Talk to an Expert

Foundations · AI line

What a meter does to customer behavior

5 min on the light depth

The frame

A meter changes how customers use your product before it changes what they pay. The value metric is the unit your price attaches to; a meter is that unit read as a measuring device while the customer works. When the meter runs on every action, each action becomes a visible cost line, and finance governs a cost line by cutting the activity behind it. For an AI capability the activity cut first is exploration, the work of finding which jobs deserve the price. The customer rations, and the usage that remains reads as demand already captured.

The first report you see will be a usage spike and a strong quarter. Read it as the first invoice cycle before customers reorganize around the bill. Then the spend commitment arrives, the quantity of usage or spend a customer is bound to for the term in exchange for a lower rate. Buyers size it against their fear of the meter, and every commitment embeds a growth assumption nobody writes down; an account can grow every quarter and still miss the number at renewal. One rule governs both: price relates to the value the customer extracts from using the product, never to use itself, so a meter on raw activity counts the wrong unit. Usage is always skewed, and the average customer an allowance was built for does not exist.

The field guide argues that usage pricing backfires the same way every time and maps the public cases to read your own product against. The exploration piece argues that variable pricing suppresses the discovery vendors need their customers to do. The commit piece argues that a spend commitment is a discount attached to a forecast, and breaks on deceleration before anything declines.

Which activity does your customer's finance team cut when your meter runs hot, and is it the work that finds your next use case?

Three things to say in the room this week

  • Before we celebrate the usage spike, I want to know what customers do once they see the bill.
  • If our meter counts an action the customer can stop, the customer will stop it.
  • Every spend commitment we signed assumed a growth rate; show me which accounts are growing slower than that.

Core reading

Go deeper: the heavy depth

Read the core in this order.

  • When usage-based pricing backfires, the field guide. Start here. Names the one mechanism and places each public case by where it puts the consumption risk.
  • Variable AI pricing suppresses exploration. Exploration done on purpose is the discovery work the vendor needs; a meter that prices it ends it.
  • The growth assumption inside AI spend commitments. The commit as a discount attached to a forecast: why it breaks on deceleration while committed revenue still prints records.

Then the shelf, lighter.

  • When the meter catches the spike. Why a record quarter can come before a ceiling when the meter is set on an action the customer controls.
  • The consumption-pricing warning, years later. The older call that consumption transfers cost risk to the customer, and the diagnostic behind it.
  • Questioning the assumption about consumption pricing. Under consumption or subscription is the same value metric decision.
  • Hard caps versus budget alerts. Read it for the posture decision about who pays for a runaway workload.
  • Volume discounts in AI consumption pricing. Why linear cost brings the quantity discount back, following behavior that lowers cost to serve.
  • From coarse volume schedules to a pricing surface. A net price for any commitment a customer might make; stepped schedules leak at consumption volumes.
  • Pricing software when you cannot see usage yet. Instrument first, price second; until then, price on bounded units the customer can verify.

The research hub on usage-based and nonlinear pricing holds the peer-reviewed work on plan choice under usage uncertainty, each study read in full with a written ceiling on what it supports.

The next module on the AI track is The unit for an AI capability: which unit above the token the price attaches to.

The shelf

Which depth did you take?