Kodak Built the First Digital Camera in 1975. Management Said Not to Tell Anyone.
In December 1975, a Kodak engineer named Steven Sasson took the first photo with a digital camera he’d built. It was about the size of a toaster. He showed it to management. Years later, he told the New York Times what they said: “that’s cute but don’t tell anyone about it.”
Kodak got a patent on the camera in 1978. The prototype never went into production. Kodak did sell digital cameras later on. But the company that owned photography let the first one sit.
It’s easy to blame Kodak’s leaders for that. I think the lesson is bigger than one bad meeting. Kodak’s business was built around film. Its customers bought film. Its stores sold film. If you’d asked those customers what they wanted, I think they’d have asked for better film.
That’s my thesis, and it’s the reason for this whole piece. Customers can tell you where today’s setup hurts. They can’t tell you what should replace it. If you start your innovation work by asking them, you’ll build a better version of what already exists. You’ll polish the old setup when you should be removing parts of it.
What customers can see, and what they can’t
Start with what happens when you ask people what they want. They describe their workarounds.
Someone who copies data between three screens all day will ask for a faster way to copy. Someone who builds a report in Excel every week will ask for a better export button. Those are real pains, and I’m not dismissing them. But each request is proof the process is broken. They aren’t asking you to remove the step. They’re asking you to make the step hurt less.
So a pile of requests is really a record of how people cope. It shows where it hurts today. It doesn’t show what the work should look like.
Think about what a customer can see from where they sit. They can see the wait. They can see the clicks, the forms and the frustration. They can’t see the labor hours behind the screen. They can’t see the equipment someone owns and keeps running. They can’t see the handoffs between teams, or the cut each middleman takes. That’s where the cost lives.
The Passenger Fallacy
I call this the Passenger Fallacy. A plane passenger can tell you the seat is cramped, the food is cold and boarding took too long. All true. But that passenger can’t design a better engine. They can’t tell you how to cut fuel costs in half. They only see the cabin.
Customer surveys fall into this trap, however many metrics they carry. They ask today’s passengers about today’s ride. So they capture how people cope with the current setup. That’s the wrong lens for disruptive innovation, because customers can’t see a setup that doesn’t exist yet. Film customers were never going to ask for a camera with no film.
To be clear, surveys are great for customer experience work. They show where today’s process rubs. But fixing that only polishes a surface built on waste and friction. You get cleaner buttons and a nicer rating screen. Every copy and paste step is still there. The cost structure underneath stays the same.
Where I first saw this go wrong
This view started with a client study.
In that study, the consultant on the project tried to put themselves into the equation. He made the car the one doing the job. The work was framed around the needs of a car, trying to transfer power from the axle to the road.
We ended up with a survey that no customer could answer. Not even one single person could answer it. Some questions needed one kind of engineer. Others needed a different kind. No one person had done the whole job, so no one person could speak to it.
That breaks a core rule of the job research I do. A survey must be answered by one person who does the job, or who did it recently. Once you need a room full of specialists to answer it, you’re no longer studying a job someone does. You’re studying a machine.
That was my first clue that the survey methods I’d been using weren’t going to work for functional innovation on their own. Something had to be added. A car doesn’t have needs. Physics does set limits, though. Power has to get from the axle to the road, and there’s a real floor on what that takes.
So I went looking for a first principles approach. I started asking what a result should cost if nothing were wasted, and how far today’s setup sits from that. That’s what I now call the Physics Gap. Everything else in this article grew out of that one failed survey.
Every industry has a cost structure
Here’s the second part of the argument. Every industry runs on a setup that has hardened over time. It’s built from labor hours, owned equipment, wait time and the cut each middleman takes. That setup decides what things cost.
A startup that copies the setup inherits the cost. It might have a nicer app, but it’s stuck on the same cost curve as everyone else. The incumbent can match it, and usually does.
This is why customer feedback can’t be the starting point. Feedback tells you where the setup rubs. The floor tells you how much of the setup shouldn’t exist at all. Those are different questions, and only the second one leads to a structure the old players can’t easily copy.
Start from the floor
So where should you start? With the result, not the customer’s complaint.
Ask what the result costs today. Count people time at a loaded rate, plus tools, outside help and the cost of redoing mistakes. Then ask what it would cost if nothing were wasted. That’s the floor. It holds only what the work truly needs: compute, materials and the checks a law or a contract requires. It holds no waiting, no handoffs and no extra sign-offs. It’s never zero.
Divide today’s cost by the floor. That ratio is the Physics Gap.
A big number means there’s a lot of waste to attack. A number close to 1 means the market already runs lean, and you should move on. People want the result to be fast, cheap and accurate. The gap tells you how far from that today’s setup really is.
This is the part of the Venture Proof method that changes the order of the work. You do the math first. You don’t wait for customers to point you somewhere.
A worked example
Here’s a made-up case to show the math. The numbers are examples, not client data.
A sales operations team pulls its last 10 feature requests. Eight of them only ease a workaround. One asks for a faster export from the sales system to Excel. One asks for a saved Excel template for the weekly report. One asks for a way to copy notes into email. One asks for a reminder to send the report on Friday.
Six of the eight trace back to one step: building the weekly account report. Nobody is asking for a new report. They’re asking for help building the one they have.
So the team stops looking at the requests and names the result. The weekly report exists so regional leaders know which accounts are at risk this week. That’s the result. The report is just today’s way of getting it.
In each region, building it takes about seven hours a week. An analyst pulls data from three systems, pastes it into Excel, cleans it, builds charts and writes notes. A manager reviews it and sends it.
