Replacing Airtable with custom software makes sense when the number of people who edit grows with headcount, a base is approaching its record cap, and the workflows in it are core to how you operate; for a couple of dozen editors keeping planning or tracking bases, keep paying for Airtable. One comparison settles it: billed editors × seat price × 36, set against a build of engineers × months × $7,000 plus what it costs each year to host and maintain. On Business list price, our example has 300 editors breaking even in month 10, 100 editors in month 32, and 25 editors not at all.
Airtable bills editors, not viewers. On the Airtable pricing page, the Business plan lists at $45 per seat per month on annual billing, a list price we checked on November 21, 2026 (check your contract for yours), and only editors are billed: read-only users are free. That single rule shapes every replacement decision on this page, because the people who only look at data cost nothing on Airtable today.
List prices checked November 21, 2026. Check your contract; negotiated prices differ.
The second half of the bill is the plan ceilings. Per Airtable's plans overview, the record limit is cumulative across all tables in a base, and each plan sets its own caps. Because the count is per base and summed over tables, a base with an orders table of 60,000 records, a line-items table of 50,000 and a customers table of 10,000 already holds 120,000 records, within 5,000 of the self-serve Business ceiling even though no single table looks large. Append-only tables are what usually close that gap, and because the cap applies to the base as a whole, the practical choices are archiving old rows elsewhere, splitting the base, or changing plans.
The published ceilings by plan:
Four things usually change an Airtable renewal quote:
Airtable is a very good tool for teams that model their work themselves, and many should keep it. Owning software pays off for specific bases, not for the account as a whole.
In words: renting for three years costs billed editors × $45 × 36 at list. Owning costs the engineering build at $7,000 per engineer-month, the Airtable subscription you still carry while building and during a one-month overlap, and an ongoing hosting and maintenance figure that begins with the overlap. We call the break-even month the point where the running total for owning drops to or below the running total for renting and never climbs back above it.
Assumptions behind the rows (swap in yours):
Rows use the Airtable Business list price as of November 21, 2026 (list price, check your contract), engineers at $7,000 a month, and an assumed upkeep figure you should change.
Reading the middle row: three years of owning 100 editors' worth of software is 4 months of Airtable (4 × $4,500 = $18,000) + $63,000 build + 33 months of upkeep (33 × $2,100 = $69,300) = $150,300. Through month 31 owning is still behind ($139,800 versus $139,500 for renting); month 32 flips it ($141,900 versus $144,000), and from then the lead grows $2,400 each month. With 300 editors the flip comes between month 9 ($129,600 owned, $121,500 rented) and month 10 ($131,700 owned, $135,000 rented). With 25 editors Airtable costs $1,125 a month, less than the $2,100 of upkeep, so the owned line never gets under the rented one. That account should stay on Airtable or simply trim editor access.
Notice how narrow the 100-editor case is: one extra build month moves break-even to month 41, and $500 more upkeep a month moves it to month 39, both past the 3-year window. Try your renewal quote in our renewal vs build cost calculator. These rows mark where two cost curves meet under the assumptions listed; they say nothing certain about what your company will spend.
For each base, record the tables, cumulative record count against its plan cap, attachment storage, monthly automation runs and API calls. Next, pull the collaborator list and sort people with edit rights into those who change records and those who only look. Demoting lookers to read-only cuts the bill even if you never build anything, and gives you the honest editor count for the math above.
Exit check: a base register showing owner, distance from every cap, true editor count and a verdict for each base (stay, migrate, archive).
Airtable's views guide describes CSV download per grid view, on web and desktop only. A full base can't be exported as one file, so each table needs its own CSV, taken from an unfiltered view with no hidden fields, because filters and hidden fields limit what gets exported. CSV leaves out record-level comments, field descriptions, base guide content and extension-only data, and attachment files are exported separately.
For anything beyond a one-off copy, pull through the Web API. Authentication is a personal access token or OAuth token sent as a Bearer header; per the authentication docs, API keys stopped working when their deprecation ended on February 1, 2024. The list records endpoint pages with an offset cursor at up to 100 records per page; long queries go through POST to listRecords because URLs must be under 16,000 characters, and a 422 LIST_RECORDS_ITERATOR_NOT_AVAILABLE error means the iterator timed out and the pass has to be restarted. Per the field model docs, attachment URLs from the API expire 2 hours after they are returned, so download files in the same pass. At 100 records per page, a base at the 125,000-record Business cap needs at least 1,250 list requests, about 250 seconds at the 5 requests per second per base limit before any retries.
