Sit with any engineering leader at a company past about $20M and ask what share of their capacity goes to new work. The answer is usually somewhere between a quarter and a third, and it is usually delivered with an apology.
Nobody decided this. There was no meeting where a leadership team agreed to spend three quarters of its most expensive resource on maintaining decisions it made years ago. It is the accumulated weight of every product that was kept, every segment that was not exited, and every integration promised to close a deal in 2021.
Complexity is the tax a company pays for every decision it declined to make.
Why it accumulates in one direction
Adding is a decision with a name attached. Somebody championed it, somebody funded it, and if it works somebody gets credit. Removing is a decision with only downside: a customer to disappoint, a colleague to overrule, a revenue line to explain away at the next board meeting.
So companies accumulate a ratchet. Every quarter adds surface area and almost no quarter removes any. Nobody is measured on removal, so nothing is removed, and the capacity available for new work erodes by a few points a year in a way that never appears on any dashboard.
By the time it is visible, it presents as “the engineering team is slow.”
The grid, and why it is uncomfortable
Put every product, segment and major integration on two axes: growth potential and technical viability. Four quadrants fall out.
Fund it properly. High on both. This is the wedge. Concentrate here, and say out loud what you are giving up to do it.
Modernize. Real demand, structurally constrained. Rebuild the foundation or lose the demand later, on someone else’s timetable.
Merge. Technically sound, commercially thin. Fold it into something that has pull rather than maintaining it as its own thing with its own roadmap and its own quarterly defence.
Retire. Low on both. Every quarter it survives is capacity the wedge does not get.
The exercise takes an afternoon. What takes months is that each of those boxes has a defender inside the building, and placing something in the fourth quadrant is a statement about a colleague’s last three years.
Doing it without detonating the company
Instrument before you decide. I have made these calls on qualitative judgment under time pressure and been right, and it was still harder to defend than it needed to be. Revenue, gross margin, support load and engineering hours per product, in one table, buys enormous internal consent for about six weeks of work.
Name the sunset list even if the dates stay soft. Ambiguity about which products are being wound down costs more energy than the decision itself. People will manage a known ending. They will not manage a rumour.
Say what you are giving up. Every simplification has a cost: a customer segment, a logo, a revenue line. State it in the same sentence as the benefit. Leaders who present simplification as free lose credibility the first time somebody finds the cost themselves.
Give removal an owner. A named person, accountable for what gets stopped, with explicit cover from the top. Without that, portfolio simplification is a slide.
What it is actually worth
At a multi-product SaaS business shaped by six acquisitions, four of them never integrated, commercial integration had run years ahead of technical integration. Investment was spread thinly and evenly, which felt fair and meant nothing had enough behind it to win.
Rationalizing it moved the business: reducing investment in the structurally constrained areas, concentrating behind two commercial wedges, and reorganizing engineering around capabilities rather than inherited product boundaries. That took the business from steep contraction back toward growth, and made the story investable enough to support two funding rounds.
The reorganization was the easy part and the socially expensive one. Naming the sunset list was the hard part and the one that actually mattered.
