Ecommerce Migration Downtime: The Cutover Plan
A well-run ecommerce migration takes the storefront offline for close to no time at all. That is the honest answer, and it usually surprises people. Both the old platform and the new one are live, hosted services. On cutover day the domain stops pointing at one and starts pointing at the other — nothing is ever switched off.
So the real cost of ecommerce migration downtime sits somewhere else: in the freeze window on the admin side, in the hours when DNS has not finished propagating and both platforms are still taking orders, and in the stock counts that drift out of date between the final export and go-live. That seam is where orders fall through. Not an outage.
Here is what actually happens during a cutover, in the order it happens, and which parts deserve the anxiety.
What ecommerce migration downtime actually means
Three different things get called downtime, and conflating them is why migration plans end up protecting the wrong thing.
| What it is | How long it lasts | What it actually costs |
|---|---|---|
| Storefront unavailable — shoppers cannot browse or buy | Minutes at most, often none | Direct lost sales. This is the fear everyone arrives with |
| Admin freeze — you stop editing the old system | Hours to days | No price changes, no new products, merchandising on hold |
| Data currency — the new store ages against reality | The length of the freeze | Overselling, missed orders, wrong stock. The expensive one |
Only the first is downtime in the sense people mean, and it is the one you are least likely to suffer. The third is the one that costs money, and almost nobody plans for it explicitly. Getting the sequence right is a separate question from whether to move at all — if you are still at that stage, our piece on the signs it is time to replatform is the better starting point.
Why DNS is the least risky part
Every DNS record carries a TTL — how long resolvers may cache it before asking again. Lower yours to a few minutes at least 24 to 48 hours before cutover, so the real change is picked up quickly rather than answered from a day-old cache.
Even then, propagation is not instant. Shopify’s own guidance on connecting a domain is that DNS changes usually take effect within two hours but can take up to two days. During that window your traffic is split: some resolvers hand visitors the new platform, some are still holding the old record.
Which leads to the rule that saves the most grief: do not shut the old platform down at the moment of the switch. Leave it running, and in most cases leave it able to take orders, for at least as long as the old TTL. A handful of orders landing on the old system that you reconcile by hand is a nuisance. An old system returning errors to a third of your traffic for six hours is real lost revenue.
One thing genuinely is asymmetric, though, and it catches teams out. Shopify issues the TLS certificate for a custom domain only after the A record and CNAME already point at Shopify, and says it might take up to 48 hours. So there is a window on the new side where the domain resolves to your store but HTTPS is not yet trusted. Confirm the domain shows as connected and the padlock is there before you announce anything, and time the switch so that window falls in quiet hours.
The freeze window is the real decision
A freeze is the point after which nobody touches the old admin. It begins at the final data export and ends when the new store is authoritative. Everything that happens in between has to be reconciled by hand, which is why its length is the single most consequential choice in the plan.
Long freeze, single export
Export once, freeze everything, launch. Simplest and safest, and it suits a modest catalogue, predictable order volume, or B2B trading that follows business hours. The cost is a day or two of stasis: no merchandising, no price moves, and orders arriving into a system you are about to retire.
Short freeze, delta sync
The bulk of the data moves days ahead; only the delta — orders, customers, stock and price changes since that bulk export — moves at cutover. The freeze shrinks to hours. It costs more engineering, and it needs a full rehearsal against a copy of the new store, because a delta process is one more thing that can be silently wrong.
The question that picks between them is not technical: what does an hour of frozen trading cost you, against the cost of building and testing a delta process you will use exactly once? Below serious order volume, the long freeze usually wins on total risk. Our free Shopify migration checklist sets out the sequencing either way, and what a Shopify migration costs covers where the engineering hours actually go.
What has to move at the last minute
Whichever freeze you choose, a specific short list is stale by launch and has to be dealt with at the cutover itself.
- Orders placed during the freeze. They need to exist in the new system, or at minimum be fulfillable from somewhere your team can see. Worth knowing before you plan this: migrated order history arrives as a record, not a live order — what does and does not transfer in a data migration is the detail behind that.
