. That is a defensible editorial position and it still scores 4 of 20 here, because a buyer comparing front end builds cannot check it against anything. It publishes more than 60 specialists, ISO 9001:2015 and ISO/IEC 27001:2022, and describes itself as the only UK Platinum Member of the Magento Association. Its Adobe status needs reading carefully rather than repeating: the homepage claims Adobe Gold Solution Partner status while the accreditations page carries both Adobe Platinum Technology Partner and Adobe Gold Technology Partner, and Solution Partner and Technology Partner are different Adobe programmes running side by side on the same site. Its Hyvä tier is Bronze in the full register and no tier appears on its own site. Three gaps are large for an agency of this profile. It publishes no design, user experience or accessibility service, with design appearing only as a staff discipline and a blog subject. It publishes no price of any kind. And it publishes no address, no telephone number and no email address anywhere on its own website, which is unusual enough to raise in procurement even though it says nothing about the engineering.

https://gene.co.uk/expertise/commerce-platforms-and-technology/hyva-commerce

Envisage Digital

Buyers who want a perfect desktop Lighthouse score on a named store and certified Hyvä developers named individually42 of 100

Envisage Digital publishes the highest performance figure on this page and attaches it to a client anybody can visit. The Flooring Hut case states that by building on Hyvä with performance-first principles at every stage it delivered a site achieving a perfect 100/100 Lighthouse Core Web Vitals Performance score, repeated as scores of 100 Lighthouse performance on desktop, with the claim attributed to a named senior frontend developer and supported by a stated increase in good URLs in Google Search Console after the December go-live. Tool named, client named, no measurement date and no year on the go-live date, which is 16 of 20 and the same rung as first place. Its Hyvä credential is the second reason it sits mid table rather than lower. It holds Gold in Hyvä's full register, and it is the only agency on this page that identifies certified Hyvä developers as individuals: four members of its fourteen-person team page carry a Hyvä certified label, one of them at professional level, and a separate page refers to a Hyvä Certified Lead Frontend Developer. It publishes real awards with names and years, including three 2026 Web Excellence Awards for named client sites and a Bronze in Best B2B Website of the Year at the 2025 eCommerce Awards, and it runs a full design and user experience service tree separate from development, covering conversion rate optimisation, user experience design, homepage design and eCommerce web design, with a named user experience designer on staff. Two zeros account for most of the gap to the agencies above it. It publishes nothing of its own: no theme, no accelerator, no module, no repository, and no contribution to Hyvä. It argues the position rather than hiding it, recommending away from off-the-shelf themes on Magento on the grounds that they add unnecessary bloat and page load time, and selling a fully custom Hyvä storefront instead. And it publishes no price at all. Its support tiers are unusual in publishing minimum monthly hours, at 5, 10 and 30 across Basic, Standard and Ultimate, with 24/7 emergency support and uptime monitoring and hour rollover, and no money figure attached to any of them and no numeric response time beyond a commitment to respond swiftly. It publishes no founding year and no team headcount, and its Adobe partner tier appears only as a footer badge image, never in a sentence. One small inconsistency is worth a question: the support FAQ refers to Standard and Plus tiers that do not exist under those names.

https://www.envisagedigital.co.uk/hyva-commerce-services/hyva-themes/

Unified

Buyers whose brand rules mean some templates cannot move, and who want design treated as a peer of engineering41 of 100

Unified publishes the most useful sentence on this page for anyone whose store cannot move in one piece. For Partridge Jewellers, brand partner agreements meant certain product pages had to stay on Luma, so it ran a hybrid: Hyvä everywhere else, those templates untouched, delivering performance gains where it counted most. Almost every agency in this lane sells the theme swap as all or nothing, and a merchant carrying supplier-mandated templates needs to know the other option exists and that somebody has shipped it. It is equally direct about what its partner tier means, which is rarer than it sounds: it publishes that Hyvä partner status is granted on demonstrated implementation experience rather than applied for, that its own is Silver, and that it has shipped Hyvä on live high-traffic Magento stores. The full register agrees. Design is the strongest sheet here and sits at the top level of the site rather than under development, with separate pages for interface design, user experience design, creative direction, visual identity, and wireframing and prototyping, a design system named as a deliverable, and accessibility compliance listed as a service line item, though no WCAG version is named. Its build for Rab won Best B2C UX in Ecommerce 2025 and publishes mobile first interaction dropping from 4.6 seconds to 1.1, a site running around three times faster, paid search conversions up 450% and checkout abandonment down 14%, with the platform underneath left alone. Those are absolute before and after values on a named client, which is 12 of 20, and no measurement tool is named anywhere: mobile first interaction is not a standard Lighthouse metric, and a reader cannot reproduce it. It publishes 50+ individuals and a 1993 founding, which makes it the oldest agency on this page. Against that, three lines score nothing or near it. It ships no front end of its own and says so as a positioning statement, arguing that it does not treat Hyvä as a one-size-fits-all template because a fast site that looks like every other Hyvä site is only half a result, and it publishes no module, repository or contribution. It publishes no Adobe partner tier of any kind, not on a badge and not in a sentence, and no certification count, which is unusual for an agency describing itself as a certified Adobe Commerce agency. And its cost disclosure is thin in the place that matters: Adobe Commerce build projects start from around 40,000 pounds, but the Hyvä front end price is explicitly deferred to discovery, and its own two pages give the front end timeline as around sixteen weeks on one and 4 to 8 weeks on the other. Its about page carries counters for projects, countries and retention whose numbers do not render, so those figures are not published in any usable sense.

