Will User Based or Credit Based Pricing Prevail in the AI Era?
My prediction is that “Credit based pricing will absorb user based pricing.”
TL;DR (Too Long; Didn’t Read)
Implications for B2B SaaS and AI leaders
Reframe your pricing debates: Shift from “seats vs credits” to “how do we balance flexibility and predictability under rapid AI-driven change?”.
Treat credits as the default design pattern: Design credits as an internal currency that can represent user access, features, and AI actions in a single system.
Embed user-based logic inside credits: Allocate a standard credit budget per user or per role so user-based pricing becomes a special case rather than the core model.
Design for prediction, not just billing: Build credit models that feed prediction engines, enabling better usage forecasts and budget conversations with finance and customers.
Use generosity and forgiveness intentionally: Calibrate overage, rollover, and pooling policies to reduce buyer anxiety about credits while preserving margin.
Align credits to value events: Set credit consumption at a level of granularity where each “spend” can be clearly tied to business outcomes (e.g., resolved cases, high‑value AI outputs).
The User Based vs. Credit Based Debate
As we enter 2026 one of the raging debates in the pricing world is whether credit based pricing will become a standard pricing pattern or if there will be a return to user based pricing.
In late 2025 there was a flurry of comments that user based pricing is the norm and there will be a return to number of users as the main pricing metric for everything from B2B SaaS applications, to AI agents, to background services accessed through APIs.
In the Dec. 3, 2025 earnings call Salesforce CEO Marc Benioff said …
“When we first started with Agentforce, we were talking about, oh, it’s going to be so much per conversation... but customers have pushed for more flexibility,”
This comment led to many headlines in the analyst and commentator community such as …
“Salesforce says per-user pricing will be new AI norm”
from Craig Hale on Tech Radar and
“Salesforce opts for seat-based AI licensing as customers demand predictability”
Lindsay Clark in The Register.
Close readers of pricing threads on LinkedIn have also noticed a lot of people leaning in on this, with many expressing their own and buyer resistance to credit based pricing models.
Even Elena Verna, who leads growth for vibe coding app Lovable and is one of the great innovators in credit based pricing, leaned in on this. In a recent webinar with Metronome, she said that many users object to credit based pricing and would prefer user based pricing as they find it more familiar and easier to wrap their heads around.
On the other hand, people like Brandon Hickie from LinkedIn have expressed clear support for credit based pricing. See my webinar with him from October of last year “The New Currency: How Credit-Based Pricing Accelerates SaaS Growth” and suggest that it may become the standard approach to pricing model design.
What is credit based pricing?
Credit-based pricing is a pricing model where customers prepay for a pool of “credits” (an internal currency) to consume specific actions or results from a product, tying price directly to usage and value delivered. This model provides flexibility while creating predictable revenue for vendors. It’s a hybrid-friendly pattern where actions like running reports, answering AI queries, or resolving complaints cost a set number of credits, allowing for transparent, granular cost management and experimentation.
Credit based pricing got a lot of attention in 2025
Credit based pricing is a flash in the pan and the world will revert to more conventional approaches
Credit based pricing is a bridge to some new normal that has not yet emerged
Credit based pricing is one of the new standard patterns
Of course all of these can be true in different parts of the market.
We may well see some scenarios where companies revert to or emphasize user based pricing. Salesforce is emphasizing user based while keeping its credit based pricing models intact.
Other companies are doubling down on credit based pricing, using credits to incent action and engagement, and developing user tools like credit wallets that make it easier to understand and manage credits.
What else could emerge from credit based pricing? That is the open space for pricing design and an area that I will explore this year using scenario planning and concept blending as my tools.
Benioff on user based pricing and flexibility
Look at the quote from Marc Beniff again. “”When we first started with Agentforce, we were talking about, oh, it’s going to be so much per conversation... but customers have pushed for more flexibility,”
Given the pace of innovation both buyers and vendors need flexibility as they are not sure how much they will use any function or how valuable it may be to them.
Is per user pricing the solution?
Benioff may like it because it is familiar to him, indeed he is one of the key people who introduced it and replaced older pricing models like per server or per site. Ellison was once at Oracle, which at that time had some horribly complex pricing models. He is one of the key people who forced pricing simplicity and clarity on SaaS. He may see credit based pricing as the industry he created sliding back into complexity.
It is not clear though that user based pricing is fit for purpose in a world of rapid change.
