Put the business model in beta along with the product
A beta that only tests features leaves the hardest question for launch day.
If you’re going to put your product in beta - put your business model in beta with it.
A company can come out of a year-long beta with forty thousand active users, a long list of fixed bugs and no evidence at all that anyone will pay. The product has been tested thoroughly. The business has not been tested, and the team now has to find out whether it exists at the moment when changing course is most expensive.
Kraus arrived at this from experience. He co-founded Excite in the 1990s and later JotSpot, a hosted wiki company that Google bought in 2006. His point, repeated in talks and already argued over on Hacker News in 2007, is that the word “beta” gives founders permission to leave the money question open. Users forgive rough edges during a beta, and they would forgive a price tag too. Almost nobody asks them to.
In practice the advice means charging, or at least asking for a commitment, while the product is still visibly unfinished. That can be a real price on a small tier, a paid plan that early users can opt into, or a pre-order page that records how many people click “buy” before telling them the feature is not ready. Each of these produces a number that a free beta never produces: the share of interested people who will part with money. That number shapes everything after it, from how much the company can spend to win a customer to whether it needs outside funding at all.
The advice holds best for software sold to businesses. Small companies that sign up for a project tool or a hosted wiki are used to paying for software, and their willingness to do so shows up quickly. A team that opens a paid tier in its sixth week learns which features customers value by watching which plan they choose. Products sold on a monthly fee from the first day, as 37signals has sold Basecamp, never had to guess whether their users would pay, because paying was how users arrived.
It holds less well for consumer products that depend on scale. A photo or messaging service that will earn money from advertising cannot test that model with a few thousand users, because advertisers do not buy audiences that small. Charging those early users would shrink the network that makes the product useful. Twitter and Facebook both spent years building an audience before settling on how to earn from it, and a revenue test forced into their betas would have measured the wrong thing. For companies like these the honest version of the advice is narrower: write the model down, decide which early signal would support it, such as how often people come back, and watch that signal from the start.
There is also a cost to testing a price too early. A product still missing its core feature may be turned down at any price, and a founder can read that refusal as a verdict on the business rather than on an unfinished tool. The fix is to treat early prices as experiments, change them often, and tell users that they will change.
The exceptions leave the main point standing. A beta is the cheapest time to be wrong, and the business model is the part founders are most often wrong about.