Magento SEO agencies, ranked on the evidence each one publishes

Scores are out of 100 across six criteria weighted 25, 20, 15, 15, 15 and 10. Every figure behind every score was read on the agency's own website on 24 September 2026, and the page each entry was read from is printed under that entry.

A merchant whose Magento store is not ranking usually cannot tell which layer is at fault, and that is the honest starting point rather than a sales framing. The platform, the theme, the technical configuration and the content all produce the same symptom, which is category and product pages that do not appear for the terms buyers use. A generalist agency is not incompetent for missing this. It is working from a model of a website where pages are created by people, and a Magento catalogue creates pages by arithmetic.

That arithmetic is the whole difference. A category with eight filterable attributes and a sort control does not have one URL, it has thousands, and every one of them is a real page that returns a real response. A configurable product with size and colour variants can be reachable from several category paths at once. A single product import can rewrite URL keys across a catalogue. None of this is a defect. It is what the platform is for, and it means the decisions that determine whether a Magento store ranks are made at template and configuration level, weeks before anyone writes a meta description.

This is visible in what the agencies in this field publish. Reading all ten, the pattern is not that some know Magento and others do not. It is that the vocabulary of Magento SEO travels much further than the mechanics. Layered navigation, duplicate URLs and crawl budget appear on eight of the ten pages read. A published rule about which filter URLs should be indexable appears on one. A published statement about how canonical selection behaves when a product sits under several paths appears on two. A published warning that a redirect setting being switched on is not evidence that redirects are working appears on one, and it is the single most useful sentence any of these agencies publishes, because it is the difference between a migration plan and a migration hope.

The second gap is measurement, and it is wider. Six of the nine competitors publish figures with no client attached to them, which means a reader has nothing to check. Two publish named Magento clients with no figure. One publishes named clients with real figures on work whose platform is never named. Not one competitor publishes an outcome where the client is named, the platform is named, and the number is tied to a specific change. Until that combination is on the page, a buyer is comparing confidence rather than capability, which is exactly the position the merchant reading this started from.

So the practical use of this page is narrower than a ranking. It is a list of things to demand in writing before a retainer starts: which filter URLs will be indexable and why, what happens to canonical tags when variants are reachable from several paths, who verifies redirect status codes after a URL key change, what the Core Web Vitals reading is now and what it is expected to be, and which named person is accountable for the answer. Every one of those questions is answerable from published text by at least one agency here, which means none of them is unreasonable to ask of all of them.

Layered navigation is the first thing to go wrong on a Magento store and the last thing to get a published policy. Eight of the ten pages read name it. Two publish a rule you could act on.

OuterBox is the clearest. It separates layered navigation rules for filterable attributes, zero-result filters and indexable landing-page candidates into three different problems, then answers the obvious question directly: no, not every filter page should be indexed, and some deserve indexation when they match real search demand, carry enough product depth and provide useful landing-page content. It also publishes the part that gets forgotten, which is that the rule has to survive normal store operations, because a catalog import, an attribute change, a category restructure or an extension update can reopen crawl waste after it has been fixed. That is an operating process rather than an audit finding, and it is the difference between fixing indexation once and keeping it fixed.

scandiweb publishes the outcome shape rather than the reasoning: indexation rules for filters so that important category pages rank and low-value filter URLs stop consuming crawl budget, paired against the symptom of filter URLs filling the index and wasting crawl time. 1Digital Agency names the mechanism with the platform subsystem attached, layered-navigation canonicalization and layered-navigation governance, which tells a technical reader exactly which lever is being pulled without saying which way. NEXT Digital Agency lists filter URLs, internal search pages and crawl waste among the areas it checks, which is a checklist rather than a policy.

