Narrow Until the Numbers Work

The Core Insight

Aulet's first commandment is that the single necessary and sufficient condition for a business is a paying customer. The list of things that do not qualify runs long: idea, product, technology, team, plan, money, strategy.

Disciplined Entrepreneurship builds a venture backward from that customer through twenty-four steps, in order. Bill Aulet teaches the sequence at MIT Sloan. The process is sequential and not linear.

Most people treat the founder and the idea as the variables that decide the outcome. Aulet argues that entrepreneurs are made, and that the original idea is the single most overrated thing in entrepreneurship. Pierre Azoulay's research puts the average founding age of the 1,000 fastest-growing ventures at 45.

Creating value for someone is only the first criterion. You must extract rent for that value, and the rent must clear the cost of running the business. The payer is the economic buyer, who is often someone other than the user. Google creates value for searchers and extracts rent from advertisers.

The Framework

The twenty-four steps answer six questions, and the numbering hides the grouping.

• Who the customer is takes steps 1 through 5 and step 9.
• What you can do for that customer takes steps 6 through 8, 10, and 11.
• How the customer acquires the product takes steps 12, 13, and 18.
• How you make money takes steps 15 through 17 and step 19.
• How you design and build takes steps 20 through 23.
• How you scale takes steps 14 and 24.

Two gates sit inside the sequence and decide whether the venture is real. The lifetime value of a customer must reach at least three times the cost of acquiring one. For a US beachhead, Aulet names 20 million to 100 million dollars a year as the target range. Anything above a billion dollars a year needs further segmentation.

Nothing gets built until step 22, because the IKEA effect makes builders fall in love and stop listening. Steps 7 through 21 run on a drawing, a brochure, a landing page, and conversations.

Key Ideas

A Market Has Three Conditions

A market segment means customers who buy the same product, buy it the same way, and talk to each other. Same way covers use case, value proposition, channel, pricing, decision-making unit, and sales process. The working test is that a sales rep switches from one customer to the next with no loss of productivity.

Two failure modes eat this step. The first is selling to everyone. The second is the abstract paying customer, which Aulet calls fun with spreadsheets. The example runs on 1.5 billion Chinese buyers, a 0.1 percent share, three brushes a year, and one dollar each. That produces 4,500,000 dollars of first-year sales and no customer anyone can name.

Eight criteria rank the survivors. The first four ask whether the customer is well funded, reachable, carrying a compelling reason to buy, and servable today. The last four ask about entrenched competition, expansion into adjacent segments, fit with the founding team, and speed to win. Then you deselect everything else. SensAble narrowed to eight industries over three months, then to industrial design, then to the toy industry, and refused international orders.

The TAM Is Annual Revenue at Full Share

The total addressable market is the annual revenue you earn at 100 percent market share, stated in dollars per year. Count end users top-down from filtered research and bottom-up by counting noses. Estimate annual revenue per end user three ways. Take what the customer spends today divided by product lifetime, so one car at 15,000 dollars over five years is 3,000 dollars a year. Then take the budget available for the problem, then comparables in nearby markets.

OnDemandKorea is the worked example. The census counted 1.7 million Koreans in the US, and further research put the working figure at 2.5 million. The team found 89 websites illegally streaming Korean dramas and measured 1.2 million unique users across them. Testing gave a 60:40 female-to-male split, which cut the number to 720,000. About 55 percent were aged 20 to 35, leaving 400,000 end users who fit the profile.

Advertising at 1.25 dollars per user per month is 15 dollars a year, so the beachhead TAM came to 6,000,000 dollars. Every other revenue line was excluded to stay conservative.

The Persona Is a Real Named Person

Step 3 builds a composite end user profile with six parts. Those parts are demographics, psychographics, proxy products, watering holes, a day in the life, and ranked fears. The ranking is forced, because end users allocate 100 points across their concerns until a weighted top five appears. SensAble's profile read as an industrial designer in toy companies, 90 percent male, aged 24 to 35, earning 50,000 to 60,000 dollars.

Step 5 replaces the composite with one real person, named, with a photo and a fact sheet. Questions about a named person have answers instead of opinions. Ed Champ ran sculpting for boys' toys at Hasbro, aged 40, earning 73,500 dollars after 14 years there. His purchasing criteria ran in order: time to market, expressing design intent, and keeping that intent intact once engineers took over. Everything downstream keys off that order, including the product spec, the value proposition, the competitive chart, and the price.

The water filtration team guessed the order wrong. Chuck Karroll manages facilities at an IBM data center with just over 20,000 blade servers growing 15 percent a quarter. His criteria run reliability, growth, cost, and greenness last. The team had built its pitch on environmental benefit, and research showed Chuck cared very little. Downtime was the thing that rang his pager. If a better representative appears, 5 percent or more better, the Persona changes.

The Buyer Is a Committee With a Clock

The decision-making unit has three primary roles and three more that block. The champion advocates from inside, the end user creates the value, and the primary economic buyer holds the budget. Influencers carry a de facto veto, and corporate IT standards or zoning rules are real veto holders. Purchasing gets neutralized rather than sold.

The water filtration team mapped its unit and found a hole. The end user was the facilities manager, the economic buyer was the data center manager, and both were risk-averse. Neither pushed. They found the champion in a newly installed VP of Sustainability who reported to the CEO and prepared the ESG report. The CIO never drove the decision and vetoed anything risky.

