Your Best Idea Already Happened. You Just Can't Find It.
Innovation. A word that can inspire a room, terrify a budget committee, and put half a conference audience to sleep, sometimes all in the same meeting.
People and organizations spend billions of dollars and thousands of hours a year on it, hire for it, build entire careers and teams around it.
And yet the pattern repeats everywhere I look. Smart people, budgets the size of the GDP of a small country, genuinely good technology, and still, anemic, seemingly random outcomes. Someone finds something real, gets celebrated once, and neither they nor the organization around them can explain how to make it happen again.
If I had to rank the problems loudly holding people and organizations back from their own best thinking, this one lands in my top three.
Approximately 2,100 words · 10 minute read
Somewhere in your company right now, a genuinely good idea is quietly rotting in a slide deck nobody has reopened since the meeting it was first presented in. I’d bet actual money on it. Every organization has a version of this: a real insight, the kind that would change a decision if anyone remembered it existed, filed away and forgotten the moment the applause ended. The idea didn’t fail. It just had nowhere to live once the moment passed.
Most companies treat innovation like lightning. It strikes, everyone’s grateful, and nobody knows how to make it happen again on purpose. Someone finds something real: a pattern in the data, a fix that works, an answer nobody had before. It gets celebrated once, maybe it changes a decision that quarter, and then the company goes back to waiting for the next flash of luck. Ask most leaders how their last great insight came about, and the honest answer is usually some version of “someone happened to notice.”
That’s not bad luck. It’s an idea that never got industrialized.
Build the Factory, Not the Fluke
“Industrializing innovation” sounds like it’s about paperwork. It’s closer to what happened to manufacturing two centuries ago, when making something well stopped depending on which craftsman was having a good day, and started depending on a process that produced good output on a Tuesday as reliably as a Friday.
A real factory doesn’t celebrate one lucky batch and stop there. It’s built so tomorrow’s output matches today’s, regardless of who’s standing at which station. Nobody touring a car plant asks which worker made this particular one great. The system is built so that question never needs asking.
Innovation, in most companies, has never had its factory built. It runs on individual craftsmanship, one sharp analyst, one curious product manager, one team that clicked for a quarter, and when that person moves on, so does the capability.
Building that floor plan is what industrializing innovation actually means: a system where good ideas don’t depend on luck, one person’s instincts, or being in the right meeting at the right time.
Fast and Broken, or Careful and Dead
Most organizations run on one of two models, and both fail in predictable ways.
In the first, people move fast and build things on their own. There’s energy here, curiosity, people trying things without waiting for permission. But there’s no shared coordination and no record of what worked. Two teams solve the same problem twice, six months apart, neither aware the other did it first. A great finding lives in one person’s head, gets used once, and quietly disappears when that project wraps up.
There’s older research on this, older than I am, in fact. It’s a seminal study nobody seems to bring up anymore, and it’s worth revisiting. It explains why this gets worse, not better, over time. Ralph Katz and Thomas Allen’s classic study of 50 R&D project groups found that teams working in isolation don’t just miss out on outside ideas, their performance actually declines the longer they stay insular, as they progressively cut themselves off from information sources outside the group. Isolation compounds. Left alone long enough, a team gets worse at the thing it was formed to do.
In the second model, everything routes through a slow, centralized process before anyone’s allowed to move. Columbia Business School’s David Rogers, writing in MIT Sloan Management Review, describes exactly this pattern: innovation never happens at scale as long as it relies on ad hoc exceptions approved by individual executives.
New ventures get greenlit because one senior sponsor likes them, then sit in traditional silos while resource allocation crawls through weeks or months of approvals. And because each project has a powerful name attached to it, almost nobody is willing to shut it down, even once it’s clearly not working.
The result isn’t rigor. It’s politics wearing a process costume.
Recent innovation-management research keeps landing on the same three failure modes, whichever industry it’s looking at: disconnected bets with no shared strategy, innovation theater that never touches revenue, and leadership that talks about innovation while budget and incentives still reward the safe, short-term choice. Structure is the missing ingredient in all three, not talent, and the opportunity sits in the space between the two extremes above: matching the right person, with the right tool, to the right problem, inside a system that captures what worked well enough to use it again.
Speed was never actually the edge. Being able to build on what you already found, without losing it, is.
The Argument Every Company Has Already Had
Two departments report “active customers.” One counts anyone who logged in once. The other only counts someone who bought something in the last 90 days. Neither team is wrong, they just built their own definition in isolation, and nobody noticed until both numbers showed up in the same meeting, in front of the same executive, who now has to guess which team to trust.
Nearly every company has some version of this story, and it’s the factory problem in miniature. No shared floor plan, so two teams quietly build two different machines and call them the same thing.