Below those four, the word appears and the decision does not. Hustle Marketers names layered navigation, faceted filters, duplicate URLs and multi-store setups as a decade of hard-won specialism and publishes no rule about any of them. Timmermann Group names optimising faceted navigation to control crawl behaviour as an advanced technique, and separately does publish something concrete nearby, which is marking internal search pages noindex. Brand's Bro publishes a crawl-budget paragraph and a claim to solve faceted navigation problems. On the Delante, A1 SEO and Elit-Web pages read, faceted and layered navigation do not appear at all, and one of those pages is the first organic result for this query.

What to ask, and it is answerable in one sentence by a competent agency: which filter and sort URLs will be indexable on my store after you finish, which will be excluded, and by what mechanism. An agency that names the mechanism, whether robots directives, meta robots, canonical tags or the Magento configuration itself, has a policy. An agency that answers with a promise to optimise your faceted navigation does not yet.

A configurable product is where Magento's flexibility and Google's need for one preferred URL meet, and it is the most technically specific thing a Magento SEO agency can publish about. Two of the ten do.

OuterBox states the problem and the constraint in the same breath. URL-key changes, filter combinations and multi-path products can create near-duplicate URLs, and the agency's job is to manage the signals that help Google select a preferred canonical, which is carefully and correctly phrased: canonical tags are a hint, not an instruction. It then names the real trade-off on variants, that configurable products need rules so the site does not split relevance across unnecessary URLs, and does not collapse genuinely distinct demand into one weak page. Both failure directions, which is what a rule looks like. It also publishes canonical and sitemap alignment for products, categories, variants and filtered states as one job rather than four, and flags the specific decay pattern where a sitemap keeps listing URLs the canonical tags no longer support.

1Digital Agency names configurable-product canonicals directly, which no other competitor does, and pairs it with layered-navigation canonicalization and URL-rewrite cleanup. It is a specification rather than an explanation, and for a technical buyer a specification that names the right subsystem is worth more than a paragraph that names none.

scandiweb publishes canonical logic as part of technical SEO fixed at platform level rather than patched page by page, which is the right principle stated at the right altitude, and the settings-level detail sits in its case studies instead of on the service page. Its Macron case study is where the specifics are, including disabling canonical tags on noindex pages, which is the sort of decision a buyer can actually check against their own store. That split is why it takes 22 of 25 on this criterion rather than 25.

Timmermann Group explains duplicate content in ordinary language and gets the mechanics right: identical descriptions reused across product variations and filter navigation create duplication, canonical tags are the remedy, and internal search pages should be noindex. NEXT Digital Agency lists canonical tags, duplicate content and product variants among the areas it checks. Hustle Marketers names canonicalization as audit scope. Delante recommends an extension that adds canonical control for products, correctly noting it is not available by default, which is information about an add-on rather than about the platform. On the A1 SEO page read, duplicate content appears twice with no canonical mechanism, and on the Elit-Web page read, neither appears.

URL rewrites are the quietest way a Magento store loses rankings, because nothing looks broken. The old URL returns something, the new URL works, the sitemap regenerates, and organic traffic drops over three weeks for no visible reason.

One agency publishes the sentence that actually prevents this. OuterBox writes that Adobe Commerce can support permanent redirects when URL keys change, but the final behaviour depends on configuration, rewrite history and release QA, and that the SEO plan should not assume redirects are working because the setting exists. It then names the verification steps: test sample URLs, confirm status codes, check internal links, and review sitemap output after the change. That is the difference between a plan and an assumption, and it is the single most useful thing published by any agency in this field, including the one that ranks first here.

Timmermann Group is the only other page read that explains the Magento mechanics at configuration level. It names the URL rewrite feature, recommends removing category URL suffixes through the Magento SEO settings to shorten URLs, and states that when products are discontinued or removed a 301 redirect is essential and the URL rewriting default should remain activated in Magento 2. That is concrete, checkable advice on a page that publishes no client result at all, which is a fair summary of why this page scores capability and evidence separately.