- Inventory levels. The one that bites. Any count exported before the freeze is wrong by go-live, and the failure mode is overselling on day one. Re-sync from whatever actually holds the truth — your ERP, 3PL or warehouse system — at cutover, not before. If that sync is itself new, it is the piece to rehearse; wiring it up properly is integration work in its own right.
- Customers created during the freeze. Small in number, awkward to explain to the ones affected.
- Price and promotion changes. The reason the freeze exists at all.
- Refunds and returns in flight. These usually stay in the old system to be completed there. Decide it explicitly and tell your support team, or they will find out mid-conversation with a customer.
Redirects go live with the switch, not after
The redirect map is a launch-day artefact, not a post-launch task. Every hour it is missing, Google and every backlink you own are pointed at 404s on a domain that is otherwise working perfectly — which is a worse signal than a site that is briefly down.
The good news is that it can be staged entirely ahead of time. Shopify’s URL redirects are defined by path rather than by domain and can be bulk-imported, so the whole map can be loaded and verified against the store’s .myshopify.com address days before the domain moves. Building that map correctly — what to map, in what priority, and how to verify it — is its own discipline, covered in how to migrate without losing SEO.
The first 48 hours
Cutover is not finished when the domain resolves. It is finished when you have proved the new store does the whole job, in this order:
- One real order, end to end. Place it yourself with a real card. Payment captured, confirmation email received, order visible to whoever fulfils it, tax and shipping correct. Everything else is theory until this works.
- Transactional email. You are sending from a different platform now. Check confirmations arrive, are not filed as spam, and that your sending records name the new sender.
- Inventory against reality. The first live sync with your warehouse or 3PL is where overselling shows up. Look directly rather than waiting for a customer to find it.
- Search Console. Submit the new sitemap and watch the coverage and 404 reports. Gaps in the redirect map surface here within days, while they are cheap to fix.
- Analytics continuity. Confirm the tag is on the new theme and conversions fire, so next week’s comparison reflects the business rather than a missing script.
- The old platform. Check what is still landing on it before deciding it is safe to retire.
Rollback deserves one honest sentence. Once real orders have been placed on the new platform, rolling back is not a revert — it is a second migration in the opposite direction, with a fresh delta to reconcile. The genuine go or no-go decision point is before the DNS change, which is where that review belongs, not on launch night at eleven o’clock.
Frequently asked questions
How much downtime does an ecommerce migration cause?
For the storefront, typically none. Both platforms are live throughout and the domain simply changes where it points. What you should budget for instead is the admin freeze — the period when you cannot publish changes to the old system — which runs from hours to a couple of days depending on how you handle the data delta.
Can we migrate without freezing the store at all?
Not meaningfully. Something has to be the last export, and anything changed after it has to be carried across by hand or by a sync process. You can shrink the freeze to a very short window with a delta sync, but you cannot remove it, and a plan that claims to has usually just moved the reconciliation somewhere less visible.
When is the best time to cut over?
Your genuine trough, which is a question for your own analytics rather than a rule of thumb. Then work backwards: TTL lowered 24 to 48 hours before, and the freeze starting early enough that the final export is unhurried. Avoid cutting over just before a promotion or a peak trading period — you want a few ordinary days to find problems.
What happens to orders placed while DNS is still propagating?
They land wherever that shopper’s resolver sent them, which during propagation means some on each platform. This is normal and manageable, provided the old platform is still accepting orders and someone is watching both queues. Reconcile the old-side orders manually once the switch has settled.
The takeaway
Ecommerce migration downtime is mostly a misnamed fear. The storefront rarely goes dark; what goes wrong is the seam between two systems that are both briefly real — a freeze that ran longer than planned, stock counts exported too early, an old platform switched off while a third of the internet was still pointed at it. Plan the freeze deliberately, keep the old system alive through propagation, load the redirects before you switch and prove one real order afterwards. That is the whole job. If you would rather have it run by people who have done it before, our WooCommerce to Shopify and Magento to Shopify migration services cover exactly this sequence.
Current store performance