Growth had stalled and everybody knew it. We put more money into marketing. It did not work. We invested in sales enablement. It did not work. We pushed the sales organization harder, which is what you do when you have run out of better ideas, and that did not work either.
The P&L looked fine throughout. Margins held, the quarters closed roughly where they were forecasted, and nothing in the statement pointed anywhere in particular.
What we eventually found, by calculating how much of our revenue came from recently developed products, was that our R&D investment was barely contributing. Enough to cover churn. Not enough to deliver growth. We had been funding development for years, and it was replacing what we lost rather than adding to it.
So the P&L was not hiding the problem. Everybody could see the problem. It was hiding the cause. And a statement that points nowhere sends you to the levers you can see: marketing, enablement, sales pressure. All of them downstream of the thing that was actually broken.
The standard P&L won’t tell you this
There is no row for revenue from things we built in the last three years. A dollar from a product launched last quarter and a dollar from a product launched eleven years ago land in the same place and are indistinguishable once they get there.
So the failure is structurally invisible. You do not find it by reading the statement more carefully. You find it by calculating an extra number: the share of revenue coming from things you recently built.
There is no universal benchmark and anybody offering you one is selling something. The comparison that does the work is against your own investment. Put the share of revenue coming from recent products next to the share of revenue you spend on R&D. If you are spending twelve or fifteen percent of revenue developing things, and four percent of revenue comes from anything developed recently, the arithmetic has already made the argument.
It reads in both directions. A strong P&L with a low share is the dangerous case, because nothing is forcing that particular conversation. It means you are being subsidized by your own past. The business paying for everything was built by people making decisions years ago, and the subsidy runs for exactly as long as that business holds. Nobody will tell you when it stops. A struggling P&L with a low share is uncomfortable and much simpler: you have found the reason it is struggling, which is worth something, because the competing explanations on offer would have cost you another year.
It is a report, not a forecast
You need several years of history to read this, and that is not the limitation it sounds like. Any company with a strong P&L and stalled growth already has three years of history. That is what the situation is made of. The number reports what happened, which is a different thing from failing to predict.
What it genuinely will not tell you is what has happened most recently. One three-year figure is a single reading, and a single reading has no direction in it.
So take a series instead. Run the same calculation across rolling windows: years six through four, five through three, four through two, three through one. Then set each window against the R&D spend that produced it, offset by however long your development cycle runs. If that is a year, the three-through-one window pairs with spend from four years ago.
Now you have direction, and you have it against the input. A rising share on flat spend means your investment is getting better. A rising share on spend that doubled is something else, and worth knowing before anybody celebrates.
For anything faster, run the same new-versus-old cut on bookings and wins, and on pipeline. Each step earlier buys time, but it costs certainty. Watch for gamesmanship: pipeline can be inflated, but nobody can inflate what a product earned three years ago. Split the pipeline anyway, because it answers a different question. Revenue tells you whether your new products worked. Pipeline tells you whether anybody is selling them.
The drain that never appears as cost
A great deal of what an organization spends is not money. It is capacity. If a large share of your revenue comes from old, low-margin business, that business is still consuming sellers, customer success, and in some companies manufacturing. Every one of those is something a new product needs in order to land.
So you can fund a new product properly and starve it anyway, because the people who would have to sell and support it are fully occupied servicing something that returns less margin per hour of their attention. The investment shows up in the P&L. The starvation does not.
Two questions worth asking in the same sitting. How much are you spending to maintain the cash cow. And how long have you been investing in the same space. Both can be entirely correct. Neither is usually a decision anybody made recently, which makes them worth re-deciding rather than inheriting.
What I found
The instinct when growth is not coming from new products is to look at R&D throughput. Velocity, headcount, delivery dates, whether the roadmap is overloaded. That was the wrong place. Our products got built, roughly when they were supposed to be. They never landed, and a product that ships and does not land is not an engineering failure.
It is one of three things, and they have three different owners. Prioritization: you built things, but not the right things, and what you built was never on any customer’s list. Definition: you built what you described and the description was wrong, so customers evaluated it and chose something else. Go-to-market: the product was right and nobody was ever set up to sell it, so it was never evaluated at all. All three look identical in the P&L, because all three produce the same result, which is spend with no revenue attached to it.
Ours was mostly definition, and underneath that it was accountability.