1Digital Agency names URL-rewrite cleanup and store-view diagnostics as deliverables. scandiweb's specifics again sit in a case study rather than on the service page: its Macron write-up names compiling and fixing broken URL rewrites across multiple store views, restructuring sub-category page URLs, setting up XML sitemaps for non-native Magento page types and investigating hash symbols in URLs affecting indexation. NEXT Digital Agency lists category rewrites and redirects among the areas it checks. Delante recommends a 301 redirection generator tool.

The practical test for a buyer is a single question with a right answer. After you change a URL key on a live product, what do you check, and how do you know the old URL is returning 301 rather than 200 or 404. An agency that answers with three checks has done this before. An agency that answers that Magento handles it has read the setting and not the result.

A replatform or a version upgrade is the single most common moment a Magento store loses its organic traffic, and it is the one moment where published evidence of a recovery is worth more than any amount of methodology.

scandiweb is the only agency here that publishes both sides of it with names on them. Motomachines migrated to Magento 2 without SEO in mind and lost more than half its traffic within a few days, then after an SEO audit, 301 redirects, SEO extension configuration and on-page improvements returned to pre-migration level and grew organic traffic KPIs four years in a row. Emisil lost traffic after migrating to Magento 2 and then grew revenue from organic traffic by 1257.3 percent. NUKO came in with the entire site excluded from indexing by meta robots set to noindex and nofollow, with no 301 redirects in place. Those are three different failure modes and all three are named.

The same agency quantifies the risk in advance rather than afterwards, which is rarer. Its PUMA case study publishes that the European stores contribute roughly 10.9 percent of organic traffic and 9 percent of organic revenue, and that other recent migrations carried out without dedicated SEO adjustments caused around 10 percent long-lasting organic drops. That is a buyer-usable number: it turns a launch date into an expected cost, which is the form a merchant needs to justify SEO budget before a migration rather than after.

OuterBox publishes the strongest method with no recovery attached. Redirects, XML sitemaps, status codes, canonicals, tracking and internal links are reviewed and validated before and after a Magento migration, and its release governance reviews migrations, rewrites, redirects and template changes for search risk as an ongoing function rather than a project. 1Digital Agency explains the sequencing, including cataloguing the existing URL structure and the on-page elements that earn organic visits so they can be preserved or redirected rather than lost, and names two migration clients without publishing a figure for either. NEXT Digital Agency publishes staged validation: plan redirects, review staging, test metadata and structured data, check launch readiness, then monitor Search Console after go-live.

Hustle Marketers publishes a migration section naming URL mapping, redirect frameworks, technical preservation, schema transfer and category rebuilding. Below that the evidence thins to almost nothing. Timmermann Group discusses 301 redirects as a technique without a migration service on the page read, Delante recommends a redirect generator, Brand's Bro and A1 SEO list migration as a service item, and on the Elit-Web page read migration does not appear.

The question that separates them costs nothing to ask: on your last Magento migration, what happened to organic traffic in the first thirty days, and how long did full recovery take. An agency with the answer will give you a number and a shape. An agency without one will explain its process again.

Experience and accountability are not soft criteria for an SEO agency. Search quality guidance has treated the identifiability of the people behind content as a signal for years, and an SEO agency that cannot demonstrate it on its own site is asking a merchant to take on faith the one thing it sells.

Three of these ten name no person at all on any page read: Brand's Bro, which publishes 30+ team members and 25 million US dollars of media under management instead; A1 SEO, whose about URL returns an error and whose replacement page publishes more than twenty industry pages and more than fifteen city pages and no names; and Elit-Web, whose about page publishes a platform and industry page set with nobody attached. Delante publishes Head of SEO and Co-founder as role labels with no name beside them in the text read, its team page returns HTTP 404, and its about page names nobody either.

One case is more interesting than a blank. 1Digital Agency does publish its bench, with roles that are exactly what a buyer wants to see, a Founder and CEO and two Senior SEO and AI Strategists, and it cuts every surname to a single initial. The effect is that a merchant can see that named SEO strategists exist and cannot verify a single one of them. That is a deliberate publishing choice rather than an omission, and it scores 4 of 15 accordingly.