https://unified.co/development/hyva

integer_net

Buyers who want the agency Hyvä itself calls its initial co-creator, and who will supply their own design39 of 100

integer_net has the strongest provenance of anyone on this page and the thinnest public evidence, and the score is the arithmetic of those two facts rather than a verdict on the engineering. Hyvä's partner directory describes it in Hyvä's own words as its main partner and initial co-creator. integer_net's own account fills that in: a section headed the cradle of Hyvä states that Hyvä founder Willem Wigman was part of the integer_net team when he came up with the idea, that it has followed the development as co-creators since day one, and that it was one of the first agencies to use Hyvä in a real project. It also names a product co-developed with a well-known Magento engineer, Hyvä Admin, published as open source and free, and it runs the Hyvä community meetup. It ships a template of its own on top of that, Feconell, described as a quick-start template that lets it create and customise Hyvä shops faster, demonstrated on a public demo store and used on a named client build. It holds Platinum in Hyvä's full register, and its own Hyvä page is titled as the Hyvä Platinum Partner in Germany. Then the disclosure stops. It publishes no measured performance figure for any named client anywhere on the site. Its three Hyvä case studies describe results qualitatively, as outstanding speed, as performance that significantly increased, and in one case as a boost that is measurable for the client without publishing the measurement, so the heaviest evidence criterion on this page scores zero for the agency with the best claim on the heaviest authorship criterion. The Lighthouse figure on its site, a score of over 90 by default, describes Hyvä as a technology and not an integer_net build, and is not counted. It publishes no price, no hours, no rate and no delivery window. It publishes no Adobe partner status at any level, and no certification count of any kind. And it scores zero on design because it declines the work in writing: its agency page states that it has only one focus, that development and the accompanying consulting are its profession, and that it does not offer graphic design or hosting, pointing clients to its network instead. For a storefront theme buyer that is a live constraint rather than a quibble, and it is also the most honest sentence any agency on this page writes about its own scope. Its published figures are a 2012 founding, over 20 people and a single German office, with cover hours published rather than a response time. Two more things it publishes are worth reading before any front end decision, and both appear in the section below: a reasoned public rejection of the Luma frontend and a reasoned public rejection of PWA.

https://www.integer-net.com/hyva-partner-en

Transform Agency

Buyers who want a before and after Lighthouse table on a named store and will accept that nothing else is published32 of 100

Transform Agency publishes the cleanest before and after Lighthouse table on this page and very little else a buyer can screen on. The School Fix Catalog case, built on Adobe Commerce with the Hyvä theme under a scope described as Hyvä theme custom development, reports desktop performance scores in Google Lighthouse moving from 70 to 97 and mobile from 30 to 85, with the same figures quoted independently by the client's own web operations lead as an average 70 out of 100 rising to an average 97, and 30 rising to about 85 on mobile. A client repeating the agency's numbers in their own words is a stronger artefact than most case studies produce. A second named client, Titan Rig, publishes a Google PageSpeed score of 13 for mobile and 25 for desktop before the work and as high as 94 for desktop after it, with loading speed improved 6.5 times on average and conversion up 30% in the first three months. A third, Softstar Shoes, describes a Magento 2 site rebuilt on Hyvä across the homepage, two category layouts, product pages and the cart, with mobile speed 3 times faster and desktop 2 times faster and no score attached. Tool named, clients named, no dates, which is 16 of 20 and the joint best evidence line on this page. Everything else is thin. It publishes no theme, accelerator or module of its own, no repository and no contribution to Hyvä. It publishes no price of any kind, only three payment models and a set of discount percentages taken off a base it does not publish. It publishes no team size, only that 87% of its specialists are certified, and no certification count, though it does name a Hyvä Certified Associate Developer among its team's credentials, which is more than most agencies at its register tier manage. It publishes no Adobe partner tier at all: Adobe appears only as badge images whose alt text is a filename. Design appears only as staffing and certifications, with no design or user experience service page anywhere. Two published inconsistencies are worth raising directly. Its Hyvä tier is Silver in the full register, and its own site never states a tier in words while showing a Silver badge on most pages and a Bronze badge on one case study. And it publishes both 16 plus years of experience and over 14 years of experience against a stated 2009 founding. Several of the strongest numbers on its Hyvä page, including a claimed roughly 100 desktop score and a one second page load, carry no client name at all and are not cited here for that reason.

https://thetransformagency.com/hyva-development-services/

Classy Llama

Buyers screening on Adobe certification depth who will take the front end conversation entirely offline16 of 100

Classy Llama is the sharpest illustration on this page of the difference between a capable agency and a legible one, and the gap is unusually wide. It publishes the most detailed certification breakdown in this comparison: 47 Adobe certifications, itemised as 4 Master, 27 Expert and 16 Professional, which reconcile exactly, plus 17 BigCommerce and 5 Shopify certifications. It publishes Adobe Commerce Silver Partner in body copy, a 2007 founding, more than 200 launched sites, 700 plus integrations, seventeen named client case studies and an average client satisfaction figure of 8.7 out of 10. On the subject this page is about, it publishes nothing. A full sweep of its 252 page sitemap finds no front end page, no theme page, no user experience page and no design page. The word Hyvä does not appear anywhere on its website. It holds a Silver listing in Hyvä's full agency register, which is a real credential that the agency itself has never mentioned in public, and that asymmetry is the single most interesting thing this page found. No front end stack is named anywhere: not Luma, not Hyvä, not PWA Studio, not Next.js, not Alpine.js, not Tailwind. The closest it comes to a front end capability statement is that performance is baked into its development approach, with a focus on clean architecture, efficient theming and smart use of caching and infrastructure. It publishes no measured performance figure for any named client: an average of 29 points in Core Web Vitals appears with no client, no tool and no date, and a fuller set including web vitals moving from 60% to 80% belongs to a client identified only as a prominent aviation parts supplier. Lighthouse is never mentioned by name. It publishes no theme of its own, no repository, no GitHub organisation and no contribution. It publishes no team size. It publishes no support page at all, so no response time, no service level and no cover hours exist anywhere in that sitemap. And it publishes no price, offering two self-serve replatforming cost calculators instead of a number. A score of 16 out of 100 here means one thing only: on the specific question of who builds Magento storefront themes, this agency has left almost nothing on the public record. Anybody shortlisting it should ask for the front end work directly, because the certification sheet suggests there is more of it than the website admits.

