Headless commerce is a migration problem
The new storefront is the part everyone looks at. Whether the replatform succeeds is decided by redirects, catalogue data and the ERP integration.
Headless commerce gets sold on performance and design freedom. Both are real. Neither is why replatforms fail, and neither is where the budget goes.
What decides the outcome is whether you can move nine years of accumulated reality onto the new system without dropping any of it. Indexed URLs with backlinks. Product data with fields nobody documented. Customer accounts, order history, subscriptions, loyalty balances, tax configuration per market.
Going headless also means you now own things the platform used to own. Search, routing, caching, image handling, the middle layer between storefront and back office. That is the trade: more control, more surface area, and a permanent maintenance obligation you should budget for.
Which is why we push clients to keep the boring parts on the platform. Payments, tax calculation and fraud screening are enormous problems with regulatory exposure, and Shopify solves them better than a custom build will. Take the front end; leave the money handling alone.
The test we apply before recommending headless at all: is the storefront actually the constraint? If campaigns are slow because merchandising needs a developer, headless helps. If they are slow because the catalogue data is a mess, headless changes nothing and you have added a system.