Step 13 puts a clock on the same map. Their new-construction cycle ran lead generation one to two months, then influencers and design engineers at two to four each. Design took six to twelve months, construction and the sale to the contractor twelve to fifteen, installation one. The average came to 2.5 years or more, and the retrofit path came to about a year. They redirected to retrofits despite fewer inbound inquiries.

Budget authority is the lever inside the map. A common individual limit is 5,000 dollars without senior approval, so pricing under it keeps the decision with one person. Operating budget against capital budget is the difference between a three-month sale and a one-year sale.

Lifetime Value Must Beat Acquisition Cost Three to One

Lifetime value is the net present value of profit from an average new customer, year zero through year five. The window is five years, because a startup's compounded cost of capital discounts anything past that to nothing. Eight inputs feed it: one-time revenue, recurring revenue, upsell, gross margin per stream, retention, product life, repurchase rate, and cost of capital. For a founder with no track record, cost of capital runs 35 to 75 percent a year. Aulet says to start at 50 percent, where a dollar in year three is worth 30 cents today.

The printed example sells a widget at 10,000 dollars with maintenance at 15 percent of list after a six-month warranty. Gross margins are 65 percent on the widget and 85 percent on maintenance, and retention is 90 percent after year one. Product life is five years, and 75 percent of surviving customers buy the replacement. The arithmetic returns 9,479 dollars, and Aulet immediately restates it as between 9,000 and 10,000 dollars. The restatement is the lesson, because the four-digit answer implies precision the inputs do not carry.

Acquisition cost gets computed the wrong way. The bottom-up version takes a salesperson at 150,000 dollars, a half-year cycle, and one twentieth of their time per customer. That gives 75,000 dollars per cycle, 3,750 dollars per sale, and 6,250 dollars once benefits, travel, demos, and trade shows go in. The number assumes a 100 percent close rate. At an aggressive 25 percent close rate the rep sells five per cycle, so three twentieths of the time goes to non-buyers. Aulet puts the realistic figure at ten to twenty times the bottom-up answer.

The right method is top-down. Add total marketing and sales expense, subtract the cost of supporting the install base, and divide by new customers. Run it over three periods: year one, years two and three, years four and five.

Build the Smallest Thing Someone Pays For

The minimum viable business product has three conditions and needs all three. The customer gets value from using it, the customer pays for it, and the feedback loop starts. Paying is the condition that separates it from a lean experiment testing one assumption.

Minimal is the hard word, because removing features fights the endowment effect. Concierge everything you can, and build only what has to respond in real time. Wealthfront put human advisors in front of clients before automating the advice people paid for.

ThriveHive sold a marketing platform from 99 dollars a month and built only the real-time integrations. Its postcard builder took an upload, showed the cost, and emailed the team, who did the work by hand. Of their beta testers, 74 percent converted to a paid subscription. Define the adoption metrics before release, with success criteria written down.

Practical Applications

Build the segmentation matrix over four to six candidate segments, filled from direct customer contact. SensAble drew roughly 90 percent of its matrix data from talking to people. Give it a few weeks, then deselect down to one beachhead.

Name the Persona before the next build decision, and use their purchasing order rather than yours. Then quantify the value proposition as as-is against possible, in the units of their top priority. SensAble measured a 50 percent reduction in time to market and refused to convert it to dollars.

Assemble the next ten customers before writing code. Same product, same sales process, word of mouth between them. The methane capture team ran ten landfill sites, got strong responses from eight, and letters of intent from more than half. If ten homogeneous excited customers cannot be found, fold the beachhead.

Price on value instead of cost, starting near 20 percent of the value you create. Set the price against the buyer's authority, the way Kinova priced its assistive arm at exactly 28,000 euros. That was the ceiling of Dutch health-insurance reimbursement, and it collapsed the sales cycle.

Match the sales motion to the lifetime value before hiring anyone. Field sales needs a lifetime value near 50,000 dollars or higher, and channel resellers take 15 to 45 percent.

Who This Is For

First-time founders holding a technology and no market get the most from this. The steps also work as a diagnostic on a company that is already selling and stalling.

Founders selling into a market they came from can skip ahead and start at the value proposition. The book targets innovation-driven enterprises aiming at global markets, so a local services business gets the wrong prescription.

This is a classroom syllabus before it is a field manual. The steps come out of a semester at MIT, and the running Bloom example is a student project Aulet critiques rather than endorses. Running all twenty-four steps before selling anything is a real failure mode, and the sequence invites it. The arithmetic also assumes a level of certainty a new company rarely has. Retention, close rate, and cost of capital are guesses at step 17, and the five-year figure inherits all three. Aulet knows this, which is why he converts every result to a range.

The Decision

Two numbers settle whether the current plan survives contact. Compute the lifetime value of your average customer over five years, discounted at 50 percent, on profit rather than revenue. Then compute acquisition cost top-down: everything spent on marketing and sales last quarter, divided by the customers you added. Report both as ranges with the assumptions written next to them.

A ratio under three to one means the business model or the beachhead is wrong, and more spending makes it worse. A ratio above it means the machine works and the next question is volume.

Everything in the twenty-four steps exists to move those two numbers.