Launch pricing: models, value metrics, and the evidence to gather first

Gather evidence before you choose a number

Pricing set from your costs, your feelings, or a competitor's page is a guess wearing a suit. Evidence means asking real prospects specific questions about money and writing down their literal answers, plus any amount somebody has already handed over for a thing like this. Your infrastructure bill tells you the floor and nothing about the ceiling, and the gap between them is where the decision lives.

The cheapest useful instrument is a pair of Van Westendorp questions, asked of about ten prospects who resemble your buyer. At what price would this be so expensive that you would not consider it. And at what price would you start doubting the quality. Do not average the two. You want the band between them, and the shape of the answers: ten clustered in a narrow range means a settled category expectation you can price against, while ten spread across an order of magnitude means you have not decided who you are selling to.

Ask both on a call rather than in a form, and stay quiet after each one; the pause is where the reasoning comes out, and the reasoning is worth more than the figure. Somebody who says forty pounds feels wrong because they would have to explain it to their finance lead has named an approval threshold, which binds harder than willingness to pay.

Then look for money that has already moved: a deposit, a paid pilot, a competitor invoice a prospect shows you; any of those outranks ten opinions. If nobody has ever paid for this job, price it low enough to make the first ten customers easy to win.

Choose the value metric before you choose the price

The value metric is the thing you charge per unit of: a seat, a thousand API calls, an invoice processed, a flat monthly fee. Choose it before the number, because the metric is structural and the number is a setting. The test is whether the metric grows as the customer succeeds, and plenty of launch models fail in the opposite direction.

Per-seat pricing on a product that reduces the headcount needed to do a job is the clearest example. The better it works, the fewer seats your customer needs, so your best outcome for them is a smaller invoice for you. An invoicing tool for freelance designers has a milder version: a solo designer is one seat forever, so seats cannot carry growth and you will bolt a usage limit onto the side within a year. The launch readiness framework puts this check ahead of every other pricing question, and it is the check founders score most generously, because a wrong metric is not visibly wrong until a couple of hundred accounts sit on it.

Hybrids are normal and often correct: a seat price with a usage allowance, or a platform fee plus a per-transaction rate. Trouble starts with a hybrid nobody can compute in their head: if a prospect cannot estimate their own bill within twenty per cent while looking at the page, the model is too complicated for a launch.

Publish your pricing, and understand the second reason to do it

Publish the numbers. The familiar argument is trust and shortlisting: a buyer who cannot see a price assumes it is high, and a sales process with it, then moves to a competitor who told them. Hiding price does not qualify leads, it selects for patience, which does not correlate with budget.

The second reason, which is newer and less discussed

A hidden price removes you from every conversation about cost that happens without you in the room. Comparison articles need a figure for the table, so writers omit you or estimate wildly. Buying shortlists get assembled in Slack threads and spreadsheets you never see, and a row with a blank price cell rarely survives the first cut. And when somebody asks an assistant what tools in your category cost, the assistant answers from something. If your page is silent it uses whatever else exists: a two-year-old review, a forum comment quoting a price you no longer charge, a competitor's comparison page. You never get to correct that answer.

Freshness compounds the problem. Analyses of AI citations consistently find that roughly half of cited pages were updated within the previous thirteen weeks, so recency does real work in what gets quoted. A pricing page carrying last year's numbers is worse than none, because it is a confident, retrievable, wrong answer with your own domain behind it. Put a last reviewed date on it and keep it true. The wider mechanics sit in answer engine optimisation for product launches.

The legitimate exception is an enterprise motion where the smallest deal genuinely requires scoping. Even then, publish a starting figure: from £1,200 a month tells the visitor whether they are in the right shop.

Three tiers is a convention, not a law

Every tier needs a real buyer you can describe in a sentence and a reason to exist that is not symmetry. Three became the default because a lot of successful products had three distinct buyers, and the pattern got copied by products with one. If you can only name two real buyers, ship two tiers and use the space to explain the product.

The top tier that exists only to anchor the middle one is the common failure. Anchoring works when the expensive tier is plausible, meaning a visitor can picture the organisation that buys it. It stops working when the tier is ornamental, because the reaction is not that the middle one looks reasonable but that the page makes no sense. If you cannot name a customer who would choose the top tier, you have built a distraction rather than an anchor.

