The first thing I ask when I walk into a new engagement is for a list of every SaaS subscription the company pays for. In the last two years I’ve done this maybe thirty times. Not once has the list the CFO sent me been complete. The actual count, after a proper audit, is usually two to three times what they thought.

A 200-person company in Melbourne I worked with last quarter had 187 active SaaS subscriptions on the books. Their CFO had told me 60. The total spend was $2.1 million a year. By the time I’d finished the audit, we’d identified $640,000 of clear annual waste. None of this is hard. It just requires actually doing it.

This is the process I run.

Week one: get the real list

The CFO has expense reports. IT has SSO logs. Security has whatever they’re using for shadow IT detection. Procurement has vendor contracts. Marketing has its own credit card subscriptions that nobody else knows about. Engineering has tools that came in via developer expense reimbursement.

You need all of it. The way I do it: pull twelve months of expense data filtered for anything looking like a software subscription, cross-reference against SSO provider logs (Okta, Azure AD, JumpCloud — whichever is in play), and then walk department by department asking heads what they’re actually using. The walking matters. People remember tools when prompted that they’d forgotten existed.

Then I categorize. The matrix I use has four buckets:

• Core systems: ERP, CRM, ITSM, the obvious stuff. Usually managed properly.
• Productivity: Microsoft 365, Slack, Notion, Atlassian. Usually overprovisioned.
• Specialist tools: design tools, analytics, marketing automation, dev tools. Where most waste lives.
• Phantom subscriptions: tools nobody can identify an owner for. Often 15-20% of the spend.

Week two: kill the phantoms

The fastest wins come from the phantom bucket. These are subscriptions where I email the listed billing contact and either get no response, get “I haven’t used that in a year,” or discover the original owner left the company eighteen months ago.

In the Melbourne client’s case, this bucket alone was $180,000 a year. Tools that had been auto-renewing since 2023, paid for via a corporate card that nobody had cancelled. The fix is purely administrative: cancel them, watch for two weeks to see if anything breaks, and move on.

You will get one or two angry emails from someone who forgot they were using something. That’s fine. Reactivating a SaaS subscription takes ten minutes.

Week three: rationalize the overlaps

This is where the real money is and where the politics start. Most companies of any size are paying for two or three tools that do roughly the same thing.

Common overlaps I see:

• Three project management tools running concurrently because different teams adopted Asana, Monday, and ClickUp at different times.
• Two BI platforms because the data team migrated to one but never killed the other.
• Multiple meeting recording / transcription tools — usually Zoom built-in, plus Otter, plus Fireflies, plus whatever Microsoft is bundling that quarter.
• Several AI coding assistants because individual developers signed up for whatever they liked and finance never noticed.

The right move isn’t always to pick one. Sometimes the cheaper move is to negotiate down the duplicates and let teams keep what they like. But you can’t negotiate from a position of strength until you know what you’re paying.

Week four: license rightsizing

Every SaaS vendor sells you more seats than you need. The standard pattern: a company buys 250 seats of something at the start of a year, headcount grows to 280, they buy another 50 seats, then headcount drops back to 240 after a reorg. They’re now paying for 300 seats and using 240.

This audit is mechanical. Pull active user data from each tool, compare to license counts, identify the gap. For most enterprise SaaS contracts you can’t reduce mid-term, but you can plan the renewal negotiation properly. Going into a Salesforce or Microsoft EA renewal armed with twelve months of utilization data is a different conversation than going in armed with vibes.

For Microsoft EA specifically, the published guidance from Microsoft on optimization is a starting point but barely scratches the surface. The negotiation is where the savings live, and you need real usage data to negotiate well.

What this is actually about

The audit isn’t really about saving $640,000, though that’s nice. It’s about establishing that someone is paying attention. The reason SaaS sprawl gets out of hand isn’t malice or stupidity. It’s that nobody was tracking it, so individual purchasing decisions accumulated without governance.

Once the audit’s done, you need a standing process. Quarterly reviews of new subscriptions, annual rationalisation, mandatory SSO for any new tool, expense policy that flags software spend for IT review before it’s reimbursed. Without that, the sprawl rebuilds itself within eighteen months. I’ve seen it happen.

The companies that handle this well treat their SaaS estate the way they treat any other significant operational expense — with discipline, periodic review, and a single accountable owner. The ones that don’t keep funding subscriptions that nobody uses, until someone like me walks in and asks for the list.