Nobody chooses Airtable because it is cheap. They choose it because an ops lead can build something useful in a weekend. The bill becomes a problem later, for one reason: every layer of the stack charges you for growing.
I call this the success tax, and it has three layers. Airtable bills per editor seat, so every new hire who touches a record costs $20 to $54 a month. Make bills per operation, so every busier month adds to the credit count. The portal builder, whether Noloco or Softr, bills per external user, so every client you onboard adds to the invoice. Together they produce a bill that climbs in step with the business and never comes back down.
I have spent years building on Airtable, Make and Noloco, and they remain the right starting point for most small teams. But I have also moved a large project off that stack onto Supabase, Vercel and Next.js, so here are the real costs on both sides. All prices are USD list prices. A figure I read from the vendor's own pricing page on 15 October 2026 is marked verified; third-party figures and my own estimates are marked as such.
What the low-code stack costs at three sizes
Airtable prices are verified from airtable.com/pricing: Team is $20 and Business is $45 per user per month, billed annually. The monthly-billed figures of $24 and $54 are third-party, from Noloco's pricing guide. Noloco's own prices are verified from noloco.io/pricing: Build is $79 a month, or $65 annual, with ten team seats included and extra seats at $6. Make's official page is an interactive calculator, so every Make figure here is third-party; Jetadmin's September 2026 guide lists the 10,000-credit tier at $10.59 for Core, $18.82 for Pro and $34.12 for Teams. Softr's prices are also third-party, from Jetadmin.
| Scenario | Airtable seats | Make operations | Portal builder | Monthly total |
|---|---|---|---|---|
| Small: 10 Team seats, light automation, small portal | 10 × $20 = $200 annual, $240 monthly (verified) | Core, 10k credits, $9 to $11 (third-party) | Softr Basic $25 (third-party) or Noloco Build $65 to $79 (verified) | About $235 to $330 |
| Mid: 15 Business seats, about 100k operations, portal for 25 team and 50 to 100 external users | 15 × $45 = $675 annual, $810 monthly (verified) | Pro at about 100k credits, my estimate $90 to $180 | Softr Pro plus extra client users, or Noloco Build plus 15 seats, $140 to $170 | About $900 to $1,300 |
| Large: 25 Business seats, 500k to 1M operations, portal for 30 team and 200 external users | 25 × $45 = $1,125 annual, $1,350 monthly (verified) | Teams at 500k to 1M credits, my estimate $400 to $900 | Softr Business plus 100 users $530 to $600, or Noloco Build with add-ons $250 to $400 | About $1,900 to $2,900 |
Airtable seats dominate at every size. The Make figure for the mid and large rows is the weakest number here: nobody publishes the higher credit tiers, so check the Make calculator with your own operation count. List price also leaves things out. Two third-party guides put Airtable's automation cap at 25,000 runs a month on Team and 100,000 on Business, counted even when a run fails, so automation volume alone can force an upgrade. WitsCode's worked example shows an eight-user Business account at $432 a month growing to roughly $600 once automations, attachment overage and Zapier are added.
What the custom stack costs to run
Supabase and Vercel charge for compute and traffic rather than for the number of people using the app, so the bill is close to flat as headcount grows. Verified from supabase.com/pricing: Pro starts at $25 a month with a $10 compute credit, 8 GB of database and 100,000 monthly active users, with compute add-ons from Micro at $10 to Small at $15, Medium at $60 and Large at $110. Auth is included. Verified from vercel.com/pricing: Pro is $20 per developer seat per month with $20 of usage credit, and viewer seats are unlimited. The email, background job and monitoring figures are my estimates; I did not pull Resend, Inngest or Sentry list prices.
| Scenario | Supabase | Vercel | Email, jobs, monitoring | Monthly total |
|---|---|---|---|---|
| Small: 10 users | Pro $25 (verified) | Pro, 1 seat, $20 (verified) | Email $0 to $20, monitoring free tier (estimate) | About $45 to $65 |
| Mid: 25 internal and 100 external users, about 100k runs | Pro plus Small or Medium compute, $40 to $85 (verified) | Pro, 1 to 2 seats, $20 to $40 (verified) | Email $20, jobs $0 to $50, monitoring $0 to $30 (estimate) | About $100 to $250 |
| Large: 30 internal and 200 external users, about 1M runs | Pro plus Medium or Large compute, $85 to $135 (verified) | Pro, 2 seats, $40 plus overage (verified) | Email $20 to $50, jobs $50 to $100, monitoring $30 to $80 (estimate) | About $250 to $450 |
Put the two tables together and the gross monthly saving is roughly $180 to $270 for small, $700 to $1,100 for mid and $1,600 to $2,500 for large. If you need SOC 2 and SSO at the database layer, Supabase Team is $599 a month, but most businesses at this size do not. The question is what it costs to get there.
What it costs to build the migration
For an app with 10 to 30 tables and 5 to 15 screens, the published 2026 ranges cluster into four bands.
- Senior freelancer or small team. Project Cost Estimator puts an internal tool at $8,000 to $28,000 and a customer-facing portal at $18,000 to $55,000, at senior solo rates of $140 to $220 an hour.
- Agency. The same guide gives $25,000 to $70,000 for an internal tool and $50,000 to $140,000 for a portal; Kolonell quotes EUR 25,000 to 60,000 for a London professional-services portal. Most agencies publish no price at all: RapidDev, Vention, LOW/CODE Agency and Agence Scroll all route you through a free assessment first, so the number you see is the number they decide you can afford.
