Exploring what’s top of mind for B2B buyers

For most of the software era, pricing was simple: pick a plan, count the seats, sign the contract. AI is changing that.

AI products cost more to run the more they're used, and their value depends on how much work they actually do. So more vendors are moving away from flat seat pricing toward models that scale with usage or results.

Gong Labs analyzed sales conversations to see how fast that shift is happening, and whether it affects who wins.

"Consumption pricing" covers two related approaches:

• Usage-based pricing charges for how much you use: credits, tokens, agent runs or API calls. You pay more when you use more.
• Outcome-based pricing charges for results: a resolved support ticket, a booked meeting, a processed invoice. You pay when the product delivers.

Many vendors now combine them with a platform fee, creating hybrid models. What they share is the idea that price should track value, not headcount.

For most of the past two years, conversations about pricing models barely changed. From February 2024 through the end of 2025, usage- and outcome-based pricing came up in about 1.5% to 1.8% of sales calls, steady enough to look like background noise.

That changed in 2026. Mentions climbed through the summer, peaked at 2.33% of calls in September and held at 2.27% in October. That’s 27% higher than a year earlier. The share is still small. But after nearly two years of flat readings, a jump this size in a few months suggests buyers and sellers are starting to put the pricing model itself on the table, not just the price.

The obvious next question is what’s driving it. The timing matches the spread of AI products, which are often priced by consumption, credits or results instead of seats, so we took a closer look at AI-specific deals.

Among AI-native companies, the share of new deals where usage- or outcome-based pricing came up rose from 6.8% in Q3 2024 to 12.4% in Q3 2026. That's nearly double in two years.

Among AI new-business deals, when usage- or outcome-based pricing came up in the first five calls, the win rate was 42.7%. The baseline for comparable deals, matched for deal size, was 34.0%. That's 8.7 points higher, or about a 26% relative lift.

This is an association, not proof that early pricing discussions caused the wins. But it suggests that when buyers and sellers talk through how the price will scale, deals tend to go better.

Bring it up early. Pricing that scales can make buyers nervous if it surfaces late. Deals that discussed it in the first five calls won more often, so get ahead of the "What if we use more than expected?" question.

Sell the math, not the rate card. Help buyers build their usage model in the deal. A seller who can say "Here's what your team will likely spend, and here's what it gets you" earns trust fast.

Treat adoption as revenue. With consumption pricing, the deal isn't done at signature. Revenue grows as customers use more, so onboarding and customer success now directly drive expansion.

Rethink forecasting and comp. Consumption revenue is less predictable than seat revenue. Revenue leaders need to forecast usage, not just bookings, and decide how sellers get credit for growth that happens after the deal closes.

Offer ways to make it predictable. Committed spend with volume discounts, pooled credits across teams, and clear overage terms help buyers say yes without fearing a surprise bill.

Know the unit before you sign. Ask exactly what a credit, token or outcome is, and map it to the workflows your team will actually run. "1,000 credits" means nothing until you know what a typical week of work costs.

Model your usage, not just the price. Build a best, expected, and worst case for your first year. If the vendor won't help you do this, treat that as a signal.

Don't let cost fears kill value. The biggest risk with consumption pricing isn't overspending. It's teams holding back to save credits and never getting the value they paid for. Set clear rules on who can use what, then encourage the use cases that pay off.

Measure value per unit. Track what each unit of spend delivers, such as time saved, deals moved or tickets resolved. That turns renewal from a negotiation into a math problem.

Consumption pricing is moving from experiment to expectation in AI deals. Sellers who make pricing a transparent, early conversation will win more often. Buyers who understand the unit and model their usage will get more value. And for both, the real measure isn't the price per credit. It's what each credit delivers.