Airtable is the best thing that ever happened to the operations spreadsheet. It gives a team structure, views, permissions of a sort, and a way to stop emailing versions of a file around. Most businesses running on it should stay on it. This page is for the ones where it has started creaking, and for the specific question of how to tell creaking from properly broken.
One of these is normal. Three at once and the arithmetic has usually already changed.
When the business logic lives in a chain of automations nobody can read end to end, you have written software without any of the tools for maintaining it.
Hiding a field is not access control. The moment somebody should not be able to see a number rather than merely not shown it, you need enforcement underneath.
Base limits, slow filtered views and syncs that take minutes. Performance work inside a hosted base has a low ceiling and you will find it.
Once you are syncing between bases to keep one operation consistent, you are maintaining a distributed system with none of the guarantees of one.
Pricing, approvals and statuses applied rather than remembered, so the process cannot be worked around by someone in a hurry on a Friday.
Enforced on the server on every request, and evidenced when a customer or an auditor asks who could see what, and when they could see it.
Who changed what, when, and what the value was before. This is the thing you will wish you had the first time an invoice is disputed.
Accounting, telephony, payment terminals and hardware, wired in properly rather than held together by a third-party automation service.
If your team is small, your records number in the thousands rather than the millions, and the rules are ordinary, Airtable is doing a good job and replacing it is a bad use of money. The same is true if what bothers you is the interface rather than the model underneath. A custom front end over your existing base is a much smaller project than a custom system, and we will suggest that when it fits. The honest test is whether the problem is how it looks or what it cannot enforce.
You do not have to move everything at once, and you usually should not. Pick the single process that hurts most, the one with the pricing rules or the approvals or the compliance reporting, and build that. Leave the rest in Airtable. If the new system reads and writes to the same records through an API, the two can coexist for a year while you find out whether the second piece is worth building. Migrations that move an entire operation in one go are where budgets go to die.
Yes, and it is often the cheapest useful step. Airtable stays the store, the custom app enforces the rules and gives people a proper interface. It buys you most of the benefit for a fraction of the work, and it leaves the door open to moving the data later.
No. Migrating the existing records is part of the work rather than an extra line on the invoice. We map the base onto a relational schema, run the import against a copy, and reconcile the record counts and the totals with you before anything at all goes live.
A focused replacement of one process is usually weeks rather than quarters, and we aim to put something usable in front of the people doing the work early. An entire operation is a bigger question, and the honest answer to it comes after a scoping call.
Airtable is a monthly cost that scales with seats and never stops; a build is one price up front and then mostly stops. Where they cross depends on your headcount, and a single process replacement crosses sooner than a whole operation. You get the build number at discovery, before anything starts, so you can do that arithmetic yourself.
Describe what you’re building, what it has to do and when it needs to be live. You’ll get back the architecture we’d use, the trade-offs, and a realistic timeline, before anything is signed.