- In-house hire. A US full-stack developer runs about $120,000 to $157,000 base, roughly $160,000 to $205,000 fully loaded, or $13,000 to $17,000 a month. A three-month build is $20,000 to $50,000 of salary plus one to three months of recruiting, and then the person is still on payroll. This only makes sense if the migration is one of many projects.
- Fixed-fee specialist build. Nobody publishes a rate card here, so this is my estimate of the market, not my price list. Working back from 80 to 150 hours at senior rates with AI assistance gives $11,000 to $33,000 on time and materials, or fixed fees in the $10,000 to $30,000 band: between freelancer and agency prices, with agency-like scope.
The payback maths
Payback in months equals build cost divided by net monthly saving: the low-code bill minus the custom run cost minus a maintenance reserve of 18 percent of build cost per year, the middle of the published 15 to 22 percent range. Solving for a 24-month payback gives the low-code spend you need to be paying today.
| Build cost | Low-code spend needed for 24-month payback | For 12-month payback |
|---|---|---|
| $15,000 | About $1,000 per month | About $2,000 per month |
| $25,000 | About $1,500 per month | About $3,000 per month |
| $50,000 | About $2,900 per month | About $5,800 per month |
| $90,000 | About $5,100 per month | About $10,200 per month |
These are my estimates. The small scenario at $280 a month never pays back on subscription savings, whatever the build costs. The mid scenario at $1,100 a month pays back a $20,000 build in about 32 months, or about 20 months at 25 seats and $1,500 a month, while a $60,000 agency build at that spend effectively never does. The large scenario at $2,400 a month pays back a $30,000 build in about 19 months and a $90,000 build in about 129.
So below roughly $1,000 a month of low-code spend, about 15 Airtable Business seats plus a portal, the financial case is weak. There the decision has to rest on non-price triggers: automation run caps, a compliance or SSO requirement, a base that has slowed to a crawl, or wanting to own your data and code. Those are legitimate reasons. They are just not subscription savings. And because the low-code bill scales per seat and the custom stack does not, every ten additional Business seats adds $450 to $540 a month of saving, so model the headcount you expect in 24 months, not today's invoice.
Quick test: does a migration pay back for you?
- Add up your Airtable, Make and portal bills for last month, including overages. Is it above $1,000?
- Will your editor seat count or external user count grow by more than 50 percent in the next two years?
- Are you already hitting a hard limit, such as automation run caps, record limits, API rate limits, or a compliance requirement the current stack cannot meet?
Two yeses and the maths is probably on your side. One yes and the case rests on that reason alone. No yeses and you should stay where you are for now.
Hidden costs people forget
The build quote is never the whole cost. These are the lines missed most often.
- Data migration. An Airtable export is per-view and flat: linked records collapse to text, rollups become static values and attachment URLs expire within hours, as Tallyfy's guide lays out. Getting that into Postgres with proper foreign keys is typically one to two weeks of data engineering.
- Formula and automation translation. Every formula field becomes SQL or application logic, and every Make scenario and Airtable automation has to be rediscovered, documented and rebuilt. Budget another one to two weeks, and expect to drop automations that only existed because the old tool needed them.
- The parallel run. Both systems run side by side for two to six weeks while people learn to trust the new one, and you keep paying the old subscription throughout. Call it one extra month of low-code spend.
- Maintenance. Project Cost Estimator budgets 15 to 22 percent of build cost per year, separate from hosting; on a $25,000 build that is about $4,000 to $5,500. Retainers for a stable business application run EUR 1,500 to 6,000 a month in Kolonell's published SLA pricing. My own reading of the SMB market is $300 to $800 a month for a few hours of fixes and $1,500 to $3,000 for one to two days a month; call those estimates.
- Training. Half a day to a day per role, and a few weeks where nobody is sure which system is the source of truth.
How AI-assisted development changes the numbers
I build with AI assistance, so I have an obvious interest here, and the evidence is mixed. On the positive side, Anthropic's internal study of 132 engineers found Claude used in about 60 percent of their work and many self-estimating around 50 percent higher productivity than a year earlier, though that is self-reported and some said debugging time went up. Brilworks, a consultancy, reports roughly two times productivity on agent-eligible tasks, and WitsCode published a three-week Airtable to Supabase and Next.js rebuild where Claude Code handled most of the schema and CRUD work.
On the cautious side, the METR randomised trial from July 2025 found 16 experienced open-source developers were 19 percent slower with AI tools on codebases they knew well, even though they believed they were faster. GitClear's finding that code duplication rises with AI use suggests some cost shifts to maintenance.
Both sides are right; the difference is the setting. METR measured incremental changes to large, mature codebases. A migration off Airtable is the opposite: a greenfield rebuild of a CRUD application whose spec already exists in the working base. Schema, authentication, forms, list views and role-based access are where AI scaffolding is strongest. Data mapping, automation translation and cutover remain human work, which is where knowing Airtable and Make from the inside matters. A two to three times speed-up on the build phase is plausible for a specialist, not for a generalist dropped into a base they have never seen.
What matters for your budget is how that gain is priced. If a developer bills hourly and AI cuts a 40-hour task to 8, they lose 80 percent of the revenue, so hourly billing gives them no reason to pass the saving on. Fixed-fee pricing is how the client captures it: scope and price are agreed up front, and the speed-up shows up as a lower number than an agency quotes for the same outcome. No agency has published a rate-card cut attributable to AI; the practitioners who have repriced are the small fixed-fee shops.
My honest summary: under $1,000 a month with nothing breaking, stay on Airtable and enjoy it. At $1,500 a month or more, growing and hitting limits, a fixed-fee rebuild onto Supabase and Next.js should pay for itself inside two years and leave you with a bill that does not care how many people you hire. In between, a proper look at your actual seats and operations is worth an hour of your time.