Pricing strategy: frameworks & methodology

Three decisions send teams back to the pricing model: bill on a different unit, split one plan into several, or stop selling a perpetual licence and start billing monthly. Each is arithmetic before it is strategy, and each fails in the same way, by being argued in a meeting rather than modelled in a sheet. This lesson gives the sequence: shortlist and score the billing unit, test it against real willingness to pay (the concept the foundations lesson sets out), then model the revenue effect before anyone edits a pricing page.

Step 1: score the candidate metrics before you test any price

Assume the value metric as the foundations lesson defines it. The work here is choosing between four to eight candidates and killing the weak ones fast. Score each against five tests:

• Does the bill rise as the customer's own outcome rises?
• Can the buyer forecast next year's spend within about 20 percent?
• Is the unit observable and auditable, so nobody disputes an invoice?
• Can it be gamed (shared logins, batched jobs, one seat driving twenty workflows)?
• What is the median annual growth of that unit inside an existing account? That number is your expansion slope, and it decides whether you grow without selling.

Nespresso's metric is the capsule, never the machine. Machines move at thin or negative margin through retail; capsules sell at roughly 0.40 to 0.55 euro for about five grams of coffee, which puts the effective price near 80 to 110 euro a kilo, several times premium ground coffee. Consumption tracks use exactly, so the metric scores well on tests one and five.

The edge case that most teams miss: a metric is only as strong as the barrier protecting it. When Nestlé's core capsule patents started expiring around 2012 and compatible capsules flooded in, the per-capsule economics leaked at exactly the point where the model made all its money. The answer was Vertuo in 2014, a system whose machine reads a barcode on the capsule rim, which re-established control of the billing unit. Choosing a metric you cannot defend is choosing a revenue line with an expiry date.

Adobe sits at the other extreme: a seat per month. Seats are predictable and auditable but correlate loosely with delivered value once one designer can produce ten times the output. Adobe's response from 2023 was to attach generative credits to the seat, a stable base with a metered edge. That hybrid shape is the usual landing point. Pure seats decouple your revenue from the value you create; pure consumption gives procurement no budget certainty and generates the shock invoice that ends the renewal conversation.

Step 2: instrument the willingness-to-pay research properly

Three instruments, chosen by the question you are answering.

Van Westendorp asks four price-perception questions (too cheap, a bargain, expensive, too expensive) and returns an acceptable range plus a rough optimum. Budget 200 or more respondents per segment. Its weakness is that it captures perception without purchase intent and is anchored on what people already pay, so it reproduces the market price you are trying to escape. Treat the output as guard rails.

Gabor-Granger walks a respondent down or up a price ladder and records intent at each point, which gives you a demand curve and a revenue-maximising point for a single package. Stated intent always overshoots. Standard forecasting practice is to count roughly half of the "definitely would buy" answers and a quarter of the "probably" ones before you believe the curve.

Choice-based conjoint is the one that answers packaging questions, because it prices bundles you have not built yet. Around 300 respondents, eight to twelve choice tasks, and you get part-worth utilities per feature plus a simulator. Run MaxDiff first if you need to rank twenty features into three tiers, then run conjoint on the resulting bundles.

Segment before you average. A blended optimum drawn from a bimodal sample produces a price that neither half of the market wants. And the cheapest study you will ever run is your own CRMCRMCustomer Relationship Management: software and strategy to manage and analyse customer interactions throughout their lifecycle.View full definition →: discount depth by deal size, win rate by price band, the price point where discount requests spike. If you are losing no deals on price, you are underpriced, and no survey is needed to tell you that.

Step 3: model the repackage with a migration matrix

Never model a repackage on averages. Build a matrix per segment: for each existing cohort, estimate the share that lands on the equivalent tier, upgrades, downgrades and leaves.

Take 10,000 accounts on a flat $100 per month, $12m in annual recurring revenueannual recurring revenueAnnual Recurring Revenue (ARR) is the normalized, predictable revenue a subscription business expects to earn from active contracts over a single year.View full definition →. Repackage into three tiers billed on a usage metric and assume 60 percent land at $79, 25 percent at $149, 10 percent at $299, and 5 percent churn. That is $5.69m plus $4.47m plus $3.59m, so $13.75m, up 14.5 percent.