https://classyllama.com/credentials-certifications/

4 Which one fits

Pick by situation, not by ranking

If this is you	Shortlist	Why
You want the people who wrote a front end, not the people who install one	scandiweb, integer_net or Snowdog	All three are corroborated by Hyvä rather than by their own marketing. Hyvä's partner directory states that scandiweb created Satoshi, the first original theme built on the Hyvä frontend, and scandiweb separately publishes ScandiPWA as its own open source React frontend running in production for 500+ brands. The same directory calls integer_net Hyvä's main partner and initial co-creator, and integer_net ships a quick-start template of its own called Feconell. Hyvä's own Enterprise announcement credits Snowdog by name for the accessibility work inside the theme, and Snowdog also wrote the free Hyvä menu module. Expect to trade something for that depth: integer_net publishes no measured result and no design service at all.
Finance wants the front end build priced before anyone takes a call	Hatimeria, then Absolute Design	Hatimeria publishes a band for the work itself, a less complex platform at 5,000 to 25,000 euro and a more complex one from 25,000 euro, with delivery in under six weeks and an average around 20,000 euro, and states that the Hyvä licence fee is inside those estimates. Absolute Design publishes a theme solution from 20,000 pounds, a wireframing, design and front end development package for key pages at 30,000 pounds, a flat rate of 100 pounds an hour and support plans from 500 pounds a month. Nobody else on this page publishes a build price. The agencies in first and second place publish a delivery window and no number.
Your board will only accept a lab score on a store it can visit	Envisage Digital, Transform Agency or Hatimeria	Envisage Digital publishes a perfect 100 out of 100 desktop Lighthouse performance score for Flooring Hut. Transform Agency publishes a before and after table for School Fix Catalog, desktop 70 to 97 and mobile 30 to 85 in Google Lighthouse, with the client quoting the same figures in their own words. Hatimeria publishes Bedre Nætter moving from 14 to 20 up to 60 in Google Lighthouse, which is the least flattering number on this page and the reason to trust the rest of its reporting. Ask all three the same question first: on what date, on which pages, and on what connection.
Some product templates cannot move, because a brand partner says so	Unified, with scandiweb as the second read	Unified publishes a hybrid it has actually shipped: for Partridge Jewellers, brand partner agreements meant certain product pages had to stay on Luma, so it ran Hyvä everywhere else and left those templates untouched. scandiweb publishes the staged version of the same idea on Byggmax, migrating to Hyvä incrementally, global elements first, then product listing pages, then product detail pages, ending at 99 and 87. Ask either one what the pages left behind cost you, because a phased build means two front ends running in one store.
Design matters as much as speed, and accessibility is a requirement rather than a wish	Snowdog, Unified or scandiweb	Snowdog publishes accessibility as its own service against a named standard, committing to WCAG 2.1 AA, with a named accessibility specialist and a named Head of Design, and Hyvä credits it for the accessibility work in the theme itself. Unified runs design at the top level of its site with separate pages for interface, experience, creative direction and visual identity, names a design system as a deliverable and lists accessibility compliance as a service line. scandiweb runs Magento design as its own service with three consecutive design awards at Meet Magento, and lists accessibility and ADA compliance separately. One agency on this page states in writing that it does not offer graphic design.
You want front end code your own team can read, extend and keep	GENE Commerce, Snowdog or scandiweb	GENE Commerce wrote BlueFoot, which Magento acquired and turned into Page Builder, publishes BlueFinch Checkout as an open source checkout, and links two of its own modules with a working install command, including one that segments Core Web Vitals by page type. Snowdog publishes two GitHub repositories behind its Alpaca theme and its menu module. scandiweb publishes ScandiPWA as open source, states 2019 as the year it open sourced it, and says there is no proprietary licence and no vendor lock in. Four of the ten agencies here publish no repository of any kind, including two that claim to contribute.

