When the internal tool outgrows Retool
Retool charges for every person who opens the tool. On the Business plan that is $50 a month per builder and $15 a month per internal user, according to Retool's pricing page today. For an admin panel six people use, nobody notices. For a fulfilment console that 60 warehouse staff sit in front of all shift, the bill grows with the size of your operation instead of the size of the software.
The bill tracks your headcount
Retool defines a builder as any enabled user who built or edited an app or workflow during the billing cycle, and an internal user as one who did not. Team runs $10 per builder and $5 per internal user per month. Business runs $50 and $15. The app itself is free. What you are buying is permission for your staff to look at it.
That model is generous early and unforgiving late. Ten users on Team costs about $50 a month, which is less than the meeting where you debated building it. The same tool with 80 ops staff on Business, say six of them editing, lands near $1,400 a month before anything else. Nothing about your software changed. You just hired.
Workflow runs follow the same curve. Free includes 500 runs a month, Team and Business both include 5,000, and Retool sells more at $75 per additional 5,000 runs per month. A nightly job is nothing. A workflow that fires once per order is a line item that moves when sales go up.
Git shows up last on the price list
Retool's Source Control docs carry an Enterprise label. The feature does what you would want: branches, pull requests on GitHub, GitLab, Bitbucket, Azure Repos or CodeCommit, and changes reviewed before they reach the published app. It is the workflow any engineer already uses for everything else they ship.
Below Enterprise you get app release versions on Team and unlimited resource environments on Business, which cover rollback and staging but not review. If two builders touch the same app, the diff you want to read before approving lives behind a sales call with custom pricing. That is a reasonable business decision by Retool and an awkward one for a team whose warehouse console is now load-bearing.
The practical effect is that change control on your most operationally important internal tool is weaker than on your marketing site. I find that the detail teams notice latest and regret most.
You can test the screen, not the query
Retool's own testing guide points at WebDriver tests. You log a headless browser into your instance with a username and password, drive the rendered app with Cypress or Playwright, and assert against data test IDs like RetoolGrid:listView1. It works, and Retool documents how to run it against a branch preview in CI.
That is end to end testing, which is the slowest and most brittle kind. The JavaScript inside a query has no unit level harness in the product, so the only way to check a pricing rule is to boot a browser, sign in, click through the UI, and read what the grid says. When a rule has 14 branches you feel every second of it.
When Stay World Class moved off Webflow and Xano onto Next.js, NestJS and Supabase, the numbers people repeat are the visible ones: Lighthouse went from 55.91 to 91 out of 100 and LCP got 77% faster. The number nobody quotes mattered more to the migration itself. The business logic became functions a test file could call directly, which is the only reason a 4 stage ETL mapping legacy IDs onto PostgreSQL UUIDs was safe to run against real customer data.
When staying on Retool is the right answer
If fewer than a dozen people use the tool, stay. At Team pricing you are spending less per month than an hour of engineering time, and no custom app is cheaper than one you never have to deploy. The same goes for anything read heavy and low stakes, like a dashboard your finance lead opens on Mondays, or a console you built to survive a six month backfill and plan to delete afterwards.
Stay too if you have no one to maintain the replacement. A Retool app that Retool keeps running beats a Next.js app that nobody has patched since the person who wrote it left. That failure is quieter and worse.
The signal to move is narrow. User count climbing past a few dozen, revenue depending on the tool staying up, and logic complex enough that you are afraid to change it. When all three are true, per seat pricing and browser only tests stop being a fair trade. Until then Retool is doing exactly what you hired it for.
