2025 Was the Year Every Pricing Page Got Rebuilt
The scale of this shift is not subtle. According to the PricingSaaS 500 Index, the top 500 B2B and AI companies made over 1,800 pricing and packaging changes in 2025 alone, averaging 3.6 changes per company for the year. That's not iteration. That's a sector that doesn't know what it's selling.
Credit-based pricing specifically exploded. According to Growth Unhinged, Kyle Poyar's survey of 230 B2B and AI companies found credit-based models grew 126% year-on-year, with 29% of respondents now using some form of credit system. Three years ago that number was close to zero.
It's not just new pricing mechanics being added. It's old ones being defended. According to BetterCloud, 68% of SaaS vendors restricted AI features to premium tiers in 2025, and 37% are planning further pricing changes soon. Vendors are still figuring out whether AI is a feature, a product, or a different business entirely, and the pricing page is where that confusion becomes visible to the buyer.
And the pressure is real. Tropic reports that AI-native tool spend grew 94.35% year-over-year among mid-market and enterprise buyers in 2025. Budget is moving fast toward AI tools. Founders feel that window closing, and pricing becomes the lever they reach for first because it feels like the fastest thing to change.
Why Usage-Based Pricing Feels Like the Safe Move
Usage-based pricing has real upside when it's built on a clear value unit. Monetizely found that AI providers using modern usage-based pricing showed 40% higher gross margins and meaningfully lower churn than those still on legacy per-seat models. That's a genuine signal, and it's why boards are pushing founders toward it.
But there's a second reason usage-based pricing is spreading, and it's less flattering. Per-seat pricing is breaking because AI is quietly replacing the seats it used to meter. Userpilot cites a Salesforce case where 90 enterprise accounts saw a 10% reduction in customer-service seat usage as AI agents took over work humans used to do. The company's own AI was cannibalizing its own pricing model.
Faced with that, usage-based pricing looks like an escape hatch. If you can't count seats anymore, count something else: tokens, credits, API calls, whatever resource the product consumes. It feels modern. It feels aligned with how the big AI labs price. And it lets a founder avoid the harder question underneath: what does the customer believe they're paying for.
That avoidance is the problem. A pricing mechanic borrowed from OpenAI's website doesn't carry OpenAI's positioning with it. OpenAI can price by token because the value unit is obvious: you get more intelligence for more tokens. Most B2B startups haven't done the work to make their value unit that obvious. So they end up with a token meter and a buyer who still can't explain, in one sentence, why they're paying for it.
The One-Sentence Test
Here's the test I run with founders before any pricing conversation goes further: can you tell me, in one sentence, what unit of value your buyer actually feels, not what resource your product consumes.
Those are two different things, and most pricing pages confuse them. A resource is what your infrastructure burns: compute, tokens, API calls, storage. A value unit is what the buyer experiences as the reason they're paying you: fewer hours of manual review, a deal that closed faster, a risk that didn't materialize, a decision made with confidence instead of guesswork.
If your pricing page talks about tokens and your buyer thinks about hours saved, you have a translation problem no pricing mechanic will solve. The credit system just makes the gap between "what we charge for" and "what they value" more visible, not less.
I've seen this play out directly. A company I worked with was pricing by API call volume, because that's what the infra team metered internally. Sales kept hitting the same objection in every late-stage deal: the buyer couldn't predict their bill, so procurement stalled it. The fix wasn't a new pricing tier. It was renaming the value unit from "API calls" to "cases resolved," which is what the buyer's own team reported up to their CFO. Same usage data. Different unit of value. Deal velocity changed within a quarter.
Resource-based question: what does our system consume?
Value-based question: what does the buyer's boss ask them to report on?
If those two answers don't match, no pricing model fixes it. Only repositioning does.
What I Tell Israeli Founders Before They Touch Pricing
Most of the Israeli founders I work with are technically excellent and commercially impatient. They want the pricing conversation to be a one-week exercise: look at three competitors, pick a model, ship it before the board meeting. That instinct is understandable given how fast the market is moving, but it's backwards.
Before I let a pricing conversation start, I ask three questions:
First, what does the customer's finance team call this spend internally. If procurement can't categorize the line item without calling you, your value unit isn't clear yet.
Second, what would the customer stop doing if they stopped paying you. Not what feature would disappear, what business outcome would stop happening. If the answer is vague, positioning isn't done, and no pricing model will paper over that.
Third, does the pricing model make the value unit more visible or less visible to the buyer. Token-based pricing can be the right answer for a developer tool where tokens map cleanly to work done. It can be the wrong answer for a workflow tool where the buyer thinks in outcomes, not consumption.
Once those three are answered, the pricing model usually becomes obvious. It stops being a guessing game copied off a competitor's page and starts being a direct expression of the value you've already defined. That's the order that works: positioning first, pricing mechanic second. Reverse it, and you're just redecorating confusion with a usage meter.
Not sure if your pricing problem is actually a positioning problem?
I work with Israeli B2B and AI founders on the positioning work that makes pricing decisions obvious instead of guesswork. Happy to look at where yours is breaking down.
