Every October I pull up a spreadsheet that I keep at arm's length the rest of the year. It lists every subscription my team or my clients are running: the tool name, the monthly cost, the seat count, the renewal date, and one column I added two years ago that I now think is the most important field on the sheet: who actually opened this last month?
The first time I did this seriously, in my second year managing tools for a top US-based company I work with, I found a project management platform that had been on autopilot for fourteen months. The project it was set up for had wrapped up before I joined. Every month, a quiet card charge, every month, nobody logged in. The fix took about eight minutes. The refund negotiation took longer.
That is the thing about SaaS sprawl. It does not announce itself. It accumulates in small increments, each one defensible at the time, and then compounds quietly until someone finally looks at the credit card statement and asks the uncomfortable question: does this tool earn its line item?
Why October, Not January
Most people think of budget reviews as a January exercise, after the calendar year closes. I do mine in October for a specific reason: most annual SaaS contracts renew in Q4. By the time January arrives, the signatures are already on the new year's contracts. October gives you four to six weeks of lead time to negotiate, cancel, or right-size before renewals lock in.
The window matters. Most SaaS vendors require 30 to 60 days' notice to cancel or downgrade. Miss it by a week and you are paying for another full year. Calendar the renewal dates. Set the reminder six weeks out, not two.
Step One: Build the Real Inventory
Most orgs have an incomplete picture of what they're running. There are the tools IT manages, the tools the finance team approved, and then the shadow subscriptions: the thing marketing signed up for directly, the three-seat plan someone put on a personal card and expensed, the free-tier tool that quietly added paid seats when usage ticked up.
To get the full picture, I pull from three sources simultaneously:
- The corporate card statement. Every recurring charge, three months back. Not just the big ones. Filter for anything with a recognisable SaaS name and anything that recurs on a monthly or annual cadence.
- The email inbox. Search for "receipt", "invoice", "subscription", and "renewal" in the accounts payable or admin inbox. The free-tier-to-paid upgrades often only send email, no card charge until the bill comes.
- Microsoft 365 or Google Workspace app integrations. Both admin consoles show you what third-party apps have been granted access to your tenant. This catches a class of tools you would never find on the card statement because they authenticated with SSO and are paid separately.
The goal is one canonical list. Every tool, every cost, every owner. If a tool has no listed owner, that is your first signal. Tools without owners are tools nobody is accountable for, and tools nobody is accountable for are tools nobody will fight to keep.
Step Two: Check Who Is Actually Using It
Seat count and usage are not the same thing. I have seen accounts where a team was paying for 20 seats of a collaboration tool and eight people had logged in during the preceding 90 days. The other 12 seats were either for people who had left, people who tried it once and went back to what they knew, or seats provisioned speculatively "in case someone needs it."
For each tool on the inventory, I look at two things:
Login frequency. Most SaaS admin panels will show you last-login dates per seat. If more than a third of your seats show no login in 60 days, right-size the plan to what you actually need. Do this before the renewal conversation, not after.
Core-feature adoption. Some tools have high login counts but thin usage. People open the dashboard, check one number, and leave. That is not the same as a team genuinely dependent on the tool. Ask yourself: if this disappeared tomorrow, what would actually break? If the honest answer is "probably nothing, we'd adjust", that tool belongs on the cut list.
A tool your team logs into is not the same as a tool your team cannot work without. The second kind earns its keep. The first one might just be habit.
Step Three: Map the Overlaps
This is where the audit gets interesting. Most mid-sized teams have at least two or three cases of tools doing the same job. Not because anyone made a bad decision, but because tools accumulate through different routes: a salesperson brought in one CRM, operations brought in another for internal tracking, and now you're paying for both.
The overlaps I find most often:
- Two project or task management tools, usually one the engineers prefer and one someone else insisted on for visibility.
- Multiple communication channels with overlapping scope, Slack for most things, Teams because M365, plus a tool someone added for async video that now mostly just has unanswered messages.
- More than one outreach or CRM platform, where different teams signed up independently and the data is now split between two systems neither trusts fully.
- Duplicate automation tools, both Zapier and another platform, each running different workflows, with no one person who owns the full picture.
For every overlap, the question is not which tool is better in the abstract. It is which one your team has actually built workflows around. Consolidation is almost always worth the short adjustment cost. Parallel tools cost money in two ways: the subscriptions, and the slow data fragmentation that makes both tools less useful over time.
Step Four: The Conversation Before the Invoice
Once you have the inventory, the usage data, and the overlap map, you have real leverage. Most people go into renewal conversations empty-handed. You now have specifics.
For tools you want to cut: Cancel before the notice period, full stop. No extended trial, no "we'll reassess in six months." If it did not earn its keep this year, the likelihood it does next year is low. The inertia of auto-renewal is a vendor feature, not a neutral default.
For tools you want to right-size: Pull your actual seat utilisation data before you call. "We're currently on 20 seats and only 11 people are actively using it" is a much stronger opening than "we'd like to reduce our plan." Vendors prefer a renegotiated deal to a cancellation. Most will agree to a lower seat count or a better per-seat rate if you give them the alternative clearly.
For tools you want to keep: Ask for a better rate anyway. Annual commitment for a monthly price, a longer term for a discount, a competitor quote as a reference. Vendors have more flexibility than their published rates suggest. The worst answer is no.
What I Do Not Cut
There is a category of tool that looks like a waste on paper but is not: the one that is cheap, has low visible usage, and quietly prevents a category of problem from ever becoming a ticket.
In my setup, a good example is monitoring. The tools that alert me to outages before users do. The usage logs look thin because they only generate activity when something breaks. Cut them and you find out they were earning their cost the first time something breaks on a Friday afternoon and nobody knows about it for two hours.
The test I apply: does this tool prevent a problem more expensive than its annual cost? If yes, it stays regardless of what the login frequency looks like. The invisible-because-it-works tools are the ones you want to protect, not prune.
What Last October's Audit Found
Across the accounts and internal stack I audited last year, I found roughly four tools that were costing money and had near-zero usage. Two were candidates for consolidation where we were paying for overlap. Three were on plans that we right-sized down by 40 to 60 percent based on actual seat utilisation.
The total saving across that set of subscriptions was roughly a third of what we had been paying. The bigger value was not the money. It was the clarity. After a stack audit, you know what you are running, who owns it, and why. When a new tool gets proposed, you can compare it against what already exists. When someone leaves, you know exactly what to deprovision.
One afternoon a year. Start with the credit card statement, end with a canonical list, and ask each tool the same question: does this earn its line item? If you cannot answer yes, you have your answer.
If you are walking into a stack audit for the first time and want to compare notes on what you are finding, drop me a line. I am happy to look at what you are working with.
Comments