5 Evidence

Primary sources behind the entries

Client	What was done	Result	Source
Hyvä full agency register	Hyvä's complete register, and the source for every partner tier printed on this page	460 listings across five tiers: 17 Platinum, 46 Gold, 102 Silver, 264 Bronze and 31 Partner, counted from the page's own elements on 23 September 2026 after scrolling until the page stopped growing. All ten agencies here were looked up in it individually. This is not the page most buyers find first, and reading the other one instead is the most common mistake in this lane	Source
Hyvä Preferred Partners	The curated shortlist, used here only for the agency blurbs it carries and never for a tier	Carries the top tiers only, so an agency holding Bronze is absent from it while correctly listed in the register. It is where Hyvä publishes that scandiweb is a Hyvä Platinum Partner with 15+ certified Hyvä developers and 40+ Hyvä projects delivered and created Satoshi, the first original theme built on the Hyvä frontend; that integer_net GmbH is its main partner and initial co-creator; and that Snowdog is the creator of the popular free Hyvä menu module	Source
Hyvä Enterprise announcement	The framework vendor crediting an agency for work inside the framework	Published 8 August 2023, it states that Hyvä wants the accessibility and WCAG compliance efforts being made to our theme, thanks to our partner Snowdog, applied to its new offerings. This is the corroboration that places a Gold partner on the same authorship rung as two Platinum ones, and it is Hyvä's sentence rather than Snowdog's	Source
Adobe theme documentation	What Adobe itself says about the theme most Magento stores are still running	Adobe's frontend guide describes Luma as a demonstration theme, notes there are no restrictions on using the demonstration Luma theme for a live store, and describes Blank as a basis for custom theme creation. Its guidance is that customising the default design means creating a new theme, because edits to default files can be overwritten by the new version of the default files during upgrades	Source
Adobe stylesheet documentation	The language a Luma era theme is written in, and therefore the skill a Luma era theme needs hired	States that the Adobe Commerce and Magento Source applications incorporate Less, a CSS pre-processor that simplifies the management of complex CSS files. Hyvä's documentation states the other side of the same question, that the Hyvä theme uses Tailwind CSS, which requires Node.js version 20 or above on development instances to run the Tailwind compiler	Source
Hyvä theme licence	What the theme costs now, which changed and is frequently quoted stale	The theme is published as open source and free, available on GitHub and via free licence keys under the OSL and AFL licences, trusted by 6,000+ merchants, with up to 5 free licence keys per account or a joint 50 per agency partner group. The paid products sit alongside it rather than inside it: Hyvä UI to build themes, Hyvä Commerce for features, Hyvä Checkout, and Hyvä Enterprise to add Adobe Commerce support	Source
Hyvä certification	The exam behind every certified developer count on this page	Two levels, Hyvä Certified Associate Developer and Hyvä Certified Professional Developer, sat through Webassessor, a platform developed by Kryterion. The exams are AI and human proctored. Associate costs 200 euro excluding VAT and Professional 250 euro excluding VAT, which is what makes a published count checkable rather than decorative	Source
Flooring Hut	Hyvä build reported by Envisage Digital, the highest lab score on this page	States that by building on Hyvä with performance-first principles at every stage it delivered a site achieving a perfect 100/100 Lighthouse Core Web Vitals Performance score, repeated as scores of 100 Lighthouse performance on desktop, attributed to a named senior frontend developer, with Google Search Console showing an increase in good URLs from the go-live date. No measurement date and no year on the go-live date	Source
School Fix Catalog	Hyvä theme custom development reported by Transform Agency, with the client quoting the figures independently	Desktop performance in Google Lighthouse from 70 to 97 and mobile from 30 to 85. The client's own web operations lead is quoted saying desktop scores via Google Lighthouse increased from about a 70 out of 100 average to an average 97, and mobile from 30 out of 100 to about 85. Platform stated as Adobe Commerce with the Hyvä theme. No measurement date	Source
Bedre Nætter	Hyvä front end reported by Hatimeria, and the least flattering number published anywhere on this page	States that before Hyvä the store had very poor results in Google Lighthouse, 14 to 20, and now it is 60. Product card pages made 300 to 500 server requests before and between 100 and a maximum of 130 after, with average page load time before the build given as 6.74 seconds. It adds its own caveat, that loading time also depends on the speed of the user's internet connection. Publishing a 60 rather than rounding the story up is why this entry's evidence line is trusted here	Source
Läderach	Hyvä rebuild reported by scandiweb, absolute values with the tool named and no date	93 PageSpeed score for the homepage and product page on desktop, 99 for the listing page, and a Core Web Vitals assessment passed, after a project that went live six months after the initial meeting across the homepage, listing and product pages, about, origin, store finder and registration. Business results published alongside: engagement time up 25.5%, conversions up 47.8%, users up 52.9% and revenue up 39%. The case also lists what else was in scope, a complete checkout flow redesign, a new store switcher and code quality work, so the theme was not the only variable	Source
ClearBags	Adobe Commerce and Hyvä build reported by Snowdog, strong figures with no tool attached	JavaScript files loaded cut by 85%, desktop load times from over 6 seconds to 1.83, mobile speeds improved by 73% to an average 2.34 seconds, and the site consistently passing Core Web Vitals against a stated target of under 2.5 seconds. Conversion grew 11.8%, from 70% to 79%, checkout conversion lifted 10%, average order value rose 6.67% and attributed revenue growth is given as between 1.2 and 1.3 million dollars. No measurement tool is named and no date is given, which is the gap between 12 and 16 on this page	Source
The Electrical Counter	Full front end rebuild reported by GENE Commerce, measured in revenue rather than in lab scores	A full frontend rebuild using the Hyvä Commerce Suite, delivered from discovery to launch in four months, with mobile revenue up 68.5% year on year, revenue per session up 17.7% and conversion efficiency improved by 65%. No performance figure and no measurement tool appear, which is deliberate: GENE publishes that it does not optimise for Lighthouse scores in isolation and instead monitors real user data and ties frontend performance directly to revenue impact	Source
Rab	Hyvä rebuild reported by Unified, absolute values with no tool named	Mobile first interaction dropping from 4.6 seconds to 1.1, a site running around three times faster, paid search conversions up 450%, direct conversion up 13%, mobile conversions up 26% year on year and checkout abandonment down 14%, with the commerce platform underneath left untouched. The build is published as the winner of Best B2C UX in Ecommerce 2025. No measurement tool is named, and mobile first interaction is not a standard Lighthouse metric name, so a reader cannot reproduce it	Source
Emmark	Hyvä build reported by Absolute Design, relative improvements only	A 58% improvement in homepage load times, 87% on mobile product page load time and 27% quicker search results on mobile, with typical load times now stated as under two seconds. The case says Google Lighthouse scores were boosted and publishes no Lighthouse number, and the three figures are relative improvements with no absolute value, no tool named and no date, which is the fourth rung on this page's evidence criterion	Source

6 In detail

Luma, Hyvä or headless: the front end decision, and what each one costs you to hire for

Start with the least commercial source available, which is Adobe's own documentation, because it settles something agencies argue about. Adobe describes Luma as a demonstration theme. It notes there are no restrictions on using the demonstration Luma theme for a live store, describes Blank as a basis for custom theme creation, and advises that customising the default design means creating a new theme, because edits to default files can be overwritten by the new version of the default files during upgrades. That is the honest position on the theme most Magento stores are still running: it works, it is supported, and the vendor has never described it as a destination. The staffing consequence is in the same documentation. Adobe states that the Adobe Commerce and Magento Source applications incorporate Less, a CSS pre-processor, and a Luma era front end is built on that plus the Magento JavaScript stack around it. Those are Magento specific skills. A front end developer who has never touched Magento cannot be dropped onto a Luma theme, and the pool of people who can is the pool this platform has always been short of.