Name tiers for the buyer rather than for metals. Solo, Studio and Agency tell a visitor which row is theirs before they read a single feature; Bronze, Silver and Gold ask them to work it out from a comparison grid. Tier names teach a buyer more about who you serve than your headline does, so they deserve the same scrutiny as the rest of your launch positioning. Check the upgrade gate too: if a new account hits it in the first hour, you have built a paywall with extra steps rather than a tier.

Design the end of the trial before you launch

More launch revenue is lost at trial expiry than anywhere else in the funnel, and it is lost to inattention rather than to price. Teams spend six weeks on the signup flow and forty minutes on day fifteen, when the decision actually gets made. Four questions need written answers before you open the doors.

1. What the product does at expiry. Hard lock, read-only, or degraded to a free tier. Read-only converts better in most products, because the customer still sees the work they did and feels the loss of editing rather than of everything.
2. What the user sees. An in-product state, not just an email. The expiry screen should show what they built, what they are losing, and one way to continue. Where the only signal is an email, your conversion rate is decided by inbox placement.
3. What the emails say and when they arrive. A reminder several days before expiry, one on the day, one afterwards. The pre-expiry message should reference what the person actually did: your trial ends Friday is a deadline, while your trial ends Friday and the four recurring invoices you set up will stop sending is a consequence. Accounts that never activated need a different sequence entirely.
4. What happens to their data. Decide the retention window, write it on the page, and honour it. Thirty to ninety days costs almost nothing in storage and converts returners who come back when the problem recurs.

Decide too whether the trial asks for a card up front. Card-required trials convert a much higher percentage of a much smaller number of starts, and at launch the no-card version teaches you more, because the people who drop out are telling you about the product rather than the form. Sequencing this belongs in the 90 day pre-launch runbook rather than in launch week.

Changing price later, and what it costs

You will change your price, and that is fine. Raising it on new customers only is close to free: change the page, keep existing accounts, move on. The cost arrives when you want existing customers to move too, and it comes in three currencies.

The first is trust. A price rise applied to people who bought at a lower number reads as a change in the deal, however carefully you word the email, and the churn shows up over the following two billing cycles rather than immediately. Grandfathering avoids that, at the cost of the second currency, which is operational: every legacy price is a plan in your billing system and a branch in your support scripts. Three grandfathered cohorts and a half completed migration is how a two-person company acquires a pricing operations problem.

The third cost matters most and gets noticed last. Changing the value metric is an order of magnitude more expensive than changing the number, because it is a re-contracting exercise rather than a price change. Every account has to be remapped from the old unit to the new one, and some customers end up worse off under any mapping you choose. Billing logic, invoices, usage reporting and every published comparison of your product assumed the old unit. Teams moving from per-seat to usage-based typically run both models in parallel for a year. That is why the metric deserves the evidence gathering and the number can be a decision you make on a Tuesday.

Pricing failure patterns at launch

Five shapes recur, and the first two do most of the damage.

• Priced from cost. The founder totals the hosting bill, adds a margin that feels responsible, and lands on nineteen a month for something that saves four hours a week. Cost sets your floor. Value sets your price, and the distance between them is the business.
• Priced against a competitor serving a different buyer. Undercutting a product built for two-hundred-person companies claims you are the cheap version of something your buyer never considered. Price against the alternative they are weighing: a spreadsheet, a contractor's day rate, or doing nothing.
• A free tier so generous nobody upgrades. The limits should sit where a user is succeeding, not where they are still evaluating. If it comfortably covers the whole job for your main use case, you have built a product you give away with a paid tier attached for the guilty.
• Annual only, sold to a buyer who cannot commit annually. Annual billing helps cash flow and hurts conversion, and at launch the conversion and the learning are worth more. Offer both, discount the annual, let the buyer choose. Small teams cannot commit for twelve months to a tool they have used for eleven days.
• Contact us on every tier. This tells a self-serve buyer they are not welcome and an enterprise buyer nothing. Where a call is genuinely required, require it on that tier and publish everything below it.

Most of these are the same mistake in different clothing: pricing decided by looking inward at your costs or your comfort rather than outward at what the buyer gets and what they can get approved. It is why pricing turns up so often in the ways launches actually fail without being named as the cause, since a wrong value metric looks like a marketing problem for six months before anyone traces it back. The launch readiness assessment scores your commercial shape alongside positioning and activation, and the rest of the launch guides take those apart.