Now stress it. If the metric under-reads real usage, more accounts sort downward: 70 percent at $79, 18 percent at $149, 12 percent churn gives $9.85m, an 18 percent decline. Same package, same prices, one wrong assumption about the distribution. That swing of roughly 33 points is why the migration matrix, not the price card, is the deliverable.

Two lines people forget: the cost of grandfathering (revenue you will not collect for as long as legacy plans exist, so put an end date on it) and the support and billing cost of running two schemas at once.

Step 4: model the trough on a licence-to-subscription move

A perpetual licence at $2,400 with 20 percent annual maintenance books $2,880 in year one from one buyer. The same buyer on a $50 monthly subscription books $600. Recognised revenue per customer drops roughly 75 percent on day one while your cost base does not move at all. The model has to answer three questions: how deep is the trough, how long is it, and what fills it.

Depth is the ratio above, applied to the share of new business you convert. Length is that gap divided by the growth in units the lower entry price buys you, plus the annual price escalation you can now apply to a recurring contract. Track annual recurring revenue, billings and the deferred revenuedeferred revenueCash a company has collected for goods or services it has not yet delivered. It sits on the balance sheet as a liability until earned.View full definition → balance through the transition, because GAAP revenue understates what is happening to cash and to the contracted base.

Second-order effects belong in the same sheet. Resellers who lived on upfront licence margin lose their economics overnight. Sales compensation built on one-off deal size stops working. Against that, once billing is recurring you can reprice the installed base every year, which is a lever perpetual licensing never gave you.

Real-world cases

Adobe announced in May 2013 that Creative Suite 6 would be the last perpetual release, having launched Creative Cloud the year before. Revenue fell from about $4.40bn in FY2012 to roughly $4.06bn in FY2013, with net income down by around two thirds, then $4.15bn in FY2014, before passing the old peak at about $4.80bn in FY2015 and reaching roughly $9bn by FY2018. The trough lasted two years and the model was worth roughly double the old one within five.

What filled it was unit volume bought with a lower entry price. A Creative Suite package ran from around $1,300 to $2,600 upfront; the subscription entry points were $19.99 a month for a single app and $9.99 for the photography plan, which reached students, freelancers and former pirates who were never going to write a four-figure cheque. Adobe reported subscriber counts and recurring revenue through the transition precisely because the GAAPGAAPThe standard set of accounting rules companies follow to prepare consistent, comparable financial statements, dominant in US reporting.View full definition → line looked worse than the business was.

Nespresso is the counter-case on durability rather than on transition. The capsule metric ran for two decades with almost no change to the model, because consumption scaled with value and the price per capsule stayed small enough to escape household scrutiny. The failure was not in the metric but in the moat around it, and rebuilding that moatmoatA lasting edge over competitors: a resource, capability or position they cannot easily replicate, letting a firm earn above-average returns over time.View full definition → cost a new machine platform and a split installed base that Nespresso still maintains.

CMO action items

• Score your candidate metrics on the five tests above and publish the scoring, so the choice survives the next reorganisation
• Run conjoint before Van Westendorp when the question is packaging, and size the sample per segment rather than in total
• Build the migration matrix with a base, a downside and a stress case, and get finance to sign the downside before you commit to a launch date
• Put an expiry date on grandfathered plans on the day you announce them

Common mistakes that kill results

Testing price before the metric is settled. If the unit changes, every price point you tested is answering a question you are no longer asking.

Modelling a single scenario. One migration assumption, taken from an optimistic sales view, is how a repackage that should add 14 percent ends up losing 18 percent.

Averaging across segmentssegmentsDividing a market into distinct groups of customers who share similar needs, characteristics or behaviours, so each group can be served with a tailored approach.View full definition →. Enterprise and self-serve buyers have different curves; a blended price sits in the valley between them.

Reading GAAP revenue during a subscription transition and panicking. Deferred revenue and the contracted base tell you whether the trough is behaving.

Leaving legacy plans open indefinitely. Every grandfathered cohort is a permanent tax on the new model and an argument your sales team will use to sell the old one.

Key takeaways

• The metric decides the revenue curve; price only decides the level
• A metric you cannot defend has an expiry date, as the capsule patent cliff showed
• Van Westendorp gives guard rails, conjoint gives packages, your own quote history gives the truth
• The migration matrix, built per segment with three scenarios, is the actual deliverable of a repackage
• A licence-to-subscription move costs roughly three quarters of year-one revenue per customer and is repaid by unit volume and annual repricing, over about two years in Adobe's case