14 Trends in AI Monetization

Welcome back to Good Better Best.

Each week, we break down real pricing, packaging, and product moves from SaaS and AI leaders and share the ideas worth stealing.

This week, we're partnering with our friends at Teneo to share 14 trends in AI Monetization. Teneo partners with software companies on the hardest version of the AI pricing problem: moving an established seat-based business onto a model built for AI. This has given them a window into what’s working, what’s not, and what we should look out for going forward.

Before that, we have two upcoming events:

Let’s get to it.

This week we observed 220+ changes. The highlights:

Check out more updates on PricingSaaS →

Teneo sorts the 14 trends into three buckets:

Here's the full list, plus my take on each. The slide below is from Teneo’s latest AI Monetization report. Grab the full report here.

AI lets one person do the work of several, which breaks the link between headcount and value. In a lot of categories, seats are now inversely correlated with value. The better your agent works, the fewer seats your customer needs.

66% of vendors now combine a fixed fee with a usage charge, up from 45% in 2023. The go-to model for many SaaS companies is Seats + Credits. The takeaway: Licenses still make sense in the AI era as buyers push for predictability, but as agentic products take off, another meter is needed to capture the upside.

AI lines are growing 100–200% at some companies, and leaders are already getting 9–12% of revenue from AI (NiCE, Similarweb, Five9). But the median is about 4%. Teneo expects that to hit 10–30% of many firms' economics in the next 12–24 months. If that's right, the pricing decisions you make this year will show up in your 2027 P&L.

We tracked 959 pricing changes across 500 companies in H1 2026, and 42% of companies launched new AI pricing within 12 months. We see 200+ changes a week on PricingSaaS now, and that pace isn't slowing down. The advantage goes to whoever can keep adapting, not whoever finds the "perfect" model.

This is the most important stat in the deck. When customers turn down AI features, 47% of vendors say it's because pricing is unpredictable. Only 39% say it's because the price is too high. We've watched this play out in real time: Cursor, GitHub Copilot, Replit, and Zendesk all took public heat over surprise bills or confusing credit changes.

AI stitches capabilities together across a product, so the natural unit to sell is the job, not the module. The vendor runs the whole workflow, the customer buys the finished result, and the meter moves closer to the work (runs, tasks, completed jobs). Expect fewer standalone add-ons and more shared wallets across products.

Credits separate the price a customer sees from the meter underneath. Vendors peg them to one of three things: their own costs (protects margin), product activity (simple, but the exchange rates get messy), or completed work (best value alignment, but you have to define "done"). We're still watching new credit systems go live every week. Cost-based credits are easy to launch, but work-based credits will age better.

Once you sell credits, the balance, burn rate, alerts, allocations, top-ups, and usage history become product features. Customers need them to stay in control, and finance needs them to forecast. This is why metering, entitlements, and billing infrastructure is having a moment. You can't run a credit model on a billing system built for annual seat contracts.

Pricing used to be a finance and sales exercise. Now it's getting designed into the product from day one, including metering, entitlements, and in-product controls. That requires real collaboration between product, finance, and GTM. If your pricing still lives in a spreadsheet owned by one person, that's the first thing to fix.

Customers want flexibility, but they need predictability more. In Teneo's buyer study, the top condition for accepting usage pricing was structured commitment, not open-ended pay-as-you-go. The fix looks the same everywhere: T-shirt-size commitments (Small / Medium / Large), usage included in the plan, and a brake pedal for finance. See Salesforce Pre-Commit, Workday Flex Credits, and SAP AI Units.

As agents reach into software through MCP instead of the UI, vendors have to decide what to meter and how agent activity maps back to entitlements, credits, or paid usage. Our data backs up Teneo here. Most vendors include MCP in existing plans. Business apps lean toward paid tiers, and standalone MCP add-ons are rare.

Evals (structured quality tests) can now score whether AI work was done well. That makes it possible to charge per resolved ticket or completed outcome, with no charge when the work misses the bar. Helply just made this move, swapping $1 per ticket for $0.50 per outcome.

I'm still skeptical this goes mainstream soon. The vendor takes on more risk, the quality test has to be agreed with the customer, and trend #10 says buyers want forecastable spend first. Evals make outcome pricing possible. Buyers will set the timeline.

With BYOK, customers pay the model provider directly and the vendor charges for the software on top. Usage can grow 10x without the vendor's costs growing 10x.

This is the harness premium in its purest form. When the model cost is the customer's problem, your orchestration, data, and workflow have to justify the price on their own.

Winning in this model takes new motions, new messaging, and teams equipped to land and expand on usage, not just seats. Customer success becomes critical: helping customers realize value, manage consumption, and avoid bill shock.

In a seat world, CS protected renewals. In a usage world, CS drives revenue.

A big thank you to the Teneo team for sharing their research with us. The full report goes much deeper on each of these, including the survey data from 300 vendors and 400 enterprise buyers.

Thanks for reading! If you’re working on AI monetization and want to learn more about how we help, book time here.

Until next time,

Rob