The Expensive Bet That People Will Want the Product
In 2020, a student of mine was building a marketplace for yoga teachers. Clients would find a teacher there, and teachers would get paid through the platform — in its own blockchain tokens.
He was in love with the idea. He had no doubt that clients would come to his marketplace to look for teachers, and no doubt that teachers would happily take their pay in tokens. I tried to talk him out of it. The plan stacked several risky assumptions on top of each other, and the biggest one — that anyone wanted this at all — he refused to test. Why test what is obviously needed?
The startup closed.
I have seen hundreds of startups die on that same bet. I have seen corporate products die on it too — launched with budgets, stakeholders, rollout plans, and a head of product who could defend every screen. The founders and managers were in love with what they were building and certain the world needed it. The fundamentals went unchecked: whether the segment exists, which Jobs it performs, whether the product creates value for those Jobs, and whether anyone communicated that value. The whole bet was one assumption: once people see the product, they will want it.
They believed the product was the center of the customer's attention. Build it well enough, show it clearly enough, explain the features carefully enough, and the customer would buy, activate, return.
The customer has to get paid, get home, submit the report, look competent, close the deal, survive the audit, wake up rested. Our product appears only if it helps one of those Jobs happen better.
The Product Is a Necessary Evil
The canon sentence is brutal: the customer doesn't need your product — they need a Job performed.
A product is a means. The end is the Job landing: outcome reached, criteria met, higher-level Job made possible.
Nobody needs a bank — people pay it to sleep peacefully, knowing the money will not disappear. Airbnb gets hired the same way: the family is buying a great weekend together, not an apartment-booking flow.
Marketing has known a cousin of this thesis for decades. Theodore Levitt taught it as the quarter-inch hole: people do not want a quarter-inch drill, they want a quarter-inch hole (the line is even older than Levitt). AJTBD extends the hole into a full picture: the situation the person is in, the criteria for good enough, the higher-level goal the hole serves, and the comparison against every other way of getting it.
Even a product the customer loves is tolerated because, right now, it is the most acceptable route to a Job. You are not walking around with an iPhone 6. Love does not stop a better route from appearing.
This is where product-centered thinking burns money. The team benchmarks features, reads competitor release notes, and treats the category as the competitive set. The mistake is letting the product become the unit of analysis.
The customer is trying to perform a Job, with success criteria, in a specific situation, in order to perform a higher-level Job. If a spreadsheet, a person, a habit, doing nothing, or a product from another category performs that Job more energy-efficiently, the customer can live without us. An indie product rarely loses to a rival app. It loses to Google Sheets and to leaving things as they are. Zoom can compete with airfare when the Job is "I want to meet with the people I need to meet." A mattress can compete with coffee when the Job is "I want to function tomorrow despite bad sleep."
The practical move is attention discipline. Take attention off the interface, the category, and the founder's favorite feature. Put it on the person's Job: the situation they are in, the outcome they want, what counts as good enough, and where the current way costs too much effort. Both selling and measuring come down to the same question: did the Job get done better than the customer expected?
Wes Tries the Genius-Founder Escape Hatch
Uber and ChatGPT make the same point at scale. Uber did not win because people had been begging for a ride-hailing product. It deleted whole pieces of the old way of getting around: own the car, find parking, call a dispatcher, wait without certainty, handle the payment. ChatGPT did not win because people had been asking Word for a chatbot. It deleted the blank page: hold the structure in your head, start from nothing, search for phrasing, rewrite the first draft.
No customer could have described either product in advance. Every one of them could describe the old route and where it wasted energy — and that is exactly what useful research collects. Useful research never asks people to design the product — it reconstructs the Jobs, the criteria, Jobs Graphs, and the current ways the new product must outperform.
Jobs and Ford were never arguing against listening to customers — only against asking customers to invent the product for you. Study the Jobs they already perform, then invent a more energy-efficient way to perform them. Later in the book, we will walk that algorithm step by step and build products people could not request in product language.
What About "Satisfy Customer Needs"?
If the unit of analysis is the Job and not the product, one mantra is still standing in every deck — satisfy customer needs — and the next chapter takes it apart.