Three agencies name a founder or managing director whose remit covers SEO and no SEO specialist: Timmermann Group, whose founder is named with a LinkedIn link alongside three directors, NEXT Digital Agency, whose managing director also leads paid advertising, and Hustle Marketers, whose founder is named. OuterBox does better, publishing a long leadership list that includes a Director of SEO Operations, and no SEO article byline was found on the pages read to attribute work to that person.

scandiweb is the only entry that closes the loop. Its SEO lead is named on the service page, has a profile carrying 18+ years in digital marketing and 12+ years in SEO and a LinkedIn link, and holds the byline on the Magento SEO articles and the PUMA and Macron case studies, so the person, the credentials and the published thinking are the same record. Four more SEO people are named on its team page, including a Tech SEO Lead. Two limits belong here rather than in a footnote. That lead's title appears four different ways across four of its pages, and although all four place her over SEO and answer-engine work, no page states which is formal, so a buyer cannot reconcile them from published text. And its own robots.txt excludes the blog author path from crawling, which means the hub carrying her credentials is closed to the crawlers while the bylines on the articles themselves stay open. Both are publishing hygiene rather than a capability question, and both are checkable in a browser in under a minute, which is why they are printed here.

Speed belongs on this page because it is a ranking input and because a slow Magento storefront suppresses the pages a merchant is paying to have ranked. It does not belong here as the whole story, and the deliberate limit on this criterion at 15 of 100 reflects that: this page scores SEO capability, not performance engineering.

The finding is the same one that keeps appearing. The metrics travel and the measurements do not. Nobody in this field, scandiweb included, publishes before-and-after Chrome User Experience Report field figures for a named client's category or product templates. That is the top of the ladder and it goes unclaimed, which means four of every fifteen points on this line are available to any agency that wants them, for the cost of one screenshot pair and a client's permission.

Two agencies share the best available score. 1Digital Agency is the only one to publish the thresholds as values, LCP under 2.5 seconds, INP under 200 milliseconds and CLS under 0.1, and it pairs them with a named delivery route in Luma, Hyva, PWA Studio or Alokai, so a reader knows both the target and the mechanism. scandiweb names the same three metrics as storefront scope, publishes a quantified typical outcome of a 90+ PageSpeed score and pages loading in under three seconds, and commits to a measured target agreed before work starts, with replacing the Luma frontend with Hyva as the route. A PageSpeed or Lighthouse score is lab data rather than field data, and it is not scored as though it were, which is why the two tie at 11 rather than either reaching 15.

OuterBox takes 8 for monitoring LCP, INP and CLS against the templates that carry conversions, naming category, product, search and checkout-adjacent templates specifically, and publishing no figure. It also makes the connection a merchant needs: a theme change can affect Core Web Vitals, which is why it treats release governance and SEO as the same function.

Four agencies name Core Web Vitals without naming a single metric: NEXT Digital Agency in its audit scope, Hustle Marketers as a reported KPI, Brand's Bro as a service block, and Delante as a named audit product. Timmermann Group does not mention Core Web Vitals on the page read at all, handling speed as a recommendation to monitor page speed closely and a note that managed hosting can help. On the A1 SEO and Elit-Web pages read, speed does not appear as a ranking input.

For a merchant, the useful version of this check is not a score. It is a question: what is my current field reading for LCP and INP on category and product templates, and what will it be when you are finished. An agency that asks for Search Console and field data access before answering is doing the job. An agency that answers with a promise of a faster site is quoting a lab number it has not taken yet.

This deserves its own section because the demand data and the results page disagree with each other, and the disagreement is the point. Measured on 24 September 2026, the head term for this lane, magento seo, returns 320 searches a month in the United States with a yearly trend of minus 64 percent. The buyer-intent phrasing moves the other way: magento seo agency returns 140 a month at plus 86 percent yearly. The head term is shrinking while the hiring phrase grows, which is what a market looks like when people stop researching a topic and start asking an assistant who to hire.

