What Pricing Actually Is

The idea in one minute#

Price is the share of the value you created that you keep. It is not a function of what the product cost to build, which is sunk and irrelevant, and it is not a function of what competitors charge, which tells you about their costs and their segment rather than about your value. Pricing from cost or from the competitor is the single most common error, and it is an error in both directions: it undercharges when you have created a lot of value cheaply, and it overcharges when you have spent heavily on something nobody wanted. The useful question is always what the customer’s alternative costs them, because that is the number your price is actually measured against.

A picture#

flowchart LR
  ALT[":i-wallet: <b>Their alternative</b><br/><small>spreadsheet, hire, agency, nothing</small>"] --> VAL[":i-trending-up: <b>Value created</b><br/><small>what you save or make them</small>"]
  VAL --> BAND[":i-scale: <b>The band</b><br/><small>your cost at the bottom, their value at the top</small>"]
  BAND --> P{":i-tag: <b>Price</b><br/><small>how much of it you keep</small>"}
  P --> CAP[(":i-coins: <b>Captured</b>")]
  P --> LEFT[(":i-handshake: <b>Left with them</b><br/><small>the reason they buy</small>")]
  COST[":i-receipt: <b>What it cost you to build</b>"] -.->|"sunk, and not an input"| X[":i-triangle-alert: <b>Cost plus pricing</b>"]
  class ALT,COST neutral
  class VAL compute
  class BAND,P queue
  class CAP,LEFT memory
  class X warn

Note that some value must be left with the customer. A price that captures everything is a price nobody accepts, because the reason to switch disappears.

How it really works#

The band#

Your cost to serve is the floor. The value the customer gets is the ceiling. Everything in between is negotiable, and the band is usually enormous, which is why pricing feels arbitrary and why small changes in approach move revenue so much.

For most software the floor is low and irrelevant to the decision; the interesting end is the ceiling, which means the work is understanding value rather than understanding cost. The exception is any product with a real marginal cost, which now includes anything with a model in it, where the floor comes up to meet you.

Value is measured against the alternative#

A customer does not compare your price to zero. They compare it to what they do today, which costs them something already:

The last row is where most products actually live, and it is why discovery and pricing are the same conversation: if nobody can state what the problem costs them today, you cannot price it and probably should not build it.

Why cost plus pricing is wrong in both directions#

If you build something in two weeks that saves a customer fifty thousand a year, cost plus pricing charges them for two weeks. If you spend two years building something nobody values, cost plus pricing produces a number no one will pay. The cost of building tells you whether the business works at a given price, which is a different and later question.

Why copying a competitor is also wrong#

Their price encodes their cost structure, their segment, their funding position and their history. Copying it imports all four. It is worth knowing as a reference point for what the market finds normal, and it is not an input to what you should charge.

Price is a product decision#

Three reasons it belongs with product rather than purely with finance. The value metric determines what the product optimises for, since customers will change behaviour to manage their bill. Tiers are a design of which problems you solve for whom. And a price change is a product migration with communication, grandfathering and churn risk, which is programme work rather than a number edit.

The uncomfortable part#

Pricing well requires asking people what things cost them and what they would pay, which feels intrusive, and most teams avoid it by inferring a price from internal logic instead. Everything in this topic is downstream of being willing to have that conversation.

What fills the box in 2026#

Billing and metering through Stripe Billing, Orb, Metronome or Chargebee, and the choice matters more once usage pricing is involved. Willingness to pay research is interviews and surveys rather than software. Anything with a model in it needs cost per task from AI Cost and Margin before a price can be set at all, because that is the case where the floor genuinely constrains you.

Sector lens#

Remember this#

• Price is the share of created value you keep, not a markup on cost.
• The band runs from your cost to their value, and it is usually wide.
• Value is measured against their actual alternative, which is often a spreadsheet or nothing.
• Cost plus pricing undercharges cheap wins and overcharges expensive mistakes.
• A competitor’s price encodes their situation, not your value.
• Some value must be left with the customer, or there is no reason to switch.

Try it#

1. For your product, write down the alternative five real customers use today and what it costs them. If you cannot fill three rows, that is the finding.
2. Compute your band: cost to serve one customer, and the annual value one customer gets. Where in that band does your price sit?
3. Ask one customer directly what they were doing before, and what that cost. Most people answer readily and nobody has asked them.

Check yourself#

1. Why is the cost of building a product irrelevant to its price, and what question is it relevant to?
2. What changes about the band when the product has a real marginal cost per use?
3. Why does pricing belong with product rather than only with finance? Give the three mechanisms.