When the interface stops being an app
Airtable Interface Designer is a good dashboard builder. It is a weaker app framework, and the line between those two things falls almost exactly where your users stop reading and start writing. Airtable prices that line directly. Per its permissions documentation, an interface-only viewer is not billable. An interface-only commenter or editor is. So the feature that turns your dashboard into a tool is the same feature that turns every user into a seat.
The read side is genuinely good
Interface Designer runs on every Airtable plan, including Free, with only the timeline visualization held back to paid tiers. You point a page at a table, pick a list or gallery or kanban layout, hide the forty fields nobody outside your team should see, and hand someone a screen that answers their question. Airtable's current-user filter is the part I keep recommending to people: add a condition where a user field is the current user, publish, and each person opens the same page and sees only their own rows.
For a review queue or a status board this beats anything you would build in a week of custom work. Read-only interface users do not count against your bill either, so a hundred stakeholders can watch a project dashboard on a Team plan and the invoice does not move. That is a real advantage and it is why so many operations teams stopped at this layer and never looked further.
I have never once told a client to rip out an Airtable dashboard that people actually open. If the thing works, it works.
Writing is where the pricing turns
Airtable's plans page lists Team at $24 per collaborator per month billed monthly, or $20 billed annually. Business is $54 monthly, $45 annually. On Team plans Airtable bills anyone with Commenter permissions or higher. On Business it bills Editor and up. The permissions chart marks interface-only viewers as not billable and interface-only commenters and editors as billable.
Run that against a write-heavy workflow. Forty warehouse staff who need to mark a shipment received are forty editors. On Team annual billing that is $9,600 a year for people whose entire interaction with your system is one dropdown and a timestamp. Move to Business for user groups or admin visibility and the same forty people cost $21,600. The dashboard that was free to read costs five figures the moment it accepts input.
Nothing here is a trick. Airtable publishes all of it, and per-seat pricing is the honest way to charge for a collaboration product. The problem is that it prices your users as collaborators when what you actually built is an application with an audience.
Validation you cannot write and permissions that leak
Cost is the part people notice. The part that bites later is that an interface cannot enforce a rule. You can set a field to view-only or editable, and you can hide records with a filter. You cannot say that a shipment must have a carrier before its status moves to dispatched, or that these two dates have to be in order, or that this number cannot exceed what the linked record allows. Airtable's own docs describe field visibility and record visibility as the two controls on offer. Neither is a constraint.
The permission model has edges that surprise people. Airtable documents that an interface-only editor can add new options to a single-select field without any base access, which is the opposite of base behaviour where only Owners and Creators can do that. It also documents that when a user's edit pushes a record outside the interface's filter, the user keeps access to it for five minutes. These are defensible product decisions. They are also the kind of thing you discover from a support article after an auditor asks.
Postgres handles this with a CHECK constraint and row level security, and the rule lives in a migration you can test and review. When we moved Stay World Class off Webflow and Xano onto Next.js, NestJS and Supabase, the four-stage ETL mapped every legacy ID to a PostgreSQL UUID precisely so the constraints had something stable to hold onto. Lighthouse went from 55.91 to 91 out of 100 and LCP dropped 77%, but the part the team noticed day to day was that bad data stopped arriving.
When the interface is the right answer
Stay on Interfaces if your users mostly read. A dashboard with a hundred viewers and three editors is close to the best deal in software, because Airtable charges you for three seats and the build took an afternoon.
Stay if the process is still moving. Writing validation rules for a workflow you will redesign in six weeks is wasted effort, and dragging a field in Interface Designer is faster than any migration you can write. Stay if the data is low stakes and a wrong row costs an apology instead of money.
Start costing a migration when your editor count climbs past roughly twenty, when someone asks for a rule the interface cannot express, or when a customer wants to use the thing and you realise every one of them would need an Airtable login. The first of those is arithmetic you can do today. The second two usually arrive together.
