Analyse an Airtable Base Without Building a Single Formula Field

Airtable is where the ops team lives. It holds your client list, your project board, your hiring pipeline, your content calendar, and the spreadsheet-but-fancy that someone built last March and nobody has dared touch since.

It is also where the actual answers about how your business is doing are hiding — buried under linked records, lookup fields, formula columns that no longer compute, and views you can't combine.

You don't need an analyst. You need someone who can read the base and answer the question.

Why Airtable is great for storing data and awkward for analysing it

Airtable is the best-in-class operational database for teams who don't want a database. The interface is forgiving. The schema is flexible. New columns don't need a migration.

That same flexibility is the analysis problem.

• Linked records are the analysis ceiling. Want to ask "how many active projects per account manager"? You need a rollup on Accounts that counts linked Projects filtered by status. That formula takes ten minutes to write, breaks the next time someone renames a status, and only answers that one question.
• Cross-table joins are formulas, not queries. You can't say "show me clients with no project in the last 60 days." You build a rollup. Then a lookup. Then a formula. Then a view. Then someone deletes the rollup field and the view breaks.
• Filtered views don't compose. You can build a view for "active clients" and another for "projects in progress." You can't ask "active clients with no projects in progress" without a third view, plus a formula, plus a hidden field.
• No time-aware analysis. "How long does a project sit in each stage on average?" requires logging status changes in a separate table and building rollups against it. The base doesn't track it for you.
• Scripting blocks need a developer. They work. But the moment you reach for one, you've left the no-code zone Airtable promised.

The traditional escape hatch is paid Sync to a warehouse, plus a BI tool, plus someone who knows SQL. Most agency owners and ops leads have neither the budget for that stack nor the patience to babysit it. So the questions don't get asked, and the base accretes views and formula fields that mostly answer the question your team had eighteen months ago.

How to connect Airtable to an AI analyst

Connect Anna to Airtable via OAuth. Pick the workspace and the bases you want to share. Confirm read-only access. Done. Sixty seconds.

Once connected, Anna can read every base you grant — tables, fields, records, linked records, attachments, comments, views, and metadata. She uses the same Airtable API the platform uses, with no rate-limit issues you'd notice.

You ask the question. She picks the tables, follows the links, runs the analysis, and writes the answer.

A 25-minute ops review in five questions

The questions an ops lead, agency producer, or operator actually needs answered.

1. Pull the project board, find the bottleneck

"Pull the Projects table. For every project closed in the last 90 days, calculate how long it sat in each stage. Show me the average days per stage."

Anna reads the Projects table, follows the Status history (or computes from creation dates and modification timestamps if you don't keep one), and returns the durations.

This single question rewrites how most agencies plan their next quarter. The team thinks "we need more designers." The board thinks "we need to fix the client review handoff." The truth is sitting in the base.

2. Find the loss-making clients

"Join the Clients table to the Time Tracking table and the Invoices table. For each client, sum hours logged and revenue invoiced this quarter. Group by profitable, marginal, loss-making — using $100/hr as the break-even."

The base has the numbers. Nobody has built the rollup that surfaces this because doing it in Airtable means three lookups, two formulas, and a master rollup field that nobody trusts.

This is the single most important conversation any agency owner has each quarter. Most never have it because the data exists in the base and the analysis does not.

3. Find the stale pipeline

"From the Sales Pipeline base, find every deal that hasn't been updated in 30 days. Group by stage and by deal owner. Show me the totals."

Sales pipelines accumulate dead deals like a fridge accumulates condiments. Anna writes the clean-up list. The pipeline reverts to something the team can trust.

4. Pull the hiring pipeline, ask the question nobody wants to ask

"Open the Hiring base. For every role we've opened in the last 12 months, calculate days-to-fill, candidates-per-role, and offer-acceptance rate. Surface any role with more than 60 days open or under 30% offer-acceptance."

These are the metrics every head of people knows in theory, tracks in practice with a status column in Airtable, and never rolls up because the rollup formula would touch four linked tables.

5. Audit the base itself

"Audit the Projects base. Which records have no client assigned? Which have a status of 'Active' but no activity in the last 30 days? Which fields are >50% empty? Which views show the same filter as another view?"

This is the question every ops lead asks themselves every six months and never gets around to answering. Anna runs the audit in one sentence.

Joining to other tools

The real lift comes when Airtable becomes one piece of a multi-source picture. If your other tools are connected:

• Airtable + Stripe. Match clients in the base against Stripe customers. Surface clients with active engagements who aren't being invoiced, or invoiced clients with no active project.
• Airtable + HubSpot. Cross-reference your Clients table against your HubSpot accounts. Surface accounts in HubSpot that aren't tracked in ops, or vice versa.
• Airtable + Google Calendar. Match project deliverables in Airtable against meetings on the calendar. Surface deliverables with no scheduled review.

These joins are not formulas you can write inside Airtable. They are the questions you sketch on a whiteboard, fail to answer, and quietly drop.

What this replaces

• The scripting block someone built in 2023 that nobody now understands
• The CSV export-and-pivot you do on the last Friday of every quarter
• The paid Sync to Postgres-to-Looker-to-dashboard pipeline that costs more than your hire would have
• The "we'd need an analyst for that" answer to every interesting operations question
• The four-table rollup field that broke when somebody renamed a status option

Questions you can also ask Anna

The audit above is the wedge. Same toolset, same OAuth:

• "For every content piece in our Calendar base, calculate days from brief to publish. Show me the slowest writers and the fastest, normalised by piece length."
• "In the Talent Database, group candidates by referral source. Show me which sources produce the highest offer-acceptance rate."
• "For every project in the Projects base, flag the ones missing a kickoff date, a deadline, or an assigned lead."
• "Across all bases I've shared, find duplicate company records. Match on domain or normalised name."
• "Pull the Tasks base. Surface every task assigned to someone who's been out for more than two weeks."

Each is one sentence. Each used to be either a developer ticket or a Friday afternoon in the base.

The take-home

Airtable is where your operations live. The numbers that matter — utilisation, bottlenecks, profitable clients, stale pipeline — are sitting in linked records the team hasn't found time to roll up.

Connect the workspace. Ask the question. Anna reads the tables, follows the links, runs the maths, and writes the brief.

The base finally stops being a place where data goes to die.