Quarterly AI Tool Audit: A Simple Template to Cancel or Keep Every Subscription

If your business is like most, your AI tool stack grew without a plan. You signed up for ChatGPT because everyone was talking about it. Someone on the team started using Notion AI. Your marketing person added Jasper. The developer added a GitHub Copilot seat. You got a trial for some tool at a conference and never cancelled it.

Now you’re paying for five or six AI subscriptions, some of which overlap significantly, some of which nobody is using, and none of which have been evaluated against each other or against the outcomes they were supposed to deliver. That’s not a small business problem — it’s the default state of most organisations that adopted AI tools during the past two years.

A quarterly AI tool audit fixes this. Here’s a practical template you can run in under two hours, every three months, to make sure every subscription you’re paying for is earning its place.

Why Quarterly — Not Annual

The AI tool market moves faster than almost any other software category. Tools that were best-in-class six months ago may have been overtaken. Free alternatives that didn’t exist last year might now cover everything a paid tool was doing. A model tier you needed for quality reasons may now be available at half the price from a competitor.

An annual audit is too slow. You’ll overpay for a full year on tools you should have cancelled in month four. Quarterly keeps your stack honest without becoming a constant distraction. Three months is also long enough to generate real usage data — you can tell the difference between a tool people genuinely rely on and one they opened twice after the initial excitement wore off.

Step 1: Build Your AI Tool Inventory (20 minutes)

Before you can evaluate anything, you need a complete list of every AI tool your business is paying for. This sounds obvious, but most businesses don’t have it written down anywhere. Pull together:

• Every subscription visible in your credit card or bank statement from the past three months — filter for any company you associate with AI
• Ask each team member what AI tools they’re using, including any personal accounts they’re using for work tasks
• Check your Google Workspace or Microsoft 365 admin for any AI add-ons or connected apps
• Review any annual or per-seat contracts signed in the past year

For each tool, record: the tool name, what it was supposed to do, who uses it, cost per month, and contract type (monthly vs annual). This list is your audit starting point.

Step 2: Assess Actual Usage (30 minutes)

This is where most audits uncover the most savings. For each tool on your list, find out if it’s actually being used — not “do we have access to it,” but “are people opening it and getting value from it.”

Most SaaS tools have usage analytics in their admin dashboards. Log into each tool’s admin panel and check: number of active users in the past 30 days, sessions or generations or queries run, and any usage trend (growing, flat, declining). For tools without admin dashboards, a quick survey of your team takes five minutes.

Be honest about what “used” means. A tool someone opens once a week to do one task counts as used. A tool that was heavily promoted at onboarding and hasn’t been touched since the first week does not. Tools with declining usage trends deserve particular scrutiny — they’re usually in the process of being quietly abandoned by the team.

Step 3: Identify Overlap (20 minutes)

Overlap is the biggest source of waste in most AI stacks. When you have ChatGPT Team, Notion AI, and Jasper all paying for access to capable language models, you’re almost certainly paying three times for capability you could get from one tool.

Go through your inventory and for each tool, ask: what is the specific thing this tool does that couldn’t be done by another tool already in the stack? If the answer is “nothing specific,” that’s a cancellation candidate.

Common overlaps to look for:

• Multiple general-purpose AI writing tools (ChatGPT + Jasper + Notion AI)
• Multiple meeting transcription tools (Otter + Fireflies)
• An image generation tool you’re barely using alongside a ChatGPT Plus subscription that includes DALL·E
• A research tool that duplicates what ChatGPT or Claude with web search can already do

Step 4: Calculate ROI for Each Tool (20 minutes)

For each tool you’re keeping, do a rough ROI calculation. This doesn’t need to be precise — it needs to be honest. The question is: is the value this tool creates clearly worth what we’re paying for it?

For writing tools, estimate hours saved per week by the people using it, multiply by an hourly rate, and compare to the subscription cost. For a tool saving two hours per week at $50/hr, that’s $400/month in value from a $30/month tool — obviously worth keeping. For a tool saving 20 minutes per week, the math is tighter and worth scrutinising.

For tools where the value is less directly measurable — research tools, analysis tools, customer-facing chatbots — use a different frame: could you get equivalent value from a tool you’re already paying for? If yes, the specialist tool is probably redundant.

Step 5: Make and Execute Decisions (10 minutes)

The audit is useless without action. For each tool, one of four decisions applies:

Keep: actively used, clear value, no significant overlap with other tools in the stack.

Cancel: low usage, clear overlap with another tool, or ROI doesn’t justify cost.

Downgrade: used and valuable but on a plan tier with features nobody uses. Most AI tools have plan tiers — if you’re on a Team plan but only one person uses it, check if a lower tier covers actual usage.

Evaluate: tool is used and might be valuable, but there’s a newer or cheaper alternative worth testing before the next renewal. Set a reminder to run a two-week trial of the alternative before the next billing cycle.

Execute cancellations and downgrades before the next billing date. Document your decisions and your reasoning — so that in three months when you run the next audit, you have context for why things are set up the way they are.

What a Healthy AI Stack Looks Like

Most small businesses with 5–20 people need far fewer AI tools than they think. A well-curated stack typically looks something like: one general-purpose AI chat tool (ChatGPT Team or Claude Team), one meeting transcription tool, one AI coding tool if you have developers, and perhaps one specialist tool for a specific high-value workflow. That’s four tools at most, and often three is enough.

The goal isn’t to use the most AI tools. It’s to get the most value from AI with the least overhead, cost, and complexity. A quarterly audit keeps your stack pointed in that direction.

Making the Audit a Team Habit, Not a Solo Exercise

The most durable version of a quarterly AI audit is one that involves the people who actually use the tools, not just the person paying the bills. A 30-minute team check-in where each person shares which tools they’re using, what’s working, and what isn’t produces better decisions than a solo review of subscription receipts — because the people closest to the work have the clearest picture of value.

What to Do When an Audit Reveals Underperforming Tools

A quarterly AI audit that finds tools generating no measurable value is valuable only if it results in action. Cancellation is the right response when a tool has been available for a full quarter with no meaningful usage and no compelling case for future adoption. Suspension (pausing the subscription rather than cancelling) is appropriate when usage is low but a specific upcoming project will need the tool. Consolidation is appropriate when two tools overlap significantly — audit results frequently reveal redundant subscriptions acquired at different times by different team members. The hardest audit outcome is a tool with high cost and low measured value but strong advocacy from the team that uses it. For these cases, ask the advocates to quantify the value specifically — if they cannot describe the business outcome the tool enables in concrete terms, the tool is probably not worth its cost.