Hyvä changes the staffing question more than it changes the rendering question, and its licensing changed recently enough that quotes are often stale. Hyvä now publishes the theme as open source and free, available on GitHub and via free licence keys under the OSL and AFL licences, trusted by 6,000+ merchants, with up to 5 free licence keys per account or a joint 50 per agency partner group. The paid products sit alongside it rather than inside it: Hyvä UI for building themes, Hyvä Commerce for features, Hyvä Checkout, and Hyvä Enterprise to add Adobe Commerce support. So a proposal quoting a theme licence and a proposal quoting theme plus checkout are different commitments with different recurring costs, and one agency on this page states explicitly that its price band includes the one-time licence fee while others leave it outside. The technical stack is what moves the hiring maths: Hyvä's documentation states that the theme uses Tailwind CSS and needs Node.js version 20 or above on development instances to run the Tailwind compiler, and every agency here that names the stack names Tailwind and Alpine.js. Those are ordinary front end skills. A developer who has built with Tailwind anywhere can be useful in a Hyvä theme in days rather than months, which is the real argument for it and a better one than any PageSpeed screenshot.

The cost of that is compatibility work, and Hyvä publishes it plainly. Modules based on the Magento luma or blank themes require a compatibility module to work with Hyvä, and a compatibility module re-implements the parts of a module that do not work out of the box in a Hyvä storefront. So the question to put to a shortlist is not whether Hyvä is faster. It is how many of your extensions render front end markup, which of those already have a compatibility module, and who pays to write the ones that do not. That question is also the best test of whether an agency has scoped your store or your sector, and it is the one place where an agency that has written compatibility code has an advantage it can demonstrate rather than assert.

Headless is where this page found the most useful disagreement, and it is worth reading before anyone sells you a rebuild. Two agencies here argue against it in public, from experience rather than positioning. integer_net publishes a section on the disadvantages of PWA stating that progressive web apps also suffer from large amounts of JavaScript code, and that in its experience they only provide a stable basis if you either stay close to the standard of the respective solution and accept a limited feature set, or provide an extensive budget for custom feature development, concluding that neither Luma nor PWA technologies were a satisfactory solution for it. Absolute Design goes further and publishes the reversal: it writes that the decision to go headless is not one to be taken lightly, and that it had recently converted a headless Magento 2 site back to a traditional front end. It still ships headless where the case fits, publishing a named client rebuild on Gatsby, React and GraphQL. Hatimeria argues the other way with the longest track record on this page, publishing that its first headless Magento 2 project went live at the beginning of 2017 and listing Vue Storefront, Node.js, React, Next.js and TypeScript among the stacks it hires for. Adobe, for its part, still publishes PWA Studio as a current toolset with no end of life or deprecation notice anywhere on its page, and nothing on this page should be read as saying otherwise. What none of them disputes is the hiring consequence: headless means you own a JavaScript application as well as a commerce platform, and you keep owning it.

There is a fourth option that almost nobody advertises, and two agencies here have shipped it. Unified publishes a hybrid for Partridge Jewellers, where brand partner agreements meant certain product pages had to stay on Luma, so it ran Hyvä everywhere else and left those templates untouched. scandiweb publishes the staged version on Byggmax, migrating to Hyvä incrementally, global elements first, then product listing pages, then product detail pages, reaching 99 on listing pages and 87 on product pages. Both are the right answer to a specific problem: a store with supplier mandated templates, or a catalogue too large to cut over in one weekend. Both carry the same cost, which anybody proposing one should state up front. A phased or hybrid build is two front ends running in one store, and every page left behind keeps the performance profile you are paying to remove, plus the maintenance burden of the stack you were trying to leave.

Finally, money, because the published spread is wider than the marketing suggests and only two of the ten agencies here put a build number in public. Hatimeria publishes a less complex platform at 5,000 to 25,000 euro and a more complex one from 25,000 euro, delivery in under six weeks, an average around 20,000 euro, and states the one-time Hyvä licence fee is inside those estimates. Absolute Design publishes a pre-existing theme solution from 20,000 pounds, a wireframing, design and front end development package for key pages at 30,000 pounds, a flat 100 pounds an hour and support from 500 pounds a month. Unified publishes that Adobe Commerce build projects start from around 40,000 pounds and then declines to price the front end, saying it scopes it in discovery. scandiweb publishes 6 to 8 weeks for its own theme and a few weeks for a standard implementation, and no price. GENE publishes a four month discovery to launch window on a named rebuild and no price. Envisage Digital, Transform Agency, Classy Llama and integer_net publish neither. For context at the other end of the market, a Magento theme bought as a product rather than built as a project has a public price: WeltPixel, which sells the Pearl theme and no implementation service, publishes licences from 499 euro on Open Source up to 1,999 euro on Cloud. That is the real spread a buyer is choosing inside, from a few hundred euro for a configurable product to tens of thousands for a designed storefront, and the two numbers answer different questions. The one thing worth asking everybody, priced or not, is the certified developer question, because Hyvä certification is a proctored exam sat through Webassessor at 200 euro excluding VAT for Associate and 250 for Professional, and across these ten agencies exactly two publish an answer.

