The number comes up on nearly every first call about a replatform, and it usually comes up sideways, as “we’ve had a couple of quotes” rather than as a question. The quotes are for the build. The build is the part that gets a line in the board paper, and in my experience it’s between a third and a half of what the next 2 years actually cost. So here is the whole number, as best I can put it together from what’s published and what I’ve sat through, and then the 3 things I’d want settled before anyone signs.
I’ve done this from every seat. At Missguided I ran the 2-year programme that moved the platform to a microservice architecture, on time and on budget, which I mention because it’s rarer than it should be. I replatformed Abbott Lyon recently, and more recently still I’ve rethemed several £50M+ retail brands in Shopify, which is the smaller job that most brands should be doing instead of the bigger one. I’ve advised brands mid-flight, Mamas & Papas among them, and I’ve been the person called in afterwards when the go-live has slipped twice and the board wants to know why. I wrote about why these projects drift last year. This is about the money, which is real, and about the thing that matters more, which is that a replatform takes a year of the business’s attention away from every other way it could be growing.
What it costs
The build. This is the quote on your desk. For a Shopify Plus build done properly by an agency at the top of the partner list, with a custom theme, your integrations, your data and your redirects, I’d expect £100,000 to £300,000 at this size. A headless or composable build runs £300,000 to £750,000 and takes most of a year. Moving off Adobe Commerce, Salesforce or SAP Hybris is at the top of those ranges because the data and the integrations are older and nobody documented them. If the quote in front of you is well under £100,000 for a £40M brand, somebody has left something out, and you’ll find out what in month 4.
The apps. Shopify’s model is a lean core and an app for everything else, which is cheaper than Magento until you count the apps. Subscriptions, reviews, loyalty, search, personalisation, email: at this size the stack runs £3,000 to £15,000 a month, and it’s often difficult to control because the cost is spread across marketing, ops and eCommerce, with nobody owning the total. It soon adds up.
The second year. The agency doesn’t leave at go-live. A retainer for a brand this size runs £8,000 to £25,000 a month, and the first 90 days after launch need a fixing budget of their own.
The drag. I called this launch drag in the earlier piece. Every month beyond the original go-live is a month of trading on a platform you’ve already decided isn’t fit for purpose, plus the roadmap that can’t start until the migration finishes. There’s a version of this that starts before the overrun, too, and anyone who has lived through a replatform will recognise it. Somewhere around month 3 the whole business starts saying “the new website will solve that”, and from that point nothing on the current site gets fixed, nothing gets tested and nothing gets improved, because why would you. For most brands I work with the drag runs £50,000 to £150,000 a month, and the typical overrun is a quarter.
Put together for a £40M brand: a £200,000 build, £80,000 of apps, £150,000 of retainer and a 3-month overrun at £75,000 a month, and the first year is £655,000 against the £200,000 the board approved. Write the business case against the bigger number, because that’s the one you’ll be paying.
The 3 things to settle before signing
What are you actually buying. The most expensive sentence in eCommerce is still “we want it to work exactly like the old site”, and it costs money twice, once in the build because every customisation Shopify didn’t want you to make is a day of agency time, and again for years afterwards because every workaround is something that can’t be updated without risk. The migrations I’ve seen land well went in the other direction. What did we build 5 years ago that nobody uses, what has a native equivalent now, what can we stop doing. If the board can’t say whether this is a like-for-like move or a simplification, it isn’t ready to sign, because the two projects cost different amounts and the agency will quote for whichever you ask for.
Who says no. The agency can’t, because every yes increases the project value and every no risks the relationship, and I say that as someone who ran one. Your Head of eCommerce is too close to the detail and your CTO too far from the trading. Somebody on your side of the table needs the job of holding the scope, translating the trade-offs into P&L and protecting the go-live date, and in most brands at this size nobody has it. It’s the role I fill most often on a replatform, and it costs a good deal less than the overrun it prevents.
What the new platform lets you do that you can’t do now. Most replatform business cases are written as a total cost of ownership comparison, which is a fair finance exercise, but the reason to do it should be a trading number. At ARNE the question that mattered wasn’t the platform at all but a licensed mobile app whose vendor was about to charge $600,000 a year to rent it, and the decision was to stop renting and build one the business owned. The Shopify work that ran alongside, moving to Store 2.0 and clearing the technical debt, was the clean-up around that decision rather than the reason for it. If you can’t say what the new platform lets you do in year one, in a number the CFO would recognise, then the platform probably isn’t your constraint, and the money is better spent finding out what is.
When not to
If the current platform is slow, ugly and expensive but the business is still growing on it, be careful. A replatform is a year of the whole team’s attention, and the brands I’ve seen regret it are the ones who did it because the agency said the platform was old rather than because they could name what it was stopping them doing. Old on its own isn’t a reason, and a number you can name is.
Questions people ask
How long does a replatform take?
Plan for 6 to 9 months on Shopify Plus at this size, 9 to 12 for headless or for a move off Adobe or Salesforce, and then add a quarter, because that’s what the overrun usually is. Don’t put Black Friday inside the window.
Should we go headless?
Only if you can say which part of the customer experience Shopify’s own storefront is stopping you building, and what it’s worth. Headless can cost a lot and adds a significant resource requirement, because somebody has to build and run it for as long as you have it. That is changing slowly as AI tools make it cheaper to build and deploy, but the use cases are still light, and for a business that is already growing and already at scale it should be treated as a pioneering move rather than an upgrade.
What does Centre Channel do on a replatform?
Three things, depending on where you are. A Board Session if the decision is whether to do it at all. The Diagnostic if you’re not sure the platform is the problem. And Advisory, in its programme shape, if the decision is made and you need someone on your side of the table until it lands.