What to ask before you hire anyone to migrate your store
Twelve questions — including what a worrying answer sounds like.
We migrate Shopify stores to Horizon, so read this knowing we’re not neutral.
We’ve also written it so it would still be useful if you used it on us. These are the twelve questions that separate an agency that has done this from one that is about to learn on your store.
A good answer describes extraction, not conversation. The agency should be able to explain how they’ll inventory every section, setting, snippet, metafield reference, script and app integration directly from the theme files — and then have you review that inventory rather than write it.
A worrying answer is “we’ll walk through it with you.” That approach relies on your team remembering several years of accumulated changes, and nobody can do that reliably. The things people forget are precisely the things that go missing.
Ask to see a sample requirements document. Its existence tells you more than any answer to the question does.
A good answer is a written remedy with a defined boundary. Something like: anything that existed and worked on the previous theme and doesn’t work on the new one gets built at no cost, for a stated period after launch.
A worrying answer is reassurance. “We won’t miss anything” is not a commitment, and any agency that has done more than a handful of migrations knows better than to say it.
The reason this question matters is that it moves the risk of an incomplete discovery from you to them, which is where it belongs.
A good answer is that custom work goes into new files, never into core theme files. Horizon updates frequently. Any customization made by editing the theme’s own block and section files gets overwritten the next time an update lands.
A worrying answer is that they’ll “be careful” or that you can “just not take updates.” Declining updates means your theme drifts from its base version, which is exactly the position that made your current theme expensive to leave.
This is the most common complaint about Horizon in public reviews, and it’s mostly a consequence of how a theme gets built rather than the theme itself.
A good answer is that they’ll map it before you sign anything. Every app scored against Horizon — works as-is, needs reconfiguration, needs a vendor update, needs a custom rebuild, or has no path today — with a cost attached to each problem item and a named owner for fixing it.
A worrying answer is “we’ll handle it” or “we’ll cross that bridge.” App breakage is the single most common source of unexpected cost in a Horizon migration, and an agency that hasn’t looked doesn’t know.
Ask specifically about your subscription platform, your reviews app, and anything that was installed by pasting code into your theme. More on what breaks →
A good answer defines the boundary in writing. A fixed price should come with a documented scope, a stated freeze date, and a clear rule for what happens to requests made after it.
A worrying answer is a fixed price with no documented scope behind it. A fixed price is only meaningful if both parties agree what it covers, and that agreement has to exist as a document rather than as a conversation.
The useful follow-up: what’s the most common reason you’ve had to raise a change order? An agency with a real answer has done this enough times to know. Our pricing, published →
A good answer distinguishes between kickoff and the point at which the clock starts. Most migration timelines depend on the merchant approving a requirements document, and the merchant controls how fast that happens.
A worrying answer is a duration with no starting condition attached. “Six weeks” is not a commitment if nobody has said six weeks from what.
Expect to be asked for a commitment in return — a named approver and a review turnaround. An agency that asks for that is one that has been burned by not having it. How long a migration really takes →
A good answer is specific about three things: URL structure preserved, the existing redirect table carried over in full, and every structured data type currently emitted reproduced in the new build. On tracking, it should include verifying each event against a recording of the current store rather than checking it visually in preview.
A worrying answer treats both as a final-week task. SEO and tracking damage is invisible on launch day and surfaces weeks later, which is exactly why it needs to be scoped at the beginning.
A good answer includes a sandbox environment and a defined review window. You should be able to click through the entire new storefront, on real data, before anything is published.
A worrying answer is a demo or a walkthrough. Watching someone else use it is not the same as your merchandising lead trying to do their actual job in it.
A good answer is that it stays in your library, unpublished and unmodified. Rollback should be a single action, available indefinitely.
A worrying answer involves deleting or overwriting anything. There is no reason to remove the previous theme, and an agency that wants to is removing your safety net.
A good answer names roles and keeps them consistent. Ask who documents the requirements, who builds, who reviews the finished work, and whether any of it is subcontracted.
A worrying answer is vague about the reviewer. Someone other than the builder should check the finished theme against the requirements. If the same person does both, there is no review.
A good answer is acceptance criteria. Every requirement should have a specific, testable condition attached. Not “the cart drawer works,” but the version that states what happens, in what order, with what displayed.
A worrying answer is “when you’re happy with it.” That sounds generous and it is the direct cause of most launch-week disputes, because it means nobody agreed in advance what finished looked like.
A good answer is specific and immediate. An agency that does this work regularly has criteria — a store already current and unblocked, a business-critical app with no Horizon path from its vendor, complexity heavy enough that a rebuild costs more than it returns, or no internal capacity to participate.
A worrying answer is that there isn’t one. An agency with no circumstances under which it recommends against the work is selling rather than advising, and everything else it tells you should be read in that light.
This is the last question on the list because it’s the one that recontextualizes the other eleven.
What to ask for, not just ask about
Three artifacts tell you more than the answers do.
If it exists, they do this systematically. If they have to make one, they don’t.
Before-and-after performance, accessibility and SEO. If they can’t produce one, they’re not measuring outcomes.
In complexity, not in logo recognition. A famous brand in a different category tells you less than a smaller one with your app stack.
Frequently asked questions
What should I ask a Shopify migration agency before hiring them?
The most revealing questions are how they will document what your current theme does, what happens if they miss something, how they prevent theme updates from overwriting customizations, and what circumstances would make them recommend against migrating at all.
How do I know if a Shopify migration quote is realistic?
Ask what documented scope sits behind the number. A fixed price with no written scope is an estimate, and the difference surfaces as a change order once the work is underway.
Should I hire an agency or migrate to Horizon in-house?
In-house is viable if you have Shopify development capacity and can be disciplined about documentation and app compatibility mapping. Those two, rather than the building, are where in-house migrations usually run into trouble.
How much should a Shopify Horizon migration cost?
Published US agency ranges put mid-market Shopify migrations at $15,000 to $75,000 and custom Plus replatforms considerably higher. The more useful question is which category of work a given quote actually describes.
What’s the biggest red flag when hiring a migration agency?
An agency that can’t name a scenario in which it would advise against the project. It suggests the recommendation was made before the assessment.
Or start with an assessment instead of a quote.
Five days, a written evaluation of your store: what you have, what would break, what you’d gain, what it would cost, and whether you should do this at all. $2,500, credited in full if you go ahead — with us or, frankly, with the document in hand, wherever you like.
Our own answers to all twelve questions are on the migration page and in our published pricing.