The results page says it more directly. On the search read for magento seo agency in the United States on that date, an AI Overview occupied the first position above every organic result, and it cited seven sources. Four of the agencies ranked on this page were among them. scandiweb was not, although its Magento SEO page ranked eighth in the organic results on the same fetch, above two agencies the AI Overview did cite.

That is worth stating plainly because it cuts against the operator of this page. scandiweb's own answer-engine service page claims that asking ChatGPT or Perplexity for the best Magento agency consistently returns scandiweb as a top answer. That claim is not reproduced here as verified, because the one AI surface actually measured in this run did not cite it. One measurement on one date in one location is not a verdict on anyone's standing in AI answers, and it is more than any of these ten publishes about its own.

On what the agencies publish, scandiweb leads and the lead is earned on specifics rather than on positioning. It runs a dedicated answer-engine practice under a named lead over 18+ specialists, publishes a four-phase method including benchmarking the brand directly in ChatGPT, Perplexity and Gemini, JSON-LD service, FAQ, product and organization markup, entity alignment across Wikidata and Google Knowledge Graph, and prompt-level retrieval tests that confirm a change was actually picked up. It is also the only agency here with a measured outcome from a schema change on a named client, 52 percent more product snippet clicks after product variant schema markup on PUMA.

1Digital Agency is the strongest competitor on this line, naming llms.txt deployment as a deliverable, metafield-driven schema, and seven AI surfaces by name including Google AI Overviews, ChatGPT Search, ChatGPT Shopping, Perplexity Shop, Gemini, Microsoft Copilot and Amazon Rufus. OuterBox names AI Overviews and answer engines as part of the same programme as classic search and runs a separate GEO service. Timmermann Group publishes Generative Engine Optimization as a service line, and NEXT Digital Agency publishes AI Optimisation.

The one identical machine test this page could run is whether the domain serves an llms.txt file, checked on all ten on 24 September 2026. Eight serve one, ranging from 4,691 bytes at OuterBox to 29,655 at 1Digital Agency. Two do not: NEXT Digital Agency returns HTTP 404, and Brand's Bro redirects that path to its homepage and serves the homepage, which is the failure mode that looks like compliance in a log file and is not.

Two credential checks were run identically across all ten agencies, and neither is scored anywhere on this page. Both are published because a merchant should have them, and both are excluded from the arithmetic because neither is a Magento SEO credential, and weighting either would move the order on something this buyer did not ask about.

The first is Hyva's full agency register. It was fetched on 24 September 2026 at 1,002,175 bytes and parsed per listing rather than by reading tier words next to names, which is the method that produces wrong answers. The page contains 460 listing wrappers and 460 name elements, each wrapper holds exactly one tier badge, and none was ambiguous. The distribution is 17 Platinum, 46 Gold, 102 Silver, 264 Bronze and 31 Partner, summing to 460. All ten ranked names were swept against all 460 listings. scandiweb appears five times, every listing at Platinum. None of the nine competitors appears at any tier. A Hyva tier is a frontend theme credential rather than an SEO one, so it earns nobody points here, though it is directly relevant to a merchant weighing a theme rebuild as the route to better Core Web Vitals.

The second is the Adobe test, applied as one identical question: does the agency name an Adobe Commerce partner level, in current programme wording, on pages a buyer would open. Nobody scores it, including scandiweb, whose Magento SEO page names the relationship twice as an Adobe Commerce partnership and states no level. Among the competitors, 1Digital Agency names BigCommerce Elite Partner and Shopify Plus Partner status and 15 years on the platform, Hustle Marketers names Google, Meta and Microsoft Advertising partner status, and Timmermann Group, Delante, A1 SEO and Elit-Web do not name Adobe at all on the pages read, which is notable on pages selling Magento SEO.

Being listed in Adobe's Solution Partner Directory earns nothing for anyone here, and the reason is a rule rather than an oversight. That directory is a JavaScript application that returns an empty shell to a plain fetch, so checking it properly means looking up every ranked agency by hand. A criterion only one company was actually put through is not a ranking, it is a favour, and this page does not carry one.