Here’s today’s cost per region:
Analyst time: 6 hours at $60 an hour is $360 a week.
Manager time: 1 hour at $100 an hour is $100 a week.
Total: $460 a week.
Now the floor. Compute could pull and match the data for about $1 a week. A person still needs to read the accounts flagged as at risk. That takes about 15 minutes, or $15. The floor is about $16 a week.
$460 divided by $16 is about 29. That’s a big gap. Across 12 regions, today’s cost is about $287,000 a year. The floor is about $10,000 a year.
The requests would have shaved minutes off a seven hour job. The math says most of those hours shouldn’t exist.
Then test it with people
The math gives you a starting point. It doesn’t give you permission. People still have to prove it.
Here’s the order I use:
Map the job step by step, so you can see where the waste sits.
Interview the people who do that work. Ask about the steps, not the features they want.
Check what they’d pay to have the gap closed.
Build a small working test before you build the real thing.
In the example, the team talks to four analysts and two regional leaders. The analysts confirm the hours. They add one thing the map missed. About one week in four, the three systems don’t match, and cleanup takes an extra hour. The leaders say they mostly skip the charts. They read the list of at-risk accounts and the notes.
That tells the team the charts may not be needed at all.
Then the team asks the head of sales operations a direct question. Would she pay a stated yearly price, well under today’s $287,000, to get the list without the build? She says yes, if she can see it work in two regions first.
So they run a test for 4 weeks in 2 regions. A short script pulls and matches the data and produces a plain list of at-risk accounts, with no charts. An analyst reads the list and adds notes. They set the rules before they start. It’s a go if a person spends under an hour a week and leaders act on the list in at least 6 of the 8 region weeks.
The result is about 45 minutes a week, and leaders act on the list in 7 of 8 region weeks. That’s a go.
Notice where the customers came in. They didn’t design anything. They confirmed the steps, corrected the map and said yes or no to a price. That’s the job they’re good at.
The counter-case
The strongest pushback I hear goes like this. “You’re telling me to ignore my customers. That’s arrogant. The companies that listen win.”
I’m not telling you to ignore them. I’m telling you to ask them the right questions at the right time. Customers are the best source on how the work runs today. They know where it waits, what goes wrong and how often. They’re the only ones who can tell you whether they’d pay. I interview them on every project.
What I won’t do is ask them to design the answer. That’s asking the passenger to design the engine. It isn’t respect for the customer. It’s handing them a job they can’t do and then blaming them when all you get is a small fix.
The second pushback is about Kodak itself. “Kodak’s problem wasn’t research. It was a business that made its money on film and didn’t want to hurt that.” That’s fair, and I can’t prove what was said in every meeting after 1975. My point is narrower. If you start from what your current customers ask for, you’ll keep getting requests that protect the current setup. That makes the business problem harder to see, not easier.
The third pushback is that surveys are useful. I agree. Use them after you’ve removed the waste. Survey the leaders using the new list. Ask what’s confusing or missing. That’s polishing a surface that’s no longer built on waste. It’s the right tool, used at the right step.
What to do Monday
Here’s a test you can run with your own backlog this week.
Pull your last 10 feature requests.
For each one, ask a single question. Does this make a workaround hurt less, or does it remove the step?
Count how many only ease a workaround. If it’s most of them, you’re polishing the old setup. That cutoff is a rule of thumb, not a law.
Group the requests by the step they touch. Pick the step that shows up most often.
Write down the result that step is for, and who needs it. One sentence. A result, not a feature.
List every step it takes to get that result today, with who does it and how long it takes.
Price today’s cost, then the floor. Divide. A number close to 1 means move on. A big number means dig in.
Talk to 3 to 5 people who do the work. Ask them to walk you through the last time. Ask where it waited and what went wrong. Don’t pitch your idea.
Ask the person who controls the budget what they’d pay to close the gap.
Write your go and no-go rules, then run a small working test before you build anything.
Where this breaks
This approach has limits, and you should know them before you start.
First, the floor is an estimate. If you set it too low, you’ll chase a gap that isn’t there. List what’s in it, line by line, so someone can argue with each part.
Second, a big gap means nothing if nobody will pay to close it. That’s why the budget question comes before the build.
Third, watch for the mistake from my client study. Make sure the one doing the job is a person, not a machine or a system. If your interview needs three kinds of experts to answer, you’ve framed the wrong job.
Fourth, some work really is about the experience. If the gap is close to 1 and the complaints are about how it feels, survey work and design work are the right tools. Don’t force a structure problem where there isn’t one.
The close
Kodak had the first digital camera in a drawer. The people it listened to were film customers, and film customers wanted better film.
Your backlog is probably full of the same kind of request. Each one is real. Most of them ask you to make a step hurt less. Start with the math instead. Find the step that shouldn’t exist, and use your customers to prove it.
Get the free download
Grab the free checklist, with the playbook included. The checklist gets you started this week. The playbook walks you through the full sales operations example, step by step.
We work through this in the Innovation Unpacked Mastermind.
If you want to test an idea of your own, book a call.
Sources
Sasson’s 1975 camera, its toaster size, the 1978 patent and no production: IEEE Spectrum
Management’s reaction: Sasson to the New York Times, May 2, 2008
Book an appointment: Click here
Email me: mike@pjtbd.com
Call me: +1 678-824-2789
Follow me on 𝕏: https://x.com/mikeboysen
Articles - jtbd.one - De-Risk Your Next Big Idea
Website - pjtbd.com
Slaying the Sacred Cows of Innovation









