Keep your pricing model,
or change it?

As of 2026, pricing models are diversifying faster than at any point before. The era of setting a price once and leaving it for a year is over. Product usage data, customer behaviour analysis, and the volatility of AI infrastructure costs are together turning price into an operational area that has to move in something close to real time.

This article covers the six models spreading fastest right now, drawn from the pricing structures we have rebuilt over the past year.

1. Pure Subscription

A flat monthly or annual fee. Still the most widely used model, but its limits are becoming clearer when it stands alone.

• Strengths — revenue is predictable, customers grasp it immediately, and the sales cycle stays simple.
• Limits — you fail to recover enough value from heavy users, and the entry barrier is high for light ones.
• Fits when — usage varies little between customers. Collaboration and productivity tools.

2. Usage-Based Pricing

Charging on a measurable unit: API calls, transactions, gigabytes, active users. This was the most widely adopted model through 2024 and 2025.

• Strengths — a low entry barrier, and revenue that moves with usage, which protects margin.
• Limits — revenue becomes hard to forecast, and the customer cannot forecast their own cost either, which creates unease.
• Fits when — value scales strongly with usage. Infrastructure, communications, payments, AI APIs.

Usage-based pricing removes the barrier to entry, but it breaks revenue predictability. That is why hybrids are becoming more common than usage-based alone.

3. Hybrid (subscription + usage)

A flat base subscription with usage above the included allowance charged on top. The two models cancel out each other’s weaknesses.

• The base tier secures revenue predictability
• Overage charging protects margin on heavy users
• The included allowance per tier has to be defined clearly, or customers lose the plot

This is becoming the default SaaS model in 2026. AWS, Datadog, and Twilio all take this shape.

4. Outcome-Based / Performance-Based

Charging in proportion to the result the customer obtains: revenue gained, cost saved, conversion improved. The strongest value alignment available, and the hardest to run.

• Strengths — “no result, no cost” from the customer’s side, which is a powerful reason to adopt.
• Limits — measuring the result and agreeing attribution is difficult, and the sales cycle lengthens.
• Fits when — the outcome is clearly measurable. Marketing and sales automation, AI agents.

5. Tiered + Add-On

Three or four base packages, with optional modules sold on top. Spreading quickly in enterprise SaaS.

• Core capability sits in the tier, supporting capability is split out as options, which raises average revenue per account
• Customers can assemble what matches their workflow
• Pricing page complexity grows fast, so the UX of the page itself becomes a design problem

6. Agent-Based / Per-Resolution

Charging per unit of work an AI agent completes: a ticket resolved, a report written, an enquiry answered. Emerging quickly since late 2025.

• Has the potential to replace per-seat charging in SaaS
• Lets the customer compare cost against a human intuitively
• Managing margin against agent operating cost (GPU, API) is the whole game

Which model should you choose

There is no single right answer to “which model should we move to”. But checking three axes narrows it down.

In practice we are asked a different question more often: should we leave the current model alone, or change it. Answering that requires seeing, in numbers, where the current structure is leaking margin.

Closing

Pricing in 2026 is moving out of the territory of “pick a model” and into the territory of “combine and operate”. Setting a price once and leaving it for a year is finished. Moving price into an operable system is the work of the next twelve months.

📩 If you would like to know where your own pricing stands, start with the Pricing Check. Three minutes is enough to see where it can improve.