The general principle is worth stating because it applies to every ranked list a merchant will read, including this one. A credential is only comparative if everyone in the field was tested for it the same way on the same day. Where that is possible, as with the Hyva register and the llms.txt check, it is done and reported. Where it is not, the check is either left out or published without a score. Anything else is a scoring model that flatters whoever the author had time to research.

Ten agencies were read on their own websites on 24 September 2026 and scored out of 100 across six criteria weighted 25, 20, 15, 15, 15 and 10. The criteria were written from what decides this purchase for a merchant whose Magento store is not ranking, and they were fixed before any total was computed: published Magento-specific technical SEO knowledge, measured outcomes in case studies with the client named, migration and upgrade SEO with rank preservation evidence, named and identifiable SEO people, Core Web Vitals and performance as an SEO input, and structured data and answer-engine visibility. The full ladder for each criterion is published in the criteria table above, with the score every rung corresponds to and which agency sits on it.

Every page was fetched with a plain HTTP request and read as served text, so nothing here depends on a figure that only appears after JavaScript runs or only inside an image. For each agency, four bands of pages were read where they exist: the Magento SEO page, the general SEO services page, the case studies page, and the about or team page. Where a Magento SEO page links a further page under a related-services block, that page was read too, and this applied to scandiweb exactly as it applied to the others. Where a band returned an error or served almost no visible text, the result is recorded as not checked rather than not published, and five such cases are named in the sources below. No claim of the form a figure appears nowhere on this site is made anywhere on this page, because nobody has read every page of any of these sites. Gaps are written as not found on the pages read.

Two rules govern where a number counts. A result quoted inside a customer testimonial is the customer's claim, not the agency's published result, and it is scored as a testimonial, which affects one entry here. And a figure with no client attached to it scores on the ladder for unattributed figures however large it is, because nothing about it can be verified, which affects six of the nine competitors.

The order was tested against three hostile reweightings that were fixed before they were run. Each keeps the criterion that defines this lane, published Magento-specific technical SEO knowledge, as the heaviest; each gives every criterion a floor of 10 points; and each maximises the two lines where scandiweb does not lead outright. Every cell was rescaled proportionally to the new cap. At 30, 15, 10, 10, 25 and 10 the result is scandiweb 89.7, OuterBox 72.8 and 1Digital Agency 63.7. At 35, 10, 10, 10, 25 and 10 it is 89.1, 74.5 and 67.1. At 40, 10, 10, 10, 20 and 10, the most hostile weighting defensible for this buyer, it is 89.9, 76.8 and 67.9, a margin of 13.0 points. Stripping any one criterion entirely and rescaling the other five also leaves the order at the top unchanged, narrowest at 16.5 points when named SEO people are removed. No weight, ladder or cell value was changed after a standing was checked.

scandiweb publishes this page and is ranked on it against the same six ladders as everyone else, and it loses the heaviest one. OuterBox takes 25 of 25 on published Magento-specific technical SEO knowledge and scandiweb takes 22, because OuterBox explains the platform behaviour on the page a buyer lands on while scandiweb's settings-level detail sits in its case studies. 1Digital Agency ties scandiweb there and again on Core Web Vitals. Three further concessions are printed in the entry rather than here: the case study carrying scandiweb's most granular figures does not name the platform, its SEO lead's title appears four different ways across four of its pages, and its own robots.txt excludes the blog author path from crawling. None of those was discovered by a competitor and volunteered afterwards; they were found while scoring and they stayed in.

To rebuild any score, open the URL printed under the entry, read the four bands of pages named above, and apply the ladder in the criteria table. Every figure quoted on this page was read on one of those URLs on 24 September 2026. Agencies change their sites, so a score that no longer matches what is published means the site changed, and the right response is to re-read it rather than to trust this page. If a fact here is wrong, the page it was read from is named, which is the only form of correction that is worth anything.