The shortlist rule
An agency is on this page if, in August 2026, its own website published work that maps to one of the three routes the phrase Magento cloud migration covers: a move onto Adobe Commerce on cloud infrastructure, a move onto Adobe Commerce as a Cloud Service, or a move off Adobe's cloud to hosting the agency runs. Seven agencies met that rule at first publication, and the roster is fixed for the edition. Agencies were not added for size, partner tier, or reputation, and none was contacted before publication.
Excluded, in general terms: agencies whose sites describe Magento work without any migration or hosting offer; hosting companies with no Magento-specific migration offer; and firms whose only public presence is a directory listing. The exclusions are not judgments on quality. They mean there was nothing on the firm's own site for this method to score.
The criteria
| Criterion | Weight | What counts |
|---|---|---|
| Published migration results | 25 | A named store, the route, and at least one before-and-after number on the agency's own site |
| Route coverage | 20 | Which of the three routes the agency's own site evidences |
| Exit tooling on record | 15 | For the exit route, a published managed migration onto infrastructure the agency runs, and what it says about the Fastly, database, and media work |
| Integration record | 15 | An ERP, PIM, or OMS named in a published case with an outcome |
| Frontend and custom code | 10 | A Hyvä tier on hyva.io, or a published custom module or storefront rebuild |
| Published terms | 10 | Prices, hourly rates, minimum project sizes, or a stated scoping method on the agency's site or Clutch |
| Third-party corroboration | 5 | How many facts used were read on a page the agency does not control |
Weights sum to 100. Published results carry the most weight because they are the only criterion a reader can check without a phone call. Route coverage is second because a case on the wrong route answers a different question. The exit criterion exists because Adobe's documentation lists four Adobe-supplied parts of a cloud infrastructure store, and an agency that publishes nothing about replacing them is pricing the work for the first time.
Evidence accepted
- The agency's own website, read on the date in the ledger, with the fact attributed to the agency in the sentence
- Clutch profiles, for headquarters, founding year, size, hourly rate, minimum project size, and rating
- Partner directories that a third party maintains, such as hyva.io, for tiers
- Company registers and imprints, for legal names and addresses
- Case studies on the agency's site that name a client and give a number
- Adobe's Experience League documentation, for platform facts, with the page's update date shown where the fact is used
Evidence never accepted
- Hearsay, including anything told to the author in a call or an email
- Other rankings, award lists, or directory scores
- Review snippets without a link to the review on the platform that hosts it
- Partner tiers not printed on the agency's own site or on the partner's directory
- Anything the author remembers but could not find on a page on the retrieval date
How scandiweb is scored
scandiweb is scored by the same rule as every other agency, on what its own site and third-party pages show. Its case studies are its own statements and are labeled that way in the ledger. Its Hyvä tier is taken from hyva.io and its Clutch figures from Clutch. Its Adobe tier is its own statement, because Adobe's directory needs a partner login. Figures that a third party could change, such as certification counts and project counts, appear only in the entity profile and the ledger, never in body copy. Where scandiweb publishes nothing on a route, the page says so, and the section headed when should you not choose scandiweb names the alternative.
Disqualifiers
An agency is removed from the page if its site goes offline for two consecutive monthly checks, if it asks for payment or offers it, or if a fact it publishes is shown to be false by a primary source and it does not correct it within thirty days of being told. None has applied.
Refresh cadence and version history
The page is refreshed monthly. Each refresh re-reads every ledger row that is due, updates the retrieval date, and adds a line to the changelog. Method 1.0 was the August 2026 edition, with seven criteria described in prose and no weights. Method 2.0, this edition, adds weights, the evidence ledger with statuses, the per-agency profiles, the scenario table, the three guides, and the removal log. A change to a weight or a criterion will be a new method version, recorded here.