A committee can approve months of work for an offer no customer has asked to buy. The expensive decision is the one that treats a tidy business case as evidence.
Someone senior wants a new offer. The paper carries a market figure, a launch month and a squad, and the people who would have to change their week have not been asked. The money moves when the internal objections run out. The buyer, the user or the operator who already has a way to do the job can still save that budget, if anyone asks them.
A business case is a pile of guesses
Steve Blank set the alternative out in Harvard Business Review in 2013. He cites Shikhar Ghosh at Harvard Business School, who found that 75 percent of startups fail. The path Blank describes is the one a steering group still funds. Write the plan, raise the money, hire the team, launch the product, and sell hard. Somewhere in that sequence the venture hits a wall. Inside a large company the same sequence has a bigger budget. The plan is a business case, the investors are a committee, and the wall arrives after the squad has been hired.
A plan written at a desk, with a five-year forecast attached, is a pile of guesses. Blank's definition of a startup is a temporary organisation in search of a repeatable business model, something a customer will pay for and the company can keep doing. The parent company may be excellent at running the model it already has. The squad with the new offer is still searching. Fund that squad the way you fund a known product line, and you will pay for a factory before you know who the customer is.
He describes three moves. Write the guesses on one page: how you create value for a customer, and how the company captures some of it. Get out of the building, and talk to users, buyers and partners about the offer, the price, the channel and how you would reach them. Put a minimum version in their hands and change the guess from what they actually do. A small change is an iteration. A new customer, problem or model is a pivot. Build in short cycles, with only the features the current guess needs, because a year of development assumes you already know the problem. Blank will not promise a win on any single venture. Across a portfolio, teams that test this way fail less often than teams that build in private and meet the customer at the launch.
Blue River dropped the original customer
The clearest picture in the essay is Blue River Technology. Jorge Heraud and Lee Redden were students in Blank's class at Stanford. They wanted to build robotic lawn mowers for commercial sites. They talked to more than 100 customers in 10 weeks. Golf courses, the customer in the original idea, saw little value in the mower. Farmers did want something else: a way to kill weeds without chemicals. That became the product. Within 10 weeks of the shift they had built and tested a prototype. Nine months later they had raised more than $3 million. They expected a commercial product nine months after the raise.
They spent the first stretch on conversations, and they threw out the customer in the pitch. Those 10 weeks killed the golf-course mower and left a different product in its place. A funding meeting that can only say yes or no to the original slide will push the team to defend the golf course.
GE talked to buyers before the factory
Blank writes that GE, Qualcomm and Intuit had started using these methods inside the company. The case he tells in full is GE Energy Storage. In 2010 Prescott Logan, the general manager, had a new battery, later called Durathon, that looked as if it could shake up the industry. The default was a factory, more production, and a launch as a line extension. Logan went looking for a business model first. His team met dozens of prospects, left the slides behind, and asked how they bought industrial batteries, how often they used them, and the conditions they ran them in. Data centres came off the list. Telecom shrank to phone providers in developing countries with unreliable grids. GE then put $100 million into a factory in Schenectady, New York, and opened it in 2012. Blank says press reports at the time described a backlog of orders.
Read that in the order it happened. The technology already existed. The $100 million factory came after the team had dropped one segment and narrowed another, on the back of conversations with the people who buy. A committee that funds the factory so the team can go and discover who wants the battery has Logan's story backwards. The listening was the cheap part. The factory was the part you pay for once you know who has the problem.
Run this before you fund the build
Start with one sentence a stranger could recognise: who they are, what they do today, and what you want them to do next. If that sentence only makes sense inside your building, a squad is premature. Put the guesses on one page, covering the problem, the offer, the price, the channel and how you will reach people. Circle the guess that kills the idea if it is wrong, and test that one. Blue River's dangerous guess was the customer. Logan's was which segment actually had the problem. Yours might be whether a buyer will switch a contract, or whether an operator will leave the spreadsheet they already trust.
Go and ask, this month. Sit with people who can refuse you: buyers, users, and the colleague who would have to carry the offer into a real week. Leave the slides, as Logan's team did. Ask how they buy the thing today, how often they use it, and what the conditions are when it has to work. A first round of conversations will not match Blue River's hundred. It will still teach you more than another draft of the business case. Give them something thin to react to in the same month. Blank's minimum version carries only the features the current guess needs. A service you run by hand, a one-page offer with a price and a way to request it, or the next live case done the new way. He also notes that open source tools such as GitHub, and cloud services such as Amazon Web Services, cut the cost of building software from millions of dollars to thousands. A six-month squad is an expensive way to buy a fact when a thin version can be stood up with the people you already have.
Agree three results with the person who can stop the money, and write them down before the test starts. The result that ends the idea. The result that pays for a second, slightly larger test. The result that would justify the build already described in the case. Write them before anyone has spent a month on the deck, while a stop is still a small decision. I want that person to hear a customer, or an operator, say no. A result they never sat through will lose to the original slide at the next funding meeting.
The sponsor has to hear the no
If the sponsor can only open the session and leave, move the date. A room full of people who already like the idea will design a test that flatters it. The signer has to hear a stranger shrug, and has to be willing to stop the work when the result misses the line you agreed. That stop is the point of running a portfolio of ideas. Fewer of them should reach a full build.
Regulated teams can do this without putting a customer or a licence at risk. Run the service by hand for a defined slice of work. Talk to a small group who fit the sentence. Put a one-page offer in front of them with a real ask. You want a response the team did not write. Interest, an early request, or a clear refusal counts. A compliment from the people building the offer only tells you what they think of their own work.
Bring the offer you are about to fund
If a thin test has already found users, and the work is now sitting with nobody accountable for scaling it or stopping it, that later stall is the subject of pilot purgatory is where good ideas go to die. The failure here is earlier. The organisation funds the build because the internal story is tidy, and it never buys evidence that someone outside that story wants the thing. What the day should put under test, once the offer is live, is in what a lean startup workshop for corporate teams should test. How to write the result that stops the spend is in test the bet before you fund it.
Collective Campus runs that day as a lean startup and product sprint. It is one day, or a short run of days, on a single live offer. The sponsor who can fund the work or stop it stays until the call is made. You separate what the room knows from what it hopes, design the smallest test the team can run with the people and money it already has, and leave with a decision to stop, change the offer, or pay for the next test.
Send the offer in one sentence, the name of the person who can refuse the money, and the guess you have not tested. We will shape the sprint around that guess, and put it in front of a customer before a build team is hired.
Four questions before you book
- Which offer is about to receive real money, and who outside the company asked for it?
- Which guess, if it is wrong, makes the rest of the spend a waste?
- What can a buyer or a user react to in the next month, short of a finished build?
- What result would make the sponsor stop the work, and will they be in the room to hear it?
