Answers vs Questions
Finding Product Market Fit
Minneapolis, 1949: a meeting that almost certainly never happened
Samuel Chester Gale set two boxes on the conference table.
Gale was General Mills’ vice president of advertising and one of the people credited with turning Betty Crocker into an American institution. In front of him were two very real products: Betty Crocker Party Cake Mix and Devil’s Food Cake Mix.
“Ladies and gentlemen,” he said, “I give you the future of cake.”
He poured the mix into a bowl.
“No measuring flour. No measuring sugar. No shortening. No wondering whether you used enough baking powder. Add water. Stir. Bake.”
He paused for effect.
“Cake. Almost instantly.”
There were approving nods around the table.
It was a very good answer.
Now imagine the next meeting.
The sales chart was not cooperating.
The sales executive spoke first.
“We need to emphasize speed. Put a clock in the advertisement. Cake in minutes!”
The operations person disagreed.
“Speed is only one part of it. Think through the steps. Gathering ingredients. Measuring. Mixing. Cleaning. We should market the cleanup. One bowl. One spoon. No flour all over the kitchen.”
The packaging executive tapped the box.
“The product is fine. The box is the problem. We need brighter colors. Maybe a birthday party.”
R&D had another theory.
“Flavors. That is the problem. Chocolate and spice are not enough. We need yellow cake. White cake. Maybe something with fruit.”
Finance cleared his throat.
“Or perhaps it costs too much.”
One by one, the team worked through every step surrounding the purchase and preparation of a cake.
Improve the advertising.
Add more flavors.
Reduce the price.
Get better shelf placement.
Make the instructions simpler.
Make cleanup faster.
Every suggestion was perfectly reasonable.
And every suggestion began with the same underlying assumption:
People want to make cake faster.
Finally, one of the home economists from the Betty Crocker test kitchens spoke.
“What if they don’t want faster?”
The room went quiet.
“What if they still want to feel like they are cooking, just being able to more reliably bake a great cake without having a lot of baking skill?”
That question sounds obvious now.
It was not obvious in a room full of people trying to improve an instant cake mix.
They had been asking:
How do we make cake making faster?
She was asking:
Which parts of making a cake feel like work—and which parts make it feel like baking?
That is a very different question.
The meeting and dialogue are fictional of course! Gale was real, and the Hennepin History Museum credits him as a driving force behind Betty Crocker. The products were real, too. General Mills’ history says home bakers testing its 1949 Party Cake and Devil’s Food mixes asked to add some of their own fresh ingredients, including eggs, even though the technology existed to have just-add-water cake cake mixes.
The answer in the room
The instant-cake idea was not foolish.
Baking from scratch involved a lot of steps. Some were tedious. Some introduced risk. Forget an ingredient, measure incorrectly or use the wrong pan and the result might be a small culinary tragedy in the middle of someone’s birthday.
A mix solved real problems.
But the team in our imaginary meeting had started with an answer:
Make cooking faster.
Once that answer was in place, every new idea became an attempt to improve it. Faster cleanup. Faster instructions. More varieties of fast cake.
The egg suggested something else.
A cake was not merely a delicious way to combine flour and sugar. It was often an expression of care. It was something you made for your family, your friends or someone celebrating a birthday.
If the box did everything, the result might be convenient—but it did not necessarily feel homemade.
Cracking an egg was technically an extra step.
Emotionally, it may have been the step that let someone say, “I baked this.”
The famous version of this story is a little too perfect. Requiring fresh eggs did not single-handedly create the cake-mix category. Frosting, decorating, new flavors, packaging, positioning and broader postwar changes all mattered.
But the messiness makes the lesson better, not worse.
The egg was not a magical answer.
It was evidence of a better question.
Finding product-market fit is hard
We all know product-market fit when we see it.
ChatGPT. Claude. Cursor. Ramp.
Once it happens, the product can feel inevitable. It’s customer led, not sales led. Word of mouth takes over. The question changes from “Will anyone buy this?” to “How fast can we keep up?”
The interesting part is what happens before that.
What do you do when customers think your product is interesting—but not essential? When pilots do not expand? When prospects compliment the demo and then return to whatever they were already doing?
One useful place to look is not only at your product.
Look at the questions you are asking.
Are you starting with a question?
Or are you starting with an answer?
Both are legitimate. The trouble begins when you confuse the two.
I started Polyverse with an answer
As many of my readers know, I started and ran a cybersecurity company called Polyverse for almost twelve years. We achieved a small level of success with high end sales to the military, but we never achieved a breakout success, and after twelve years the company ceased operations in 2024.
In hindsight, I started with an answer at Polyverse.
