What end-of-life actually means in the room
Adobe has been clear about the date. The room is less clear. The CFO has heard "EOL" three times in three years and discounted it. The CTO knows the truth and has been told to make the case again with a tighter number. The ops director wants to know whether the trade portal still works the morning after.
That is the audience for this decision, and the vendor matrix on slide six does not speak their language. The choice is not Adobe vs Shopify vs commercetools. The choice is which migration path survives contact with the catalogue, the ERP and the contract pricing logic that the business has spent a decade tuning.
In rescue work we see the same pattern. The EOL date is treated as a deadline for a tender, not a deadline for a delivery. The tender lands in autumn, contracts sign in spring, discovery starts in summer, and the real implementation gets nine months to do eighteen months of work. The migration path you pick has to assume that timeline, because that is the timeline you will get.
Three honest paths, and where each one breaks
There are really three migration paths worth taking seriously, and each one breaks in a specific place.
Stay on Adobe Commerce, move to the cloud SKU and accept the upgrade tax. This is the lowest-risk option for businesses with heavy B2B customisation. It also has the worst long-term economics if your storefront extensions are old and your front end has not been touched since 2021. The break point is the front end, not the platform.
Move to Shopify Plus with a B2B build. The licence and operating model are attractive and the merchant tooling is mature. The break point is trade-account credit, multi-warehouse stock allocation and ERP-driven pricing at any real depth. Teams underestimate how much of their current Magento behaviour is a custom module that they will have to rebuild as a Shopify Function or an app.
Move to a composable stack with a separate PIM, search and CMS. This is the right answer for a minority of businesses and the wrong answer for most. It is right when the catalogue is large and structurally complex and the in-house team can run a vendor-managed integration estate. It is wrong when the original problem was that nobody had time to maintain one platform.
"The cost of a replatform is not the new licence. It is rebuilding ten years of pricing logic that nobody documented properly and one person still remembers."
The numbers most plans get wrong
The line item that wrecks most replatform budgets is integration, not build. ERP connectors, OMS handoff, tax engines, punchout, single sign-on and the freight rules that nobody wants to own again. On larger catalogue projects this is routinely sixty per cent of the real cost and it is almost always estimated at thirty.
The second line item is data. Customer accounts, saved lists, contract pricing, historical orders and the tax history that finance needs for audit. Migrating this cleanly is a project on its own. Migrating it badly is a six-month tail of support tickets that drains the team that was supposed to be building the next phase.
The third is the people cost during the run-in. A long parallel run on two platforms doubles the operational load. Most teams cannot sustain that for more than a quarter without something else slipping.
What to do this quarter, not next year
If the decision is still open, the most useful work right now is not a vendor selection. It is a clean inventory of the customisations you actually use, the integrations you actually rely on and the trade-account rules that actually drive revenue. Most teams discover that a third of what they thought was load-bearing is not, and that a small set of things they had stopped thinking about are.
Pair that inventory with a hard look at the front end. If the storefront is already a separate head, the platform decision narrows considerably. If it is not, the platform decision is also a front-end decision and the budget needs to reflect that honestly before the tender goes out.
EOL is not the event. The decision you make in the next two quarters is. The platforms that survive contact with reality are the ones where the choice was made by the people who will have to run it on the Monday after launch.
