What founders, investors and authors say works — summarised in plain words, linked to
the source, and tried on the businesses we run. When enough of them have tried it, we say whether it worked.
121 classics · all free0 being tested0 with a verdict5 businesses before a verdict
Jeff Bezos: Founded Amazon in 1994 and grew it from an online bookstore into a giant.
“Your margin is my opportunity.”
Where established sellers enjoy fat profit margins, a newcomer willing to run on thin margins can undercut them and win customers, so the incumbents' profitability becomes the newcomer's opening.
When it applies: Incumbents in your niche charge well above their costs and customers can easily compare prices.
Contested. Thin margins leave little room for error and can start price wars that a small business loses to better-funded rivals. Pricing below cost to push rivals out can be illegal predatory pricing in some jurisdictions, and low-price strategies by dominant firms draw antitrust scrutiny. Low prices also tend to attract the most price-sensitive, least loyal customers.
Joel Dean: Columbia University economist who pioneered 'managerial economics' in the 1950s.
“The active approach in probing possibilities for market expansion by early penetration pricing requires research, forecasting, and courage.”
A new product has to choose between starting high and cutting the price later (skimming) or starting low to win lots of buyers fast (penetration). Dean says a low launch price pays off when buyers react strongly to price, bigger volume brings costs down, and competitors are likely to copy you soon.
When it applies: Buyers in your market switch readily for a lower price and a rival could copy the product quickly.
Hubert Horan: Transport-industry veteran of airlines and urban transit whose essays dissected Uber's finances.
“Its multibillion dollar subsidies completely distorted marketplace price and service signals, leading to a massive misallocation of resources.”
The playbook: raise large sums from investors and charge customers less than the service really costs, taking share from incumbents who must cover their costs from fares, then rely on scale or weakened competition to raise prices later. Horan, a transport-industry analyst and critic of the model, documents Uber doing this and argues its cheap rides came from roughly $20 billion of investor money rather than from being more efficient; ride-hailing and food-delivery apps are the usual examples.
When it applies: A company can fund years of losses from outside capital and believes scale or market dominance will later let it charge more than cost.
Contested. Unit economics: if every sale loses money, growth multiplies the losses, and profit depends on later price rises or cost cuts that may never come (Horan reports Uber's GAAP profit margin was -135% in 2015). Customers won with subsidised prices may leave when prices rise. Legal: in the US a predatory-pricing claim must show prices below cost plus a reasonable prospect of recouping the losses later (Brooke Group v. Brown & Williamson, 1993); in the EU, prices below average variable cost by a dominant firm are presumed abusive and recoupment need not be proven (France Telecom v Commission, 2009). Ethical: Horan argues much of Uber's later margin came from cutting driver pay, not efficiency. Supporters point out Uber reported its first full-year operating profit in 2023; critics reply that it took over a decade and billions in losses.
Chris Anderson: Longtime editor of Wired magazine; wrote The Long Tail and Free.
Because serving one more digital user costs almost nothing, you can give a basic version free to a big audience and earn your money from the small share who pay for a premium version. The free tier does the marketing, and the few payers cover everyone else.
When it applies: Serving an extra free user costs close to nothing and the product spreads through use or word of mouth.
Patrick McKenzie: Software entrepreneur and writer, known online as patio11; spent six years at Stripe.
“I think a lot of uISVs, myself included when I was starting, undervalue their own software.”
Small software sellers usually price too low because they think about their own effort and the product's flaws instead of what the result is worth to the buyer. Price on the outcome you deliver; buyers who need it will pay, and the ones who fight hardest over price tend to be the costliest to serve.
When it applies: You sell software or a service that saves a business time or money and your price was set by gut feel or by copying cheap competitors.
Madhavan Ramanujam: Former partner at pricing consultancy Simon-Kucher; co-wrote the book Monetizing Innovation.
Georg Tacke: Led pricing consultancy Simon-Kucher from 2009 to 2019; co-wrote Monetizing Innovation.
Most new products miss their revenue goals because companies build first and settle the price last. Ask target buyers what they would pay, and for which features, before building, then shape the product, bundles and tiers around those answers.
When it applies: You are about to build a new product or major feature and can reach likely buyers before much is built.
Not tried yet
None of our businesses has tried it yet.
Value-based pricing: price on worth to the buyer, not on cost
Thomas T. Nagle: Pricing strategist; co-wrote The Strategy and Tactics of Pricing, a standard textbook.
Set prices from the economic value your product creates for a particular group of customers compared with their next-best alternative, and make that value visible to them, instead of adding a markup to cost or matching competitors. Groups that get different value can be offered different prices or packages.
When it applies: Your product saves or earns customers an amount you can estimate compared with what they use today.
Randal C. Picker: Longtime University of Chicago law professor who studied antitrust and technology.
“The actual history of razors-and-blades is much richer than the standard story suggests.”
The familiar playbook is to sell the base item cheaply, even at a loss, and profit on refills or add-ons. Picker shows Gillette, its supposed inventor, did not do this while its patents protected it (1904-1921); it cut handle prices only after the patents expired, to match rivals, and its customer base grew from wartime government orders and a cheap older model.
When it applies: You sell a durable item that needs repeat purchases of a consumable or add-on, and customers find it easiest to buy those from you.
Contested. Selling the base at a loss pays only if customers keep buying the high-margin refills from you; without some lock-in they buy refills elsewhere. Forcing refills to come from you (tying, blocking third-party refills) can raise antitrust and consumer-protection problems and anger customers. The famous origin story is historically wrong, so do not rely on it as evidence.
Not tried yet
None of our businesses has tried it yet.