We had product definers. Good ones. They did the market work, wrote the requirements, and got things through the funnel and into development. Not one of them had any accountability for whether the product succeeded once it existed.
They were incentivized to make new products, not to make new products successful.
That is not a character flaw and it is not unusual. It is what happens when an organization measures the part it can measure. A launch is a date. It is controllable, visible, and attributable to a named person. Success is slower, noisier, and depends on sales coverage, pricing, support readiness and a dozen other things the definer does not control. So you measure the launch, and after a few years you have built a function that is very good at producing products and holds no stake in what becomes of them.
What changed, and what it cost
The roles changed. The objectives changed. A few of the people changed.
Three reactions, and they are worth expecting.
Some read it as blame. If you make somebody accountable for an outcome they do not wholly control, and you do not say plainly that you will back them when a sound decision goes badly, it looks like you are building a file. That reading is predictable, and the people most likely to arrive at it are often the ones you most want to keep.
They were incentivized to make new products, not to make new products successful.
Some agreed it was right and could not do it yet. This was the largest group. Making a definer accountable for commercial success asks for things a definition role never required: pricing judgment, enough of the funnel to see where deals stall, the ability to brief a sales team and tell whether the brief landed. Those are learnable, and they are not automatic. The structure change is what made the gaps visible, which is a feature only if you then do something about them.
And some already worked that way. The change did not improve those people. It stopped penalizing them.
I remember one of the definers laughing at another, because he had so many more products released and in development.
The other one said: because I am busy making my products successful.
It stung, and it was completely right. Under the measure we had been using, the person doing the job properly looked like the underperformer. That is not a motivation problem and it is not a talent problem. It is a scoreboard doing exactly what it was built to do, and it had been working perfectly for years.
The review they had to invent twice
The company was working the same problem from the top, and got it wrong once first, which is the useful part.
It started with new product pricing reviews. Executives looking at what was going into development across the whole company, with the explicit goal of driving margin up. Sensible, and it worked, in the sense that what came to the reviews got steadily more margin-rich.
Then somebody noticed that groups were proposing high-margin products with no high-volume customers behind them. Superb margin on the page. Almost no contribution in practice. The gate was being passed by optimizing for the gate.
Which is the same failure as mine, one level up. Measure the part you can measure at the moment you can measure it, and you will get exactly that, because people are not stupid and they respond to what is being scored. A margin percentage at proposal stage is knowable, checkable and attributable. Whether anybody will buy it in quantity is none of those things, not yet.
So the reviews changed. Not what are you going to build and at what margin, but what happened to the thing you built.
One of my people presented into those reviews on a product that had been delayed in development. While it sat, a competitor shipped something similar. He still exceeded revenue and margin anyway.
That is the behavior the whole exercise is trying to buy. The delay was not his to prevent and the competitive position got worse while he waited, and he still delivered the outcome. A pricing review would have scored the proposal he wrote a year earlier. Only a success review can see what somebody did with a bad hand.
Who owns whether this makes money
For any new product you are funding, one question sorts most of this out. Who owns whether this makes money?
Not who owns the specification. Not who owns the launch. Not who owns the engineering. Who owns the outcome, by name, through the period in which that outcome is supposed to appear.
Then the harder question. Who gets to decide to kill a product that is not going to be successful?
You often cannot know that at the start. You find out once you are already building, which is exactly when stopping is least popular, but not nearly as expensive as finishing it and consuming go-to-market and sales resources on a loser. And if the person who owns the outcome cannot stop the work, they do not own the outcome. They own a result they are not permitted to influence. That is not accountability, it is exposure.
Kill a loser as soon as you know it is a loser. Early, it gives back the thing new products are shortest of, which is capacity. Late, it gives back almost nothing, because the money is spent and the sellers and the support people have already given it a year.
If the answer to either question takes more than a few seconds, or if it is a committee, or if ownership quietly transfers at launch to somebody who was not in the room for the decision, then you already know what your new product revenue share will look like in three years. You have simply not calculated it yet.
Calculate the number
One number, and nobody will hand it to you: the share of your revenue that comes from products you developed in the last three years, set against what you spent to develop them.
A P&L reports what the business earned. It does not report how old that is. So go and work it out.
If the share is healthy, you have learned something cheaply. If it is not, you have found it while the mature business is still paying the bills, which is the only comfortable time to find it. The alternative is finding out when the subsidy stops, and by then the decisions that would have fixed it are three years behind you.