Having a rollback plan for your website's launch day

What a rollback plan actually is

You've been told launch day will go smoothly, and in most cases it does. A rollback plan isn't a sign that something's expected to go wrong. It's a short, written answer to one specific question, agreed on in advance: if something does go wrong, what happens next, and who gets to decide?

Without that answer settled ahead of time, launch day problems have a way of turning into launch day arguments. Someone has to decide, in the moment, whether a broken checkout page is bad enough to reverse the whole launch. That's a worse time to figure out the process than the calm week before it, when nobody's watching a live error message.

Think of it as insurance you hope never to use. Writing it costs an hour. Not having it, on the day it matters, costs a lot more than an hour.

The rollback plan fits into a larger picture, covered in full at how to redesign a website without losing what works, where this is one piece among several that all fit together.

Reverting DNS back to the old site if needed

Mechanically, reversing a launch means pointing DNS back at wherever the old site was being served from before the switch. If that server or hosting account still exists and hasn't been shut down early, the technical reversal can happen fast once someone decides to trigger it.

The part that takes longer is everything around that decision. Confirming the old site still functions correctly. Telling the right people. Working out what actually happened on the new site in the meantime, so the same mistake doesn't repeat on the next attempt.

DNS reverting is the mechanically easy part. Deciding to pull that lever is the hard part, and it's the part worth planning for on paper, not in the moment.

Keeping a backup of the old site available, not deleted

This only works if the old site, or a full backup of it, still exists somewhere after the new one goes live. That sounds obvious written down, but old hosting accounts get cancelled the week after launch more than people expect. It happens to save on a bill nobody's actively thinking about anymore, once the new site looks like it's working fine.

Ask directly: how long does the old site or its backup stay available after launch? A week is a very different answer than a single day. Either can work. Just know which one you're getting before launch, not after it's already too late to matter.

Put the answer in writing. A verbal promise made during a busy planning call is easy to forget by the time anyone needs it.

Defining ahead of time what counts as 'go back'

Not every launch-day hiccup deserves a full rollback. A typo in the footer doesn't. A broken checkout page probably does. A contact form silently failing to deliver messages almost certainly does too, since it can go unnoticed the longest.

Writing a short list of what would actually trigger a rollback, before launch day, while everyone involved is calm, turns a stressful judgment call into something closer to a checklist. If this specific thing happens, revert. If that happens instead, fix it live and move on. The list doesn't need to be long. It needs to exist before you need it.

A short list also stops one anxious voice from triggering a rollback over something minor, and stops one confident voice from ignoring something that actually matters.

Who makes that call on launch day

One person should hold the authority to say "revert it" without first needing to schedule a meeting. On a small team, that's often the business owner directly. On a larger project, it might be whoever's technically running the migration itself, since they can see the problem fastest.

What matters most is that everyone agrees, before launch, on exactly who that person is. Deciding this in the moment, with a broken site already live and customers starting to notice, adds a delay nobody wants sitting in the middle of a real launch-day problem.

Name the person out loud, in the plan itself, not just in someone's head. It should be obvious, on launch day, who picks up the phone.

Communicating a rollback to customers if it happens

If a rollback does happen, a short and plain message beats total silence every time. A note on your social accounts, or a pinned message saying the site's back to normal while an issue gets sorted, is usually enough on its own. Customers tend to care less about the hiccup itself than about whether someone is visibly handling it.

Having that message loosely drafted ahead of time means you're not writing it from scratch while also juggling everything else a rollback involves in the moment. A few sentences saved as a draft is all it takes.

None of this is meant to make launch day sound riskier than it is. Most redesigns launch without needing any of it. The plan exists so the rare case that does need it isn't also the case nobody prepared for.

This plan works as a contingency for the normal sequence a launch is supposed to follow, laid out in full at what happens on website launch day. It also protects specifically against the downtime scenario explained at will your website go down during a redesign.

If you're still finalizing the redesign contract, this is exactly the kind of thing worth raising early, before anything is signed. See questions to ask before you commit to a website redesign for the rest of that list. A short conversation now saves a harder one on launch day itself.

More from the blog

All posts