Exit check: record counts per table match the base, every attachment is in your object storage, and record comments you need are copied out separately or consciously left behind.
Linked records become foreign keys or join tables, single and multiple selects become lookup tables or enums, and formulas, lookups and rollups each get a decided home: a generated column, a view or application code. The rescue runbook walks through schema mapping, constraints and row-level security in detail, so we won't repeat it here. For a replacement driven by seats rather than an app backend, the extra work is in the next step: the screens editors use.
Exit check: the schema loads the full export with constraints switched on and no rows rejected you can't explain.
Begin with the interface, form or view the most editors use daily, since that is where the seats are. Each billed editor becomes a user with a role in your app; people who only read get a dashboard or report. Because readers are already free on Airtable, the business case rests on what editors do, so their screens need to be at least as fast to use as the grid they replace. Capture a real week of edits and make those the acceptance tests.
Exit check: the busiest editor group completes a full week of its work in the new app behind a feature flag.
Automations are rewritten as background workers with retries and logging. For integrations and the parallel run, Airtable's webhooks send a ping with only the base and webhook IDs; you then pull the changes from the list-payloads endpoint. Webhooks created with a personal access token or OAuth expire and are disabled after 7 days unless refreshed, and failed pings retry up to 13 times over about a day before notifications are disabled, so schedule the refresh and alert on gaps. Sign-in runs through your identity provider, and each edit writes an audit event.
Exit check: every automation has an owned equivalent, webhook refresh is monitored, and SSO is enforced.
Keep Airtable as the place editors work for one billing month while webhook-driven sync copies changes into PostgreSQL, then flip the direction per editor group. Reconcile record counts and key totals per table every day.
Exit check: a full month of daily reconciliations with zero unexplained differences.
Downgrade migrated editors to read-only ahead of renewal, so Airtable's next term counts only people still editing there. Take a final export and download every attachment. Know the recovery windows before deleting: per Airtable's trash documentation, deleted tables, views, fields and records stay in base trash for 7 days, deleted bases and workspaces stay in workspace trash for 30 days, and Enterprise Scale admins can choose between 30, 60, 90 and 180 days.
Exit check: the billed editor count has dropped, a last full export sits in storage, and the old base shows no edits after cutover day.
Our AI pod teams hand the tedious work to coding assistants: field-to-column mapping, tests built from a week of recorded edits, paged extraction scripts. Nothing merges without a human reviewer and a passing test suite.
Moving off Airtable usually takes four roles. A tech lead owns the base register and the order of cutover. Between one and three senior full-stack engineers build the editor screens and API. A data engineer joins when bases are near their caps or attachments run to many gigabytes. QA works part-time through the overlap month. Many teams begin with a pod of 3 to 5 senior engineers (a tech lead plus engineers), and a full pod is about seven. Moving a single base can start with a tech lead plus one or two engineers.
Ryz engineers on your team start from $7,000 per month (our published range is $7,000–$15,000 per engineer per month, mostly mid-level to senior), quoted per team. Once you say go, engineers can typically start in under two weeks. Ryz engineers work on your team, reporting to your leads. Want numbers for your bases? Scope your team.
Only if enough people edit. Readers are free, so count billed editors, multiply by your seat price and by 36, and set that beside the build cost at $7,000 per engineer-month, the subscription you carry until cutover, and yearly upkeep. With Business list price and our assumptions, 300 editors break even in month 10, 100 in month 32, and 25 not within three years.
Download a CSV per table from an unfiltered grid view, knowing that CSV leaves out record-level comments, field descriptions and base guide content and that attachments come separately. For a complete copy, use the Web API with a personal access token, paging 100 records at a time and downloading attachments within 2 hours of each response.
The drivers are how many bases move, how much behavior hides in formulas, rollups and automations, how many outside systems read or write the data, and whether the overlap has to span a quarter-end. One busy base and an entire workspace are very different projects, which is why we size them per base.
Engineers times months times rate, then a yearly upkeep line for hosting, storage and the hours someone spends keeping it healthy. Ryz engineers begin at $7,000 a month; we price the team once the base register exists, so the figure matches the bases being moved.
Many teams should do exactly that. Leave Airtable in place for planning and tracking bases where people change the structure themselves, move the high-volume or seat-heavy base to owned software, and return editors who only read to read-only access, which Airtable doesn't bill.
Questions we didn't answer? Email info@ryzlabs.com.
Thanks — your message has been sent. We’ll get back to you soon.
Something went wrong while sending your message. Please try again or email info@ryzlabs.com.