Field NotesNext step
How do you audit the AI tools you already pay for?
Most small businesses don’t buy too many AI tools. They stop checking the ones they have. A simple audit to decide what stays, what gets fixed, and what goes.
Short answer
An AI tools audit is a short review of every AI tool and AI feature your business pays for. You list them from your billing statements, score each one on adoption, output quality, measurable impact, and overlap, then put each into one of three columns: keep, fix, or cut. It takes an afternoon and a spreadsheet.
How do you run an AI tools audit?
List every tool, score each one on four criteria, separate dormant tools from broken ones, and decide keep, fix, or cut. Then write down why you kept what you kept.
- inventory
List everything you pay for, including AI features inside tools you already use.
- score
Rate each tool on adoption, output quality, measurable impact, and overlap.
- diagnose
Separate dormant from broken for every low scorer.
- decide
Keep, fix, or cut. No maybes.
- document
Write a note for each tool you keep: its job, its owner, and the number you track.
How do you find every AI tool your business is paying for?
Pull the last 90 days of card and bank statements and list every software charge, not just tools with “AI” in the name. Then mark which ones include AI features, even ones nobody has turned on.
AI now lives inside tools you don’t think of as AI tools. Check each of these:
- Email platforms: writing assistants, subject line and send-time features
- Your CRM: lead scoring, call summaries, email drafting, forecasting
- Social schedulers and design tools: captions, image editing, layout
- Customer service software: chat, ticket routing, suggested replies
- Project management tools: task prioritization and workload features
- Standalone AI assistants anyone on the team pays for directly
For each, write the monthly cost and who uses it. Don’t filter yet.
How do you score whether an AI tool is worth keeping?
Rate each tool from 1 to 3 on adoption, output quality, measurable impact, and overlap. Low totals are cut or fix candidates. High totals are earning their place. Everything in between needs a decision.
| Criterion | 1 | 3 |
|---|---|---|
| Adoption | Rarely used | Part of a documented workflow the team runs |
| Output quality | Needs heavy rework | Passes your standard with light review |
| Measurable impact | No number at all | A specific, tracked result |
| Overlap | Duplicates another paid tool | The only tool doing this job |
What’s the difference between a dormant AI tool and a broken one?
A dormant tool works but nobody built a workflow around it. A broken tool was used properly and still produced poor output. A dormant tool needs an owner and a workflow. A broken tool needs to be cut.
Ask one question about every low scorer: was there a documented workflow with a named person running it? If not, the tool is probably dormant. If yes, and the output was still poor, it’s broken. Mixing these up means cutting something you never activated, or building workflows around something that won’t work.
What do you do with each tool after the audit?
Keep tools that score well and have an active workflow. Give dormant tools one 30-day fix with an owner and a weekly check-in. Cut broken and duplicate tools this week, not next quarter.
- Keep: protect it by keeping its workflow written down and owned.
- Fix: one owner, one documented workflow, one weekly check-in for 30 days. Then it moves to keep or cut. No second round.
- Cut: cancel it. What you already spent isn’t a reason to keep paying.
How do you consolidate overlapping AI tools?
Group tools by the job they do, then keep the one your team actually uses. Before buying a specialist tool, check whether your CRM or main platform already includes that feature at your current plan.
Overlap usually comes from buying reactively: a demo, a recommendation, a newsletter. One well-configured tool with a clear owner beats three tools with partial adoption. Add a specialist tool only when your main platform hits a real, documented limit. Most of the work is configuration, which is the part nobody does.
What does an AI audit tell you about your business?
The pattern matters more than the list. Mostly dormant tools mean you need a rollout process. Mostly broken tools mean you buy on demos. Good tools with no numbers mean you need measurement.
For new tools, set a rule: a named problem, an owner, a 30-day success measure, and an exit trigger before anyone signs up. For measurement, see How do you measure whether an AI workflow is working?
How often should you audit your AI tools?
Every quarter, and whenever business slows down. A slower stretch is the best time to audit, document workflows, and fix prompts, because you have time and months of real usage to judge from.
Use that time to improve what you have, not to shop for new tools. Pick one audit, one workflow to document, and one fix, and finish them before the calendar fills again.
More questions
How long does an AI tools audit take?
For most small businesses, an afternoon. The slow part is pulling statements. Scoring and deciding go quickly once everything is on one list.
Should I cut a tool my team likes but I can’t measure?
Put it in the fix column. Give it an owner and one number to track for 30 days. If it can’t show a result by then, cut it.
What if I paid for an annual plan?
Decide based on what keeping it costs from here and what it produces, not what you already spent. Stop renewal now, even if access runs out later.
Do I need special software to audit AI tools?
No. A spreadsheet with columns for tool, cost, user, score, and decision is enough.
Is the goal to spend less on AI?
The goal is a set of tools where each one has a clear job, an owner, and a result. Spending often drops, but that’s a side effect.