Pricing Optimization

Companies that have not changed their price in a year usually say the same thing: “We're afraid we'll lose customers.” What is interesting is that this fear rests on no data. How customers would react to a price increase has often never been tested. What is not tested gets managed by fear.

Pricing is the decision the team spends the least time on and the one with the most direct effect on revenue. A marketing campaign or a new feature shows up in revenue months later. Price changes in a day.

When you need it

The signals: you have not changed your price for a long time and the product has improved a lot in that period. Price objections are very rare in sales conversations; a sale that closes too easily is usually a sign of low pricing. Customers of very different sizes pay the same price: expensive for the small one, cheap for the large one. Discounting has become a reflex rather than a rule. Revenue per customer does not grow in proportion to the value the customer gets from you.

It is too early when the product has not proven it creates value. Playing with price while retention is low just moves the problem. If you cannot describe the customer's value, price stays just a number; understand the value first.

What most teams miss

When pricing optimization comes up, “should we raise or lower the number” comes to mind. That is the weakest lever. Two stronger ones: what you attach the price to (the value metric) and how you package it (tiers).

It is not the price number but what the price is tied to that decides your revenue: when the customer grows, do you grow with them?

The value metric is the thing that rises with the value the customer gets from you: seats, transaction volume, connected records, messages sent. A good value metric does three things: the customer can understand it, it correlates with value, and it grows naturally as the customer grows. If a customer pays a flat price while usage grows tenfold, they are taking the value from you without the number changing.

The second thing teams miss is packaging. A three-tier structure (typically “good, better, best”) is not only a price change; it is the customer's decision frame. The middle tier is often the one you aim for; the top tier gives it a reference and the bottom is the entry door. Build tiers around customer segments, not around feature lists.

Third, the existing customer base. Price optimization often focuses on the price for new customers. Yet existing customers are generally the group on the oldest, lowest price, and that is where a price increase has its biggest effect. Most teams leave those customers out of scope, out of fear.

Fourth, price is a communication job. The same increase is accepted when explained well and causes losses when explained badly. What is changing, why, when, and what the customer will get must be said clearly and on time.

The playbook

1. Measure the current state

Pull the distribution of revenue per customer, discount rates, distribution across tiers and the price-objection rate. Which customer segment gets the most value and which pays the least? That table shows where the price is wrong.

2. Question the value metric

How do customers get value from your product? Which unit grows with that value? Is the current price tied to that unit? If not, write it down as a candidate change. Sometimes changing only the tier boundaries gives the same result without raising the number.

3. Research willingness to pay directly

Talk to customers and ask about price perception openly. The Van Westendorp price sensitivity method shows the acceptable price range with four simple questions: at what price is it so cheap you doubt the quality, at what price is it a bargain, at what price is it getting expensive, and at what price is it too expensive. It gives direction even with small samples. Work from the customer's own words, not guesses.

4. Test at small scale

Do not change price across the whole customer base at once. Try it on new customers or a specific segment for a defined period. Read conversion and average contract value together: if conversion dips slightly but revenue rises, the price is moving the right way.

5. Build discount discipline

Who can give how much discount, under which conditions? These rules must be written and the discount must buy something: a longer contract, bigger volume, a reference. A discount with nothing in return turns price into just a negotiable number.

6. Apply a planned price increase to existing customers

Identify customers on the old price and plan an increase that is announced in advance and linked to value. Start with the customers who get the most value and read the reaction. The share of customers who stay without objecting tells you how low the price really was.

What to measure

• Average revenue per account (ARPA) and net revenue retention (NRR). The two measures that best show the effect of price and packaging; both lag.
• Conversion rate and average deal size. Must be read together in a price test; a leading indicator.
• Discount rate and distribution. Shows whether discounts are given within discipline or as a reflex.
• Churn after a price increase. Check whether it exceeds expectations in the ninety days after the increase.

Pitfalls

Seeing price as only a number. Changing the number and leaving the value metric and packaging alone usually yields a small gain. The big difference comes from changing the structure.

Copying a competitor's price. A competitor's price reflects their cost, their goal and their product, not yours. Look at it as a reference, do not use it as arithmetic.

Raising without explaining. If customers do not understand why it went up, they see the increase as arbitrary. The communication plan is half the increase.

Setting price once and forgetting it. The product evolves, the customer profile changes, and price should be reviewed along with them. Build the habit of an annual price review.

When it does not work

If you do not set the price, optimization has less room to work. When a tender specification, a public tariff or a buyer's procurement policy has already drawn the ceiling, redesigning your value metric and tiers has no counterpart.

Rule of thumb. Count how many of your last five quotes had the number written by the customer's side; if most, shift your effort to scope and contract terms.

The second limit appears where you want to tie price to usage. In 2024 Kyle Poyar wrote that a majority of usage-based businesses run a hybrid model, especially those selling to smaller customers, who “might be more sensitive to unplanned expenses.” Our reading: if you sell to small customers, a pure usage metric makes the bill unpredictable; consider a cap or a fixed base fee.

The counter-argument: critics of this caution say each month of waiting costs revenue. They have a point; still, where price is not the deciding factor you cannot tell a loss after an increase from a product problem.

How AI changes this

In classic software, as a customer grows, the value they get rises, the price rises, and your cost barely moves. In an AI product a third measure appears: the inference cost of every use. Price and cost no longer have to be tied to the same thing.

If you sell per seat but your cost accrues per use, that is exactly where the two come apart. The heaviest seat is usually the one getting the most value, yet it is also the most expensive to serve, and it pays the same as everyone else. Raising the price does not fix this; it nudges the average and leaves the gap where it was.

Rule of thumb. Before touching the number, write both metrics side by side; e.g. if value tracks user count and cost tracks call count, the gap exists. Close it first: a usage allowance, a tier limit or a hybrid structure.

Our reading: a price increase only makes sense after that. The Pricing Model Picker can show you which model fits in a few questions.

Start this week

1. Sort your customer base by price paid and look at how much the twenty lowest-paying customers use the product.
2. Book half-hour conversations with five customers and ask the four Van Westendorp questions.
3. List the last three months of discounts; for each, write what you got in return, and flag those with nothing in return.