Usage-based pricing: charge for how much customers use
Kyle Poyar: Pricing expert and former OpenView partner; writes the Growth Unhinged newsletter.
“The real enemy of usage-based pricing isn't subscriptions.”
Charging at least partly by consumption (tasks, API calls, credits, resolutions) makes it cheap to start and lets revenue grow as customers get more value, which suits automation and AI products where the number of seats no longer tracks value. Poyar, a long-time advocate, reports that pure pay-as-you-go has cooled and most usage-based companies now mix a subscription base with usage charges.
When it applies: The value customers get rises with a countable unit (tasks run, messages sent, records processed) rather than with the number of people using the product.
Not tried yet
None of our businesses has tried it yet.
Offer one standard discount for paying a year upfront
Jason Lemkin: Founded SaaStr, the large community for software founders; sold EchoSign to Adobe.
“Then, give a 20% discount for annual prepayment.”
Rather than letting discounts be haggled deal by deal, set list prices somewhat higher and offer one fixed discount, around 20%, for paying a year in advance. The business gets cash now and a year's commitment; the customer gets a clear, fair reason for the lower price.
When it applies: You sell a subscription, cash for growth is tight, and customers trust the product enough to commit for a year.
Not tried yet
None of our businesses has tried it yet.
Free trial vs freemium: trials convert more, freemium signs up more
Kyle Poyar: Pricing expert and former OpenView partner; writes the Growth Unhinged newsletter.
Lenny Rachitsky: Former Airbnb product lead; writes Lenny's Newsletter, one of Substack's largest.
“On average, 3%-5% is a GOOD conversion rate for a freemium self-serve product, and 6%-8% is GREAT.”
In a survey of more than 1,000 mostly business software products, a time-limited trial of the full product turned a much larger share of sign-ups into payers within six months (good: 8-12%) than a permanent free plan (good: 3-5% for self-serve). Free plans attract more sign-ups in the first place, so the choice is between reach and buyers who are closer to paying.
When it applies: You sell self-serve software and must choose between a permanent free plan and a time-limited trial.
Hema Yoganarasimhan: University of Washington marketing professor who studies online businesses using large-scale data and experiments.
“We find that, on average, shorter trial lengths (surprisingly) maximize customer acquisition, retention, and profitability.”
In a large randomised field experiment at a leading software-as-a-service company, new users given a 7-day free trial subscribed at higher rates than those given 14 or 30 days, and the short trial also did well on retention and profit. Long trials let interest fade (long idle spells near the end went with fewer conversions), though experienced users gained more from longer trials than beginners.
When it applies: You offer a free trial of a subscription product and new users can see its value within a few days.
Sid Sijbrandij: Co-founded GitLab and took it public in 2021.
“Features that appeal most to an individual contributor are open source and free.”
Release the core product as open source to win users and contributors, and sell proprietary features on top. Decide what is paid by who wants a feature most: things hands-on users care about stay free, things managers and executives care about (approvals, compliance, security, reporting) go into paid tiers, as GitLab does.
When it applies: Individuals inside organisations use your product hands-on, their managers hold budgets, and a free open-source version can spread on its own.
Contested. The Open Source Initiative and many developers say open core is proprietary software wearing an open-source label. Moving features from free to paid destroys community trust (Sijbrandij himself warns against it). Several open-core companies (MongoDB, Elastic, HashiCorp, Redis) later moved to non-open licences, prompting community forks. Large cloud providers can host your open-source core and capture the revenue.
Can't test here
It needs an open-source project with an active outside community of users and contributors, which takes years to build.
Jason Cohen: Founded WP Engine, a large WordPress hosting company, and Smart Bear Software.
“It's not going to be as big a deal as you think, so long as you're honest and generous.”
Most young companies set their first price too low and should raise it as the product improves. Tell existing customers plainly, in a personal note from the founder, why the increase is needed, let anyone who objects keep the old price, and expect almost everyone to stay.
When it applies: Your product has improved since launch, you charge less than comparable offers, and thin revenue is holding back support or development.
Ben Thompson: Writes Stratechery, a widely read newsletter on the strategy of tech companies.
“it is impossible for a user experience to be too good.”
Consumers, unlike business buyers, pay for things that never show up on a spec sheet: design, ease of use, polish and brand. A seller who stands out on experience can keep charging well above cheaper 'good enough' rivals for years, as Apple has with the iPhone, instead of being undercut the way disruption theory predicts.
When it applies: You sell to consumers who can see and feel a difference in experience, and you can deliver that difference every time.
Neil Davidson: Co-founded Red Gate Software, a British maker of database tools.
“What matters now is not how much you've spent, but what people are prepared to pay.”
Your price should come from what customers believe the product is worth, not from what it cost to build, because development money is already spent. Since buyers value it differently, offer versions (by features, speed of access or type of customer) so each group pays closer to its own limit, and remember the product includes support, documentation and reassurance, not just the code.
When it applies: You sell software or a digital product to buyers who differ a lot in how much it is worth to them.
Not tried yet
None of our businesses has tried it yet.
A rake too far: keep the marketplace take rate modest
Bill Gurley: Benchmark venture capitalist who backed Uber, Zillow and OpenTable early.
“High rakes are a form of friction precisely because your rake becomes part of the landed price for the consumer.”
A marketplace's cut of each sale shows up in the price buyers pay and the margin sellers keep, so a high cut pushes both sides toward alternatives and invites a cheaper rival. A small cut on a lot of volume tends to last longer than a big cut on less.
When it applies: Setting, or thinking about raising, the commission a marketplace takes from each transaction.
Not tried yet
None of our businesses has tried it yet.