Flexibility and Predictability are What Matters
What people on both sides are really looking for is a combination of flexibility and predictability. The challenge is that these two goals are often in conflict. The more flexibility, the less predictability.
User based pricing is seen to solve this as (i) the number of users is known (though it is not known if the user will actually use any specific functionality, or use anything at all) and (ii) users can be given access to all functionality.
Ellison believes that Salesforce AI is creating “three or four times or 10 times more value” than its conventional products so that there is plenty of room to increase prices to cover the growing compute costs.
User based pricing only works for the vendor if it can convince the buyer to pay for functionality that is not being used. Value connection is weak in most cases.
Are there other ways to solve the flexibility vs. predictability paradox?
Yes. Well designed credit based pricing can do this, especially when combined with a good prediction model.
Credit based pricing asks the buyer to commit to buying a pool of credits but allows them to use them in different ways. Well designed credit pricing models can also have pooling provisions (unused credits can be moved between users) and roll over provisions (credits not used in one time period can be ‘rolled over’ into the next time period. New ways to use credits can easily be introduced (new functionality, new actions).
This leaves open the question of “How many credits should the buyer commit to?” This has been the source of a lot of frustration, especially among buyers/users of vibe coding apps.
There are a few solutions to this.
Vendors can be generous, providing enough credits that most users are likely to have enough to do what they need to do for the key use cases
Vendors can be forgiving, allowing users to exceed their monthly credit quota by some amount and then adjusting the credits for the next period
Prediction models can be built that help both parties understand the probable pattern of use and the number of credits needed
It has become increasingly easy to build prediction models. Several years ago I introduced the metric ‘predictive engagement.’ This is a measure of engagement that predicts future engagement. Not all use predicts future use. In fact, some use patterns actually signal churn. If a user begins downloading content that may signal they are preparing to move somewhere else.
For pricing, and especially credit based pricing, one wants to design the pricing model so that future use is predictable. Design for prediction needs to be part of credit pricing design.
Credit Based Pricing Can Subsume Per User Pricing
I recently designed a pricing model for the digital services of a large engineering company. One division used a pricing system based on points. They had a standard price per point and a set of heuristics (an approach to problem solving that employs a pragmatic method that is not fully optimized, perfected, or rationalized, but is nevertheless “good enough” as an approximation). Once we had examined that system we were able to devise different point systems for each solution, all of which could use the same price per point. This makes it much easier to price deals that combine multiple solutions. One simply adds up all the points and multiplies by one price per point. It is not far from this point based pricing to credit based pricing.
It is easy to use the number of users in the heuristics to calculate points. One can even accommodate different types of user and patterns of use with a point system. The same is even more true of a credit system, which is a point system plus provisions for redistribution of credits between people or between time periods.
Conclusion
Will user based or credit based pricing models prevail? Credit based. Where it makes sense the credits can be used to integrate the number of users into the credit based system.
Why will a credit based approach prevail?
Credit based pricing is more flexible and can more easily accommodate innovations and different usage patterns.
Credit based pricing is more predictable, one can design credit based pricing models that feed prediction engines and build in feedback loops so that the predictions improve over time.
Credit based pricing is easier to align with value. When designing these models one begins by finding the level of granularity that makes it possible to map credit consumption to value.
User based pricing is a subset of credit based pricing and user based models can be implemented within a credit framework.
Contact me at steven.forth@gmail.com for a deep dive into the design and adoption of credit based pricing.
This lands for me.
What’s really being debated here isn’t users vs. credits as pricing mechanics. It’s how much uncertainty the seller absorbs versus how much gets pushed onto the buyer.
User-based pricing feels predictable because the variable is familiar, not because risk disappears. Credits feel risky because they surface uncertainty that was always there. Usage just makes it visible.
The interesting move is treating credits as infrastructure, not as the offer. Once credits become the internal accounting system, you can express seats, roles, actions, outcomes, and even forgiveness policies inside a single framework. At that point, “per user” isn’t replaced. It’s contextualized.
The practical question going forward is not which metric wins, but how pricing systems evolve as buyers move from needing familiarity to needing flexibility, without forcing that transition faster than buyers can justify.
Excellent post Steven!
I believe credits won not by design, but that means that we owe customers three things as pricing people:
* Transparency: show the credit/real money exchange rate, even when it hurts
* Predictability: don't change rates mid-contract or too frequently
* Reconciliation: credits should map credits to both costs AND outcomes
I hate it, but credits are here to stay.