Your Procurement Queue Has 47 SaaS Renewals This Quarter and Finance Approved Six Blind

It is Monday, September 21, 8:14 AM. Your CFO opens the SaaS renewal calendar. 47 contracts up between October 1 and December 31. Six have a vendor quote sitting in her inbox with a 14 percent uplift and a countersign deadline this Friday. The other 41 are set to auto-renew on the anniversary date under the terms the previous ops lead signed in 2024. Total Q4 renewal spend on the sheet reads $1.87M. The line for real usage data next to each contract is blank on 39 of the 47 rows.

You bought Vendr at $30K a year to help negotiate the top ten. You bought Zylo at $48K to inventory the tail. Your Head of Ops runs procurement on Fridays between the exec digest and payroll. Six renewals get a real review each quarter. Forty-one get a signature on the anniversary and a Slack message to finance that the seat count stayed flat. The SaaS line on the P&L grew 34 percent year over year. Nobody could name the three vendors driving half the growth.

What procurement is at Series B

Procurement is a running function with five parts. A live inventory of every contract, its renewal date, and its owner. A usage read pulled from the tool itself, not the vendor's dashboard. A benchmark on what the market pays for the same seat count. A negotiation cadence that opens 90 days before renewal, not 14. A signoff loop that closes the file with a decision, a discount captured, and a seat count reset for the next term.

Not one of those five parts runs off Vendr's quarterly review call. Not one runs off Zylo's inventory export. The tools inventory contracts and negotiate top-of-list renewals. The function runs when a human reads the inventory, pulls the usage, sets the benchmark, opens the vendor thread 90 days out, and closes the file. Your Head of Ops runs the function on six contracts a quarter. The other 41 renew on autopilot with the vendor's number on the invoice.

What auto-renewal costs the P&L

Take a typical Series B SaaS stack. 187 tools in production, 47 renewals a quarter, top ten summing to $1.2M. The middle 20 sum to $520K and the tail 17 sum to $150K. Vendr covers the top ten under the retainer and lands 8 to 14 percent off list on six of them a quarter. That is $90K to $170K in captured savings on $1.2M of top-of-list spend.

The middle 20 auto-renew at list. Three of them raised list prices 12 to 22 percent in the last cycle. Two more silently added a data egress line item that adds $28K a year on the current usage curve. One added a per-seat AI feature at $18 a seat a month that got turned on by default for all 340 seats. The middle bucket ate $180K of unbudgeted spend in the last four quarters. Nobody caught it until the CFO ran the SaaS variance report against last year's budget in month two of Q3.

The tail 17 auto-renew at list on annual agreements nobody remembers signing. Four are duplicates of tools already in the stack under a different name. Two are on 2023 pricing tiers that are 40 percent above the current published price. One is on an enterprise contract that got signed for a workflow the team stopped running in Q1. The tail carries $70K of pure waste per year. It renews clean every anniversary because nobody has time to open the drawer.

Why the tools you bought do not close the loop

Vendr negotiates the contracts you route to it. It cannot read your Okta login data, your seat activity, your usage telemetry, or the Slack thread where the tool's owner said last month they were sunsetting it. It negotiates against what you send. It does not decide which contracts deserve negotiation, and it does not chase the usage read that would justify a 40 percent seat cut on the renewal.

Zylo inventories the stack. It surfaces the 47 renewals, tags the owners, and flags the tail. It does not open the negotiation. It does not read the product usage inside each tool. It does not draft the counterproposal against the vendor's uplift. Your Head of Ops still owns the read, the negotiation, the signoff, and the follow-up on every contract Vendr does not touch.

The stack grew faster than the human running procurement on Fridays. The inventory got clean and the negotiation loop stayed thin. The middle 20 renewals kept clearing at list. The tail kept renewing on 2023 terms while the SaaS line ran ahead of headcount.

What a fractional AI procurement function does

An agent stack runs the five parts on a cadence you did not have to prompt. On the 1st of every month, the inventory agent reads Zylo, reads the Okta app catalog, reads the AWS marketplace subscriptions, and reads the three shadow tools charged to the CTO's Amex last month. It reconciles the four sources into one contract sheet with owner, renewal date, current spend, and last usage read. The sheet lands in the CFO's inbox at 8:30 AM.

On the 91st day before each renewal, the usage agent opens the tool's admin console, pulls the last 90 days of seat activity, and writes the read into the contract row. Seats logged in fewer than four times in 90 days get tagged for cut. Features enabled by default get tagged for downgrade. The agent drafts the counterproposal against the vendor's last quote plus the benchmark from three anonymized peer contracts in the same tier. The draft lands in the procurement channel with the owner tagged.

On the 30th day before renewal, the negotiation agent opens the vendor thread with the counterproposal, tracks the reply cadence, and escalates to the ops operator when the vendor stalls past five business days. The signoff loop closes the file with the discount captured, the seat count reset, and the next renewal date logged. The 47 renewals stop auto-renewing at list. The middle 20 clear against a benchmark. The tail gets cut, downgraded, or negotiated to current market.

The unit economics of the trade

Path A keeps Vendr at $30K, keeps Zylo at $48K, keeps the Head of Ops running procurement on Fridays, and clears six top-of-list negotiations a quarter. Captured savings land $90K to $170K a year against $1.87M of quarterly renewal spend. Middle and tail keep auto-renewing. The SaaS line grows 20 to 35 percent a year on the same seat curve.

Path B keeps Vendr on the top ten where the retainer earns its keep and keeps Zylo as the inventory source of truth. It runs a 14-day sprint to stand up the agent stack across inventory, usage, and negotiation on the middle 20 and the tail 17. It runs at $4K to $8K a month on API spend and tooling plus a fractional ops operator at $4K to $6K a month who owns the escalations. Year one lands $240K to $420K in captured savings and cuts against the same renewal calendar, and the SaaS line growth curve flattens against seat growth, not vendor uplift.

The three questions to run against every renewal on the Q4 calendar

Read the renewal sheet before the next signoff cycle. Every contract deserves three questions before the countersign lands. The questions score whether the seat count is buying a running function or a line item on the P&L. Run them once against the 47 rows this week and a third of the queue reprices itself.

Which usage read backs this renewal? Name the last 90 days of active seats, the feature adoption rate, and the owner who confirms the tool still runs a workflow. If the answer is a vendor dashboard export, the read is marketing telemetry, not procurement data. Pull the number from the admin console before the countersign.

Which benchmark backs this price? Name the peer contract, the seat tier, and the discount captured on the last two negotiations. If the answer is the vendor's list price, the negotiation never opened. The renewal is a signature on a number the vendor picked.

Which owner runs the function this seat supports? Name the person, the cadence, and the output the seat produces. If nobody owns the workflow or the workflow stopped running in Q1, the seat is a subscription to a memory. Cut it before the anniversary.

If your CFO is signing 41 renewals blind this quarter, the version where the middle 20 clear against a benchmark and the tail gets cut is fourteen days away. Vendr negotiates. Zylo inventories. An agent stack runs the function. Scope a sprint and read the first captured savings line in the Q4 close.