A finance lead at a mid-market fintech company recently forwarded us a support ticket that made her wince. A prospect had signed up for a "starter" plan advertised at a flat monthly rate, connected the product to two data sources, and watched the bill triple once usage-based charges for API calls and AI-assisted reconciliation kicked in. Nothing on the page was technically false, the starter price was real and the overage terms sat in a footnote, but the page was built for a world of simple per-seat math and the product had quietly moved past that world months earlier. The prospect did not churn because the product was bad. They churned because the price they experienced and the price they were shown were two different things.
That gap is showing up across every industry Aero works in: a biotech SaaS tool billing by sample processed instead of by user, a professional services platform charging per AI-generated document, a media analytics product metering agent queries against a dataset. The underlying shift is the same everywhere. Pricing is moving from a fixed cost per person to a variable cost per unit of AI work, and most pricing pages have not caught up.
This shift is not a niche trend. Gartner projects that 40 percent of enterprise applications will feature task-specific AI agents by the end of 2026, up from less than 5 percent in 2025, a pace of change that is disrupting the seat-based pricing conventions most SaaS pricing pages were designed around, according to Gartner's own research. The stakes for getting this right are not abstract either: Gartner separately estimates that 234 billion dollars in enterprise application software spend is at risk from agentic AI, as buyers reassess what they are paying for and how, according to Gartner. When that much budget is being re-evaluated, a pricing page that cannot clearly answer "what will this actually cost me" is not a copywriting problem. It is a conversion and retention risk.
Most pricing page advice is written for a world of three flat tiers and a feature comparison table. That advice still works for simple products. It breaks down the moment a "unit" enters the picture, whether that unit is an AI agent action, a document processed, a query answered, or a seat plus usage on top. The job of an AI pricing page is different from a traditional SaaS pricing page: it has to teach the buyer what a unit is, show them roughly how many units they will use, and let them see the resulting cost before they commit, not after the first invoice.
Two B2B software companies in adjacent markets both moved to hybrid pricing this year: a base subscription plus metered charges for AI-driven features. The first company kept its existing pricing page structure and simply added a line reading "plus usage" under each tier, with a link to a separate rate card. Sales calls became pricing calls. Self-serve signups spiked in the first week of each billing cycle and dropped sharply once usage charges appeared, a pattern their own billing dashboard made obvious once they looked for it.
The second company rebuilt the page around three things: a plain-language definition of their billing unit shown before any price, an interactive slider letting a visitor estimate monthly cost from a rough usage number, and a visible answer to "what happens if I go over." Their base tier pricing and feature set were nearly identical to the first company's. The difference was that a prospect could leave the page with a real number in mind instead of a range. Billing surprise tickets fell in the cycles after the redesign, and pricing objections surfaced less often late in the sales process, because the objection had already been answered on the page itself.
This is Aero's own practical lens for auditing a pricing page built around AI or usage-based billing, not an established industry standard. Five questions we run against any pricing page before it ships:
A page that fails two or more of these is very likely trading short-term signups for long-term trust, the same trade the fintech company in our opening example made without meaning to.
Usage-based pricing is also the first place many teams encounter a harder question: who, or what, is doing the buying. As AI agents start evaluating vendors and completing purchases on a human buyer's behalf, pricing pages need to be legible to more than one kind of visitor. We covered what that means for checkout, subscription, and account design in designing for agentic commerce. Pricing legibility and agent-readable commerce are two sides of the same shift: buyers, human or automated, need a page that states its terms plainly instead of one built to be puzzled out.
Not necessarily. The conversion drop we see is not usually caused by usage-based pricing itself. It is caused by usage-based pricing shown on a page designed for flat pricing, where the buyer cannot tell what they will actually pay.
A public rate card is good practice, but it is not sufficient on its own. Few buyers will multiply a per-unit rate by their expected usage without help. A calculator or worked example does that work for them and tends to reduce both abandoned signups and post-signup surprise.
Hand it to someone outside your company and time how long it takes them to say a specific monthly dollar figure out loud. If they cannot do it in under a minute, or if they need to contact sales to get there, the page is failing the legibility test above.
Not sure if your pricing page can survive contact with a usage-based world? Talk to Aero.
Usage-based and hybrid pricing are becoming the default for AI-powered products, but most pricing pages still assume a simple per-seat world. Here is Aero's practical framework for pricing pages that build trust instead of anxiety.
AI design tools now touch most product teams' workflows, and that convenience carries a real risk: interfaces that quietly converge on the same patterns. Here is Aero's practical framework for using AI design tools without losing what makes a product recognizably yours.
AI personalization is now standard across SaaS, fintech, and healthcare products, but accuracy alone doesn't earn user trust. Here is Aero's practical framework for designing personalization that discloses what it's doing instead of quietly guessing.
