09/14/2026
Migrate Now or Later? How to Decide
Should you make a big systems change now, or put it off?
Ask whether the shape is right before you ask how much the move will hurt. A migration costs you once; a structure that doesn't match how your business actually works costs you every week, for as long as you keep it. If the shape is wrong and the only argument left against moving is effort, move now.
Moving to a new platform, splitting one system into several, or consolidating tools always costs something up front. That sounds obvious written down. In the moment it isn't: the migration cost is concrete and easy to list, while the cost of the wrong structure is spread thinly across months of small frictions nobody adds up.
The argument I overruled
I run several websites. For a while they all lived in one shared code repository, which made one change easy to roll out to every site at once.
When I proposed giving each site its own repository, the AI assistant I work with argued against it. Its case was reasonable: five separate sets of dependencies instead of one, five upgrades to keep up with, and no more single commit that improves every site together. All true.
But almost all of that was effort, and effort was the wrong thing to decide on. The question that mattered was whether the structure matched how the sites actually live. Each one has its own audience, its own release schedule and its own reasons to change, and one shared repository tied every site's updates to every other site's.
I decided I'd rather take some pain now and get the structure right, and we split them.
A better way to weigh it
Before deciding, answer these in order:
- Do the boundaries match reality? Things that change for different reasons, on different schedules, for different people, usually want to be separate. Things that always change together usually want to stay together.
- Is the cost one-time or recurring? A week of migration work is a one-time cost. A daily workaround is recurring, and recurring costs win over any long enough period. (The login tax is one recurring cost you can put a number on in an hour.)
- What becomes harder to undo the longer you wait? More data, more integrations and more habits all build up on the current structure. Most migrations get more expensive with time, not less, which is one reason to buy software that has an API in the first place.
- What genuinely breaks? Name the real risks: lost data, downtime customers would notice, a team retraining mid-season. These are reasons to plan the move carefully, not reasons to avoid it.
- Is the argument against it still the original argument? If the structural case against a change has fallen apart and only “it's a lot of work” is left, the case has weakened. Say so, rather than finding a new reason to reach the same answer.
Now, later, or not at all
- The boundaries don't match how the business works
- Move. The wrong shape costs you every week you keep it.
- Only “it's a lot of work” is left against it
- The case against has weakened. Put a date on it.
- The right shape isn't clear yet
- Wait. Migrating to a structure you'll replace is paying twice.
- It collides with a launch, peak season or an audit
- Schedule the migration rather than dropping it.
- The pain is genuinely small
- Leave it. Not every imperfect setup is worth rebuilding.
“Later” is a legitimate answer. “Never, because it's hard” is not.
When “later” is the right answer
Waiting isn't always avoidance: the diagram above covers when it's the honest answer. What “later” shouldn't mean is “never, because it's hard.” Put a date on it, or decide honestly that the current shape is good enough.
Facing a move like this? Talk it through with me before the cost of waiting grows.
FAQ
How do I know if my current setup is the wrong shape?
Look for recurring friction: changes that always need coordinating across unrelated parts, one team waiting on another for routine work, or workarounds that have become normal.
Isn't it risky to take advice from an AI on decisions like this?
The AI's list of costs was accurate. It just weighed the wrong thing most heavily. Use it to surface the costs and risks, and keep the judgment about what matters to your business for yourself.
What's the biggest mistake businesses make here?
Treating the size of the migration as the size of the problem. The work is visible and finite. The cost of staying put is invisible and keeps growing.
Researched and drafted with AI, then reviewed and edited by the site's editors.