The fix isn’t complicated: one definition, owned by someone, with a visible record of when and why it changed, so the next person who needs it doesn’t have to rebuild it from scratch. That’s the assembly line. It works the same way whether you’re applying it to a customer count, a pricing model, or the process behind last quarter’s best idea, a shared, versioned starting point that anyone on the floor can pick up and keep running from.
Ask yourself right now: if two people on your team answered the same question today, would they get the same answer?
The Skeptic’s Version of This
You could reasonably argue this is just good discipline with a new label, the same rigor teams have always needed and mostly skipped. There’s truth in that. What’s different now is the cost of skipping it. A one-off inconsistency used to be a minor annoyance you cleaned up in a follow-up email. Confidently repeating the wrong answer at scale, with a tool that sounds sure of itself, is a different kind of failure, one that erodes trust in everything that comes after it. The discipline is old, what’s new is the price of ignoring it.
Why One Good Answer Isn’t Enough
An analyst finds a great insight and presents it once. If the reasoning behind it only exists in their head and a spreadsheet nobody else opens, that insight is a one-time event, not a capability the company has. The next person who needs a similar answer starts from zero.
A tracked, shared idea changes that. The insight stays available, stays correct, and stays usable by whoever needs it next. Reproducibility, in plain terms: ask the same question twice, or have someone else ask it, and you get the same answer both times, with a clear record of how it was built.
This is the same argument Bansi Nagji and Geoff Tuff made in their now-classic Harvard Business Review piece on innovation portfolios, applied one level down. Their point wasn’t that companies need more innovation, it was that resources need to be allocated deliberately, across core work, adjacent bets, and genuinely new territory, rather than handed out by instinct. A single insight, found once and never repeated, is what happens when that same deliberate allocation is missing at the ground level. A great analysis, produced once, is a nice story people tell in a meeting. A system that keeps producing trustworthy ideas, quarter after quarter, is a real advantage.
The Floor Plan
Strip away the jargon and it comes down to five stations on the line, none of them exotic, and none of them theoretical. Versions of all five already exist and are working right now, just usually in isolation from each other.
- A place where what worked gets written down, not left in one person’s head. Engineers have had this for decades in version control, git, where every change to a codebase is tracked, attributed, and reversible. Amazon runs the same idea for ideas instead of code, its famous six-page memo and PR/FAQ process forces a proposal into a shared, reviewable document before anyone commits real resources to it.
- A fast lane for testing the next idea, so a promising thread doesn’t sit in a queue for weeks. Booking.com built exactly this, an internal platform that lets any team launch a controlled experiment quickly, which is how a company running tens of thousands of tests a year still moves faster than most competitors running a handful.
- A quality check before anything goes company-wide, one more set of eyes, not a committee.
- A clear line back to business value, so the system never becomes an end in itself. A structure that produces beautifully governed answers nobody needed is still a failure, every idea on the line has to earn its place by tying to an actual outcome, not just the appearance of activity.
- A person who owns the final call on anything going out the door, to a customer, a board, a regulator.
That’s the floor plan. It’s not specific to any one department or any one kind of idea, it’s just what it looks like when a company decides a good result is worth being able to produce more than once. The pieces already exist. Almost nobody has put all five on the same floor. None of this requires a breakthrough, it requires assembly.
The Person Already Running the Line
The people who bridge a business question and a trustworthy answer are rarely hired into that role directly. They’re usually already inside your building, doing the work informally without the title. The analyst who’s already fielding “what does this number actually mean” questions from three different departments. The team lead who’s already fluent in both the business problem and the data behind it.
If you’ve read my piece on how great teams are built through the middle, this is the same person, just wearing a different jersey. They’re rarely the one getting credit in the room. They’re the one everyone quietly relies on to make sense of things, and most of them have never been told that’s a real job. It’s easy to miss them precisely because the value they add doesn’t show up as a single, attributable win, it shows up as the absence of confusion, the meeting that didn’t spiral into an argument about whose number is right.
You probably already know who that person is on your team. Find them. Give them a governed tool and real ownership over the shared definitions, and their next good answer becomes part of the system instead of another one-off result someone has to rediscover later.
The Tenth Question Gets Cheaper
The tenth question your team answers this way should be faster and cheaper than the first one, because the definitions, the trust, and the tooling already exist by the time that tenth question comes in. That’s what sustained innovation looks like: not a string of lucky one-off wins that each start from scratch, but a coordinated system where every answer makes the next one a little easier than the last.
The next edge in analytics won’t belong to whoever finds the insight fastest. It’ll belong to whoever builds the system where nothing has to be found twice.
I’d like to know: where has your team gotten a great result once and never managed to repeat it? What actually happened to it?