Framework

The ACE framework: how to choose a value metric

The ACE framework is our test for choosing a value metric. It scores each candidate metric on three things: Auditability, Cost-Margin Fit and Ease. A metric that scores high on all three is a price you can defend to the customer and still earn margin on. Cristian Varga developed it at Smarter Monetization to test the pricing axis of AI and SaaS products.

What a value metric is

A value metric is the unit you charge for: seats, API calls, documents processed, tickets resolved. It is not the pricing model. The model is how you charge (flat, tiered, usage-based or hybrid), and many models can wrap the same metric. The metric is what you charge for, and the right one rises as the customer succeeds.

The three pillars

Each pillar has two criteria. You score each criterion from 1 to 5, so a perfect metric scores 30.

How to run it

List four to six candidate metrics from your value map. Score each one on the six criteria, where 1 is poor, 3 is average and 5 is excellent. Add up the pillar scores. Then test the top one or two with customers using one question: "If we charged per this, would that feel fair?"

Don't stop at the total, because two metrics can tie and fail in different ways. Look at how the score splits across the pillars. A metric that is strong on Auditability and Ease but weak on Cost-Margin Fit is easy to sell and easy to lose money on.

A worked example

The scores below are illustrative. They come from the scorecard further down this page. Four common metrics for a support automation product.

Active users and resolved tickets both score 24 out of 30, but they behave very differently. With active users, a heavy account pays the same as a light one, so price stays flat while your cost goes up. Resolved tickets gives up some Ease and gains five points on Cost-Margin Fit, because revenue rises with the work done. For an AI product with variable cost, that trade is usually the right one.

Workspaces is the trap. It scores 23, nearly perfect on Auditability and Ease, close to zero on Cost-Margin Fit, and has almost no connection to value.

Why we built it for AI products

Agentic workflows changed the cost structure under pricing. One agent task can run dozens of inference calls, and each call gets more expensive as context builds up. A tangled support ticket can cost $1.50 to resolve while a simple one costs 3 cents. A seat metric ignores all of that, which is why ACE puts so much weight on Cost-Margin Fit, and why cost alignment is a criterion at all: AI products have real marginal cost.

Of the three pillars, Auditability is the one that most often kills a metric. Once a buyer can't check what the agent did, every outcome-pricing conversation turns into a discount conversation.

Where ACE fits

ACE is the value-metric test inside our value-based pricing method. It is also the first check in the pricing teardown, where a wrong value metric is the first of five places revenue leaks. The usual symptom is heavy users paying the same as light users, which puts a low ceiling on your revenue. Seats on a product whose value comes from usage cap the upside. Per-gigabyte pricing on a product whose value is speed rewards the wrong behavior. Twilio charges per API call because that is when value lands. Stripe charges on dollars processed because that tracks the customer's revenue. DocuSign charges per document because that tracks deal flow.

A high score doesn't settle the question on its own. On an enterprise AI engineering platform, per-developer came out on top in ACE. We still stress-tested it in buyer sessions by asking whether developer headcount would grow, hold or shrink if the product worked. If buyers said it would shrink, the per-developer metric would need to change before launch.

Related reading: the SaaS pricing models guide, the outcome-based pricing guide, and the pricing teardown.

Score your own metrics

Questions about the ACE framework