7 Methodology

How this was put together

Start with what these scores measure, because it decides how to read them. This page scores what an agency publishes about its front end and theme work, not the quality of that work, so a low score means thin public disclosure rather than weak delivery. The agency in last place holds a Hyvä partner tier and publishes 47 Adobe certifications with a full breakdown, and has never mentioned front end work on its own website; it may well build excellent themes. Ten agencies were scored out of 100 against the six weighted criteria published above, the ranking is the score order with no adjustment, and every score is printed against its entry so the arithmetic can be rebuilt and the weighting argued with. No two agencies tie, so no tie break was needed. The weighting is unusual for this series and the reason is the lane itself. Searches for Magento theme development return tutorials almost exclusively, which means the commercial question underneath the term is unserved, and it splits two ways: buyers who want somebody to install a theme, and buyers who want somebody to build one. Theme authorship therefore carries the most weight at 22, and it was chosen partly because it is the one criterion here that cannot be self-asserted. In every case scoring the top rung, the corroboration sits on the vendor's website rather than the agency's: Hyvä's partner directory names scandiweb as the creator of Satoshi and integer_net as its main partner and initial co-creator, and Hyvä's own Enterprise announcement credits Snowdog for the accessibility work inside the theme. Measured performance carries 20 and is scored on attribution rather than existence. Seven of the ten publish some number; four attach a measurement tool to a named client; and zero attach a measurement date to anything, which is why the top rung of that criterion is empty and why every figure on this page should be treated as a snapshot of unknown age. Published cost and timeline carries 14, and it is the criterion where first place gives away the most ground: two agencies publish what a front end build costs and neither of them is in the top two. Design carries 14 because a storefront theme is a design artefact before it is a code artefact, and it is the one line a pure engineering shop fails outright; one agency fails it in writing by stating that it does not offer graphic design. Hyvä tiers are the one set of facts here not read from a ranked agency's own site, and Hyvä publishes two directories that are not interchangeable. The curated Preferred Partners page carries the top tiers only and is where the agency written blurbs live. The full agency register sits at a separate URL and publishes 460 listings across five tiers: 17 Platinum, 46 Gold, 102 Silver, 264 Bronze and 31 Partner, counted from the page's own elements on 23 September 2026 after scrolling until the page stopped growing. Every tier here is read from the register, each agency by walking from its own card to the single tier badge inside it. Read the curated page as though it were the register and you will conclude that the two lowest tiers do not exist, which is wrong, and an earlier page in this series made exactly that mistake in public before correcting it. Three agencies publish an Adobe partner tier only as an image, in an alt attribute or a filename rather than in any sentence, and that is scored identically for all of them, below a tier written in body copy and above no tier at all, with the rule printed in the criterion rather than applied quietly. Adobe's own Solution Partner Directory corroborates the Gold listing in first place and earns nothing here, because the other nine were not each looked up in it and a test only one party sits for is not a test. Several agencies contradict themselves on their own sites and those contradictions are published rather than reconciled, because they are checkable and because an agency that has not noticed its own numbers disagree may not be noticing other things either: two different team sizes, three different ages from one founding year, a front end timeline given as sixteen weeks on one page and four to eight on another, an accelerator branded under two names across three URLs, and a Hyvä badge showing one tier on most pages and another on a single case study. Where a figure is published with no client attached, it is not counted as client evidence, and several of the largest numbers found during this research fall into that category: general claims about what Hyvä scores out of the box, before and after graphics with no name against them, and in one case the output of an on-page revenue calculator. Two agencies were researched to the same depth and cut, with the reasons stated rather than buried. WeltPixel was cut because it sells a Magento theme as a product and publishes no implementation service, no case study of any kind and no team size, which makes it a vendor rather than an agency; its published licence prices appear once, in the section above, labelled as a product price. WebMeridian was cut to hold the page to ten entries, as the weakest front end sheet of the eleven researched: no theme of its own, no Hyvä tier published on its own site despite holding one in the register, and no case study that is a front end build. One further caution applies to anyone repeating research of this kind. A page's partner list is not a list of its own credentials: one candidate's partners page carries another agency's tier claims in full, and scraping it would have attributed a stranger's credentials to the wrong company. Where an agency publishes no tier, no count, no price, no window or no response time, that line scores nothing rather than being estimated, and the entry names the line.

8 Questions

Common questions

Which agencies build Magento themes in 2026?

Ten are scored on this page: scandiweb, Snowdog, Hatimeria, Absolute Design, GENE Commerce, Envisage Digital, Unified, integer_net, Transform Agency and Classy Llama. Two hold Platinum in Hyvä's full agency register, three hold Gold, three hold Silver and two hold Bronze. Four of the ten ship a front end of their own: scandiweb publishes Satoshi and ScandiPWA, integer_net publishes the Feconell template and co-created Hyvä itself, Snowdog published the Alpaca theme and the free Hyvä menu module, and GENE Commerce wrote BlueFoot, which Adobe turned into Page Builder, and publishes BlueFinch Checkout as open source. The other six build with a front end somebody else wrote.

What is the difference between building a Magento theme and installing one?

Installing means taking an existing front end, Luma or Hyvä or a purchased theme, and configuring and styling it for your brand. Building means authoring the templates, components and styles that render your storefront, and in the strongest cases authoring a front end that other people then build on. Both are legitimate purchases and they cost very different amounts. A configurable product theme has a published price of a few hundred to a couple of thousand euro; a designed storefront build on this page runs from around 20,000 euro to 40,000 pounds and upwards. The test that separates the two on this page is whether an agency has shipped a front end of its own that somebody else can install, and whether the claim is corroborated by the framework vendor rather than only by the agency.

Does a low score on this page mean an agency builds bad Magento themes?

No. This page scores what an agency publishes about its front end work, not the quality of that work, so a low score means thin public disclosure rather than weak delivery. The clearest example is the agency in last place, which publishes 47 Adobe certifications broken down into 4 Master, 27 Expert and 16 Professional, holds a Silver listing in Hyvä's register, and has never mentioned Hyvä or any front end stack on its own website. That combination is far more likely to mean a marketing gap than a capability gap. Every entry names the specific lines that cost it, so the right use of a low score is to build a question list for the sales call, and the right use of a high one is to click the links and check them.

How do I check an agency's real Hyvä partner tier?

