TL;DR: Informatica, the standard example of usage pricing that worked, reported renewals below forecast in February 2025. Its CEO told investors that when a customer's executive sponsor changed, the account team knew and the renewal team didn't, and that changing the teams' incentives was the fix. Usage pricing makes each renewal a fresh purchase, and org charts and comp plans were built to sell once. AI token meters make it worse: usage swings with the customer's agents, so commission on raw usage rewards noise. Compute the share of variable compensation tied to usage; that is the pricing model you actually run. Make one owner accountable for each defended renewal while writing 2027 comp plans.
Want to listen to this article?
Orion Playbook newsletter subscribers can listen to the AI-generated audio version for free. Already a subscriber? Log in here
"To begin with, renewal rates were lower than forecast." That is how Informatica's CEO began his account of a disappointing quarter in February 2025. The company is the standard example of a move to consumption pricing that worked, and it had just closed 2024 with $827M of cloud subscription ARR. The commercial half of the transition had worked; the miss came after the sale. (The fourth-quarter call is rarely the part of a case study anyone reads.)
The cause he gave sat in the org chart. A customer's executive sponsor changed. The account team knew it. The renewal team didn't, or at least not early enough, and the teams around the account were not paid to work together. The fix was to change their incentives.
In a seat model, a renewal rides on inertia, so a gap between the people who watch an account and the people who renew it costs little. Consumption removes the inertia. The customer re-decides on usage at every renewal, making each one a fresh purchase, and the seat-era org chart was built to sell once. The fact that decides the renewal, a sponsor gone or usage sliding, has to travel from the meter through the seller to whoever signs, and it stops at the first handoff nobody is paid to make.
A common diagnosis puts the failure in Customer Success. The record puts it between the teams that touch the account after the sale, not inside any one of them. Snowflake, which charges by consumption, took the other route. Its CFO said the first thing it did under a new CEO was blow up Customer Success, and every rep carried a big consumption quota. It removed the function; Informatica kept its teams and changed how they were paid. Both were reaching for the same thing: someone paid on what happens once the contract is signed.
The number the board watched was built to miss this. Informatica's headline migration measure, the average ARR uplift when a customer moved from self-managed licenses to cloud, went from 2x to 1.9x, then to a guided 1.5x to 1.7x for 2025. An uplift is a price increase measured on survivors. A customer that cut back partway through the move barely registers in it, and partial churn was on the list of misses from the earnings call. Boards read the average. Renewals happen one account at a time.
Both companies' fixes ran through compensation plans. That makes the payroll a pricing document. Add up target variable pay across sales, success, and renewals, and work out what share points at the new unit. That share is the model the company is actually running. Snowflake published its split in 2022: roughly 70% bookings and 30% consumption in a new territory, moving to about 30/70 in a mature one. A company announcing a consumption strategy while paying almost entirely on bookings is running the old model by its own payroll. Nobody publishes a benchmark for that share. Computing your own takes an afternoon and a spreadsheet.
The AI meters arriving this year make all of this worse. They count tokens, credits, and agent tasks, and as I argued in an earlier issue on changing what you charge for, nobody has settled what the unit should be. Token consumption swings with whatever the customer's agents did that month. A team paid purely on usage is partly paid for weather, and it learns to push consumption the customer may contest at renewal. The other measure is defended usage, the consumption a customer would argue for when the renewal comes up. Carry both, with one owner answerable for the second. And yes, defended usage is harder to measure, with simple comp plans beating correct ones more often than consultants admit.
A CEO asks who is paid when a renewal is defended, and compares that name with the one who owns the forecast. A function leader finds the one incentive that contradicts the stated model and changes it. A revenue-operations lead models next quarter's pricing change account by account and looks for the cohort that comes out worse. Most 2027 comp plans are being written now. Somebody in your company already knows which renewal is in trouble. Check whether anyone is paid to say so.
All the images were generated with AI (ChatGPT Images, Gemini Nano Banana, Claude Opus) by Gérard Métrailler.
Sources
Callahan, Greg. "Inside Software 09.06.26: The Spreadsheet Was the Easy Part." Inside Software, September 7, 2026. https://insidesoftware.substack.com/p/inside-software-090626-the-spreadsheet. Accessed 2026-10-17.
Informatica Inc. "Informatica Reports Fourth Quarter and Full-Year 2024 Financial Results." Form 8-K Exhibit 99.1, February 13, 2025. https://www.sec.gov/Archives/edgar/data/1868778/000186877825000004/infa-2024q4x8k_ex991.htm. Accessed 2026-10-17.
MarketBeat. "Informatica (INFA) Q4 2024 Earnings Results & Call Transcript." February 13, 2025. https://www.marketbeat.com/earnings/reports/2025-2-13-1114625-stock. Accessed 2026-10-17.
Straus, Andrew. "Sales Compensation in a Consumption Pricing World." Snowflake, March 24, 2022. https://www.snowflake.com/en/blog/sales-compensation-in-a-consumption-pricing-world/. Accessed 2026-10-17.
Vaswani, Anoushka. "Navigating Consumption-Based Pricing with Mike Scarpelli, CFO of Snowflake." Lightspeed Venture Partners, May 14, 2021. https://lsvp.com/stories/navigating-consumption-based-pricing-with-mike-scarpelli-cfo-of-snowflake/. Accessed 2026-10-17.
