Walk a mid-market close, a warehouse, or a support queue and you are walking a software path. Invoice to cash lives in one product. Exceptions live in a spreadsheet someone swore they would retire. Buying those tools as if they were toner is how you get seven CRMs and one actual process.
Vertice’s Q2 2026 read of licenses it tracks put 65% of SaaS licenses as unused or underutilized (under 50% of purchased seats actually used), up from 62% a year earlier. Fully unused “shelfware” ticked down slightly; underuse did the damage. That is not a reason to fire every tool. It is a reason to stop treating a renewal as a rubber stamp.
RFP basics that survive a demo
Demos are designed to make the happy path look inevitable. Score the unhappy path.
- What work dies if this tool is down for a day?
- Show month-end, a chargeback, a warehouse that is offline, a customer who is not in the dropdown — without a services SOW.
- Who is allowed to add a seat, and how is that logged?
- What is the exit: export, escrow, or a second vendor already on the bench? If export needs a professional-services ticket, price that ticket now.
- Which overlapping tool does this retire, and on what date?
If the product cannot show the unhappy path, you are not buying software. You are buying a project. Write the RFP as a process review, not a feature bingo card. A pretty matrix that never names an owner is how overlapping CRMs survive.
A seat that nobody can turn off is not a subscription. It is a tax.
Seat waste
Three piles, not one:
- Ghosts — paid seats tied to people who left, or who have not signed in for 90 days.
- Overlap — two products doing 30% of the same job. At renewal, one of them is on trial whether the vendor knows it or not.
- Underuse — the tool is opened, but most of the purchased capacity is idle. Vertice’s 50% threshold is a blunt instrument. It is still more honest than “we still need it.”
True-down unused seats before the quote locks. “Standard uplift” is often a sales sentence. If it is not in the paper, it is a request. An 8% increase on a tool that replaced two others can be cheap. A 0% increase on a tool nobody opens is still a loss.
Who signs
Three chairs, one meeting: the person who signs, the person who uses it, and the person who would run the export. If one chair is empty, you will renew by default.
IT should not be the only signature on a tool that runs invoice-to-cash. Finance should not be the only signature on a tool that can take the warehouse down. A founder who can add seats from a credit card is how shadow IT becomes the operating system.
Put a rule on new logos: no entry without a retirement plan for an old one, and a named workflow owner. Hire for that, or keep paying for catalogs.
SaaS management categories
The market split in 2026 is not “one platform to own software.” It is three jobs. Mix them up and you will buy a dashboard when you needed a negotiation, or a negotiation when you needed a kill list.
| Category | Job | Usually owned by |
|---|---|---|
| Discovery / SaaS management | Find the stack, usage, and ghosts (Zylo, Torii, Productiv, and the spreadsheet-plus-IdP version below ~1,000 people) | IT / ops |
| Procurement / negotiation | Benchmarks, true-down, the 8% “standard” increase (Vendr, Tropic, Vertice, or a broker who actually reads the order form) | Finance / procurement |
| Operating ownership | Named workflow, kill date, unhappy-path RFP. No vendor replaces this | The operator who runs the process |
Those names are examples of the category, not a ranking and not a recommendation. Below a thousand employees, an export from identity, a pull from accounts payable, and a 90-day last-login report will get you through the first cycle. Above that, you will miss ghosts without a discovery pass. In either case, the operating skill is the owner and the kill date. The software does not buy itself well.
Sources
FAQ
Do we need an RFP for every renewal?
You need the unhappy-path questions and a stay / shrink / leave meeting. A 40-page RFP for a $6,000 tool is theater. A Slack poll after the invoice is how overlap survives.
Is unused-seat waste a reason to rip everything out?
No. Vertice’s 65% figure is unused or underused licenses in its dataset, not a law that your stack is 65% wrong. True-down ghosts. Put overlap on trial. Keep the tool that retired hours or another license.
Who is allowed to add a seat?
Someone named, logged, and tied to a workflow. A founder credit card is not a process.
Which SaaS-management product should we buy?
This page does not pick one. Match the category to the job. Discovery is not negotiation. Neither is a substitute for an owner.