Companies care about cybersecurity.
Seems pretty indisputable, right? Every company said security was important. Cyberattacks were growing. The consequences were enormous.
So the rest appeared almost mechanical.
If companies care about cybersecurity, then better cybersecurity should be valuable. Build technology that makes attacks dramatically harder, and companies should buy it.
We did build exceptionally strong technology. It worked. Some of the most elite cyber units in government understood it, valued it and used it.
But broad commercial product-market fit remained elusive.
My answer contained a hidden assumption.
I had anthropomorphized my customer, the idea of a company.
“Companies care about cybersecurity” makes a company sound like a person: a single mind that recognizes danger, compares solutions and buys the best protection.
But a company is not a person.
It is a collection of people.
The CISO may care deeply about reducing risk. The application owner cares about uptime. The developer cares about whether a new security tool breaks the build. Procurement cares about vendor terms. Finance cares about the budget. An executive may care intensely about security right after an incident and much less six months later.
Everyone can agree that cybersecurity is important while no individual has enough urgency, authority and incentive to adopt your product.
If a company were one person, selling Polyverse might have been much easier.
A company cannot love your product.
A person can.
Who would be very disappointed?
That leads to a more useful first question:
Is there a specific group of people who love your product—and cannot imagine returning to life without it?
Not people who like the idea.
Not people who describe it as innovative.
Not people who agree to another meeting.
People who would fight to keep it.
Sean Ellis turned this into a wonderfully brutal product-market-fit test: ask active users, “How would you feel if you could no longer use this product?” His heuristic is that a strong signal appears when roughly 40 percent answer “very disappointed.” The question was later popularized through Superhuman’s approach to measuring product-market fit.
Forty percent is not a law of nature. Surveying the wrong users will produce a meaningless number. And in an enterprise product, buyers, users, administrators and executives may value entirely different things.
But the question is powerful because it measures loss rather than approval.
People are generous with approval.
Loss is harder to fake.
When a pilot ends, does anyone demand that it be restored? When the product is unavailable, does work stop? Does a champion train coworkers, push the product through security and procurement, and spend political capital to get the contract renewed?
That is what love looks like in B2B.
It does not require everyone at the company to adore you. Procurement does not need to send you a valentine.
But somewhere in the account, real people must care enough to create pull.
“Meh” is a powerful signal
If the response is “meh,” you have learned something.
If the response is “The other tools are good enough,” you have learned something even more important.
Praise during a demo is not pull.
A pilot that survives only because the founder keeps nudging everyone is not pull.
A feature wish list untethered to an urgent workflow is not pull.
Your next move may not be another feature.
It may be a better question.
Return to the imaginary General Mills meeting.
If you ask, “How can we make cake faster?” the answers are predictable:
Remove another ingredient.
Eliminate another utensil.
Shorten the baking time.
Improve cleanup.
Those may all be good ideas. But the answer has already been smuggled into the question.
What would you ask instead?
Perhaps:
Walk me through the last cake you made for someone you cared about. Which parts were annoying? Which parts worried you? Which parts made it feel like yours?
Or:
When you put the cake on the table, what did you want the people around it to believe?
Or, most simply:
Which steps do you wish would disappear—and which steps would you miss?
Those questions leave room for surprise.
They might reveal that measuring flour is a burden, while cracking an egg feels like participation. They might reveal that the baker wants convenience but still needs confidence, authorship and pride.
The same discipline applies to technology products:
Who is the person experiencing the problem?
What happened the last time they experienced it?
What do they use today, and why is it good enough?
What would break if your product disappeared tomorrow?
Who would fight to get it back?
Those questions do not ask customers to design the product for you.
They help you understand what the product must accomplish.
Answers are hypotheses
Starting with an answer is not a mistake.
Sometimes the answer comes first. You invent a technology. You discover a new capability. You see a future that does not exist yet and decide to build it.
That is conviction.
Just call it what it is and proceed accordingly.
If you begin with an answer, treat it as a hypothesis and figure out how to test it as fast as possible. Decide what customer behavior would prove it right. More importantly, decide what evidence would make you change your hypothesis.
If nobody loves the answer enough to miss it, resist the temptation to keep polishing it indefinitely.
Go back to the questions.
The imagined General Mills team could have spent years making instant cake even more instant. Instead, the market offered a clue: perhaps the goal was not to remove every step.
A cake has two recipes.
One produces the food.
The other produces the feeling that I made this for you.
Product-market fit often begins when you stop asking how to sell the answer you already have—and start asking which part of the customer’s life they cannot imagine losing.
Sometimes the breakthrough is not removing one more step.
It is discovering which step must remain.