A business model is the arrangement by which a product turns its promise into revenue. It covers the questions of who pays, what they are paying for and how often they pay. Alexander Osterwalder and Yves Pigneur set it out as nine building blocks in Business Model Generation in 2010. The value proposition sits at the centre of that canvas, the market sits to the right of it, the resources and partners serving that market sit to the left and the money runs along the bottom. A promise made to a chosen segment says what the customer receives. It does not say how the organisation gets paid for delivering it, and the business model is the answer to that second question.
The sections below set out six ways a product is paid for and what each one demands of the product. They then show how a model narrows the strategy above it. The page ends with what a change of model costs once customers are already living on the old one.
The six ways a product is paid for
Six arrangements cover most software products, and one product often runs two of them at the same time.
| Model | What the customer pays for | What it requires to work |
|---|---|---|
| Subscription | Access for a period, usually a month or a year | Value that recurs, since a customer whose job is finished cancels |
| Transaction | A share of something the customer completes, such as a payment or a booking | Volume the product can see and settle, and a share small enough to leave the customer better off |
| Usage | Units consumed, such as calls made, records processed or storage held | A meter the customer trusts, and a bill they can forecast before the month ends |
| Marketplace | A fee on a trade between two other parties | Enough supply and demand present at once for a trade to clear quickly |
| Advertising | Attention, sold to somebody other than the user | Reach large enough to interest a buyer of attention, and data about who is watching |
| Licence | The right to use one version, bought once | A next version the customer will pay to have, and a support obligation that ends |
Advertising is the row teams misread most often. It is the only one of the six where the person using the product and the person paying for it are two different people. The team then has to judge the product on behalf of somebody who never appears in a user interview.
What each model demands of the product
Each row of that table is a standing obligation. Three of those obligations catch products out often enough to be worth naming here.
Subscription demands that the value keep coming back. If a product does a job once, it collects a month of revenue and then a cancellation, so the promise has to describe work that keeps arriving. Payroll arrives every month. A data migration happens once.
Usage demands a meter the customer believes and a bill the customer can predict before it lands. If the meter counts something the customer cannot reproduce from their own records, the invoice turns into a support ticket every month. A large bill nobody forecast is one of the most common reasons a usage priced account leaves after a good quarter. The product worked, and the model punished the customer for it.
Marketplace demands liquidity, which means enough buyers and sellers in the same place at the same time for a trade to clear quickly. Below that point the product can work perfectly and earn nothing. If a seller lists twice and waits both times, that seller stops listing.
How a business model narrows the strategy
The model settles what every feature is judged on. Two teams can hold the same promise and still build different products, because they are working under different models. Under subscription the question asked of a feature is whether it holds a renewal. Under advertising the question is whether it holds attention, so a feature the user would happily pay for survives only if an advertiser benefits too. That constraint comes from the money side of the canvas, and no amount of customer research removes it.
Clayton Christensen described the hardest version of this in The Innovator's Dilemma in 1997. An established cost structure makes the cheap end of a market look unattractive, so the incumbent turns that end down for reasons its own model makes sound. The decision is rational inside the model and wrong outside it, and that is why better analysis rarely helps.
What changing a business model costs
If a model has stopped fitting, changing it is possible and expensive. Adobe announced in May 2013 that it would stop selling perpetual Creative Suite licences and offer its creative applications only through Creative Cloud subscriptions. The product barely changed. The company changed a great deal, and the cost of a move like that falls in three places.
- Existing customers sit on terms somebody has to honour, buy out or break.
- The sales compensation plan pays people on a number that no longer exists.
- Reported revenue drops for as long as the transition runs, because money recognised once now arrives across twelve months.
None of the three is a product problem. All three still reach the product team, and they arrive as pressure to soften the change. A model changed halfway carries both sets of costs and neither set of benefits.
Choosing the model settles the shape of the money and leaves the amount open. Inside any of the six rows, two decisions are still in the organisation's hands, and the first is what the customer is charged while the second is the unit that charge is measured against. Most teams leave both of them until the week before launch.