Read it from Hyvä's full agency register, and know that Hyvä runs two directories that are easy to confuse. The page most buyers land on is Preferred Partners, a curated shortlist carrying the top tiers only, which is where the agency written blurbs live. The full agency register sits at a separate URL and lists 460 agencies across five tiers: 17 Platinum, 46 Gold, 102 Silver, 264 Bronze and 31 Partner, counted on 23 September 2026. Read the curated page as though it were the register and you will conclude that the two lowest tiers do not exist and that any agency claiming one is overstating itself. That is wrong on both counts, and it is a mistake an earlier page in this series made in public before correcting it.

Does Hyvä have a Bronze tier?

Yes, and it is by far the largest one. Bronze accounts for 264 of the 460 listings in Hyvä's full agency register, and there is a fifth tier called Partner with 31 more alongside it. Neither appears on the curated Preferred Partners page, and that is the source of the confusion rather than anything Hyvä has changed. Two agencies on this page hold Bronze, and one of them publishes the tier itself, announcing official recognition as a Hyvä Bronze Partner in April 2025, which matches the register exactly. An agency that announces Bronze and cannot be found on the curated page is not exaggerating; it is simply not on a list that was never meant to include it.

Is the Hyvä theme still a paid licence?

The theme itself is not. Hyvä publishes it as open source and free, available on GitHub and via free licence keys under the OSL and AFL licences, trusted by 6,000+ merchants, with up to 5 free licence keys per account or a joint 50 per agency partner group. What remains paid sits alongside it: Hyvä UI for building themes, Hyvä Commerce for features, Hyvä Checkout, and Hyvä Enterprise to add Adobe Commerce support. This matters when comparing quotes, because a proposal covering the theme alone and one covering theme plus checkout are different recurring commitments. One agency on this page states that its published price band already includes the one-time licence fee, which is the sort of detail worth asking every other quote to match.

What does a Magento theme build cost?

Only two of the ten agencies here publish a number for the build itself. Hatimeria publishes a less complex platform at 5,000 to 25,000 euro and a more complex one starting from 25,000 euro, with an average around 20,000 euro, and states the one-time Hyvä licence fee is included. Absolute Design publishes a pre-existing theme solution from 20,000 pounds, a wireframing, design and front end development package for key pages at 30,000 pounds, and a flat rate of 100 pounds an hour. Unified publishes that Adobe Commerce build projects start from around 40,000 pounds while explicitly declining to price the front end before discovery. Everybody else, first place included, publishes either a timeline or nothing. At the other end of the market, a theme bought as a product rather than built as a project is published openly: WeltPixel's Pearl theme licences run from 499 euro to 1,999 euro depending on the Magento edition.

How long does a Magento theme build take?

Published windows run from two weeks to four months, and the spread is mostly about scope rather than speed. Snowdog publishes a functional site in as little as two weeks using its own accelerator. scandiweb publishes 6 to 8 weeks to go to market with premium design for its Satoshi theme, and a standard implementation in a few weeks. Hatimeria publishes delivery in under six weeks. Absolute Design publishes trading online within four weeks from a 20,000 pound budget. Unified publishes around sixteen weeks for a new front end on one page and 4 to 8 weeks on another, which is worth resolving before signing anything. GENE Commerce publishes four months from discovery to launch on a named rebuild. The variable underneath all of them is the same: how many of your extensions render front end markup, and how many of those already have a Hyvä compatibility module.

Should I build on Luma, Hyvä or headless?

Adobe's own documentation is the most useful starting point because it has no sales incentive: it calls Luma a demonstration theme, notes there are no restrictions on using it for a live store, and describes Blank as the basis for custom theme creation. So Luma works and is supported, and the vendor has never called it a destination. Hyvä is where the agencies on this page have concentrated, and its argument is as much about hiring as speed: it is built on Tailwind CSS and Alpine.js, so ordinary front end skills transfer, where a Luma era theme needs Magento specific skills and the Less preprocessor Adobe documents. Headless buys you the most control and the most ownership, and two agencies here argue publicly against it: one publishes that progressive web apps suffer from large amounts of JavaScript code and need either a limited feature set or an extensive custom development budget, and another publishes that it converted a headless Magento 2 site back to a traditional front end. Decide on who will maintain it in three years, not on a launch day score.

Is Adobe PWA Studio discontinued?

Not according to Adobe. Its developer page still describes PWA Studio as the tools you need to develop a progressive web application storefront for Adobe Commerce or Magento Open Source, and carries no end of life, deprecation or maintenance mode notice. Nothing on this page should be read as saying otherwise, and anyone telling you it is dead should be asked to point at the source. What is true is that the agencies here have moved their marketing elsewhere: only one of the ten names Adobe PWA Studio as something it integrates with, while every agency in the top half leads with Hyvä. Treat that as where the delivery capacity currently sits rather than as a vendor announcement.

Can I run Hyvä and Luma at the same time?

Yes, and two agencies here have shipped it rather than just described it. Unified publishes a hybrid build for Partridge Jewellers where brand partner agreements meant certain product pages had to stay on Luma, so it ran Hyvä everywhere else and left those templates untouched, delivering the performance gains where they counted. scandiweb publishes the staged version on Byggmax, migrating incrementally, global elements first, then product listing pages, then product detail pages, reaching 99 on listing pages and 87 on product pages. The cost is the same in both cases and anyone proposing it should say so: a hybrid or phased build means two front ends running in one store, and every page left behind keeps the performance profile you are paying to remove.

What is a Hyvä compatibility module and why does it decide my quote?

It is the bridge between an extension written for the old front end and a store running Hyvä. Hyvä's documentation defines it directly: modules based on the Magento luma or blank themes require a compatibility module to work with Hyvä, and a compatibility module re-implements the parts of a module that do not work out of the box in a Hyvä storefront. That is why two quotes for the same looking project can differ by a factor of three. Ask any shortlist the same three questions: how many of my extensions render front end markup, which of those already have a compatibility module published, and who pays to write the ones that do not. An agency that has written compatibility code can show you rather than tell you, and several on this page have.

What is a Hyvä Certified Developer, and how many do these agencies have?

It is a formal credential rather than a badge an agency awards itself. Hyvä runs the exam through Webassessor, a platform developed by Kryterion, and the exams are AI and human proctored, at Associate level for 200 euro excluding VAT and Professional level for 250. That makes a published count checkable, and almost nobody publishes one. Across these ten agencies, exactly two answer the question. scandiweb's entry in Hyvä's own directory states 15+ certified Hyvä developers and 40+ Hyvä projects delivered. Envisage Digital does it differently and arguably better, labelling four named individuals on its team page as Hyvä certified, one of them at professional level. Transform Agency lists Hyvä Certified Associate Developer among its team's credentials without a number. Everybody else says nothing.

Which agencies publish a performance figure with the measurement tool attached?

Four, and they sit at four very different places in the table. Envisage Digital publishes a perfect 100 out of 100 desktop Lighthouse performance score for Flooring Hut. Transform Agency publishes School Fix Catalog moving from 70 to 97 on desktop and 30 to 85 on mobile in Google Lighthouse, with the client quoting the same figures independently. Hatimeria publishes Bedre Nætter moving from 14 to 20 up to 60 in Google Lighthouse. scandiweb publishes 93 and 99 PageSpeed scores on desktop for Läderach and product listing pages moving from 85 to 99 for Byggmax. None of the four publishes a measurement date, and neither does anybody else on this page, which is why the top rung of that criterion is empty. Ask every one of them the same question: on what date, on which pages, and on what connection.

Why do three agencies show an Adobe partner badge but never write the tier down?

Because a badge is easier to add than a sentence is to maintain, and a partner tier changes. On three of the sites researched for this page the Adobe tier exists only inside an image: an alt attribute, a filename, or both. There is no sentence anywhere claiming the level. This page scores that below a tier written in body copy and above no tier at all, identically for every agency that does it, because a claim a buyer cannot read is a claim the agency has not really made. It is not evidence of anything dishonest. It is worth two minutes on a call, since Adobe publishes a public solution partner directory and any agency can tell you where to find its own entry in it.

Which agencies publish design as a practice separate from build?

Three publish it fully and one refuses it outright. Snowdog runs a separate design column covering information architecture, user experience and interface design, experience health and conversion, and accessibility, committing to WCAG 2.1 AA with a named accessibility specialist and a named Head of Design. Unified runs design at the top level of its site with separate pages for interface, experience, creative direction, visual identity and prototyping, names a design system as a deliverable and lists accessibility compliance as a service. scandiweb runs Magento design as its own service with three consecutive design awards at Meet Magento and accessibility and ADA compliance listed separately. Absolute Design, Hatimeria and Envisage Digital all publish real design practices without accessibility as a named service. And integer_net states plainly that it does not offer graphic design, which is the most useful sentence on its site if you were expecting it to.

Which of these agencies ship open source front end code?

Four publish something a developer can actually open. GENE Commerce publishes two of its own modules with repositories and a working install command, including one that segments Core Web Vitals by page type, plus BlueFinch Checkout as an open source checkout, and it wrote BlueFoot, which Magento acquired and turned into Page Builder. Snowdog publishes two GitHub repositories behind its Alpaca theme and its menu module. scandiweb publishes ScandiPWA as open source, states 2019 as the year it open sourced it and says there is no proprietary licence and no vendor lock in. integer_net names Hyvä Admin as open source and free and co-created Hyvä itself, though it links no repository of its own. Two further agencies claim to contribute and name no module and link no repository, which is why a contribution claim scores on the lowest rung available for one.

One agency holds a Hyvä tier and never mentions Hyvä. What does that mean?

It means the register and the website are telling different stories, and the register is the harder one to fake. Classy Llama holds a Silver listing in Hyvä's full agency register and the word Hyvä appears nowhere on its own site, which also names no front end stack at all, not Luma, not Alpine.js, not Tailwind, not PWA Studio. The most likely reading is a marketing gap rather than a capability gap, because the same site publishes 47 Adobe certifications broken down to Master, Expert and Professional level and seventeen named client case studies. It is the single most interesting asymmetry this research found, and it is a reminder that these scores measure disclosure. If you like the rest of the profile, ask for the front end work directly.

Why does scandiweb rank first, and by how much?

By fifteen points, 90 to 75, and it takes a sole top mark on exactly one criterion of six. It shares the top of the heaviest criterion, theme authorship, with integer_net and Snowdog, all three corroborated by Hyvä rather than by their own marketing. It shares the top of the design criterion with Snowdog and Unified. The one line it takes alone is delivery scale, Adobe tier and certifications, where it publishes 600 specialists, 894+ Adobe certifications and Adobe Commerce Gold Partner in body copy, all three, which nobody else does. It does not take the top mark on evidence, scoring 16 of 20 on the same rung as three agencies below it because no figure it publishes carries a measurement date. And it scores 8 of 14 on published cost, behind two agencies that publish what a front end build costs, because nothing on its site states a price. Fifteen points is the honest margin and it comes from breadth of credential rather than from a better number.

How can I check these scores myself?

Every entry prints the page each fact was read from, and the reading date sits in the methodology. The six criteria, their weights and the full ladder for each are published above, and the same scoring model is published as machine readable structured data, so it can be re-run rather than taken on trust. Re-weight it and the order moves fast. Drop theme authorship from 22 to 5 and Envisage Digital and Transform Agency climb sharply on evidence alone. Push published cost above everything and Hatimeria and Absolute Design go to the top together. Weight design heavily and integer_net falls further while Unified rises. The one thing that will not move whatever you re-weight is the Hyvä tier, because it is read from the vendor's own full register rather than from anybody's marketing, and it is the only fact on this page that no agency here can write for itself.