The distraction tax.

Every product organization I have been inside eventually hits the same three markers. Not because the leadership made a mistake. Because the company succeeded.

The first marker is a mature product that has to keep working. Paying customers, support tickets, integrations that broke quietly six months ago that someone now needs to fix. As the business scales, the operational surface grows with it — availability, security, observability, supportability, maintainability, compliance. None of those were on the founding team's mind at eight engineers. All of them are baseline expectations now. Most of the engineering team's day is spent keeping that surface healthy.

The second marker is a team that has evolved. The engineers who shipped the original may be gone — promoted out, moved on, hired into a competitor. The codebase carries the fingerprints of every generation of stewards that passed through — different architectural opinions, different naming conventions, different ideas about the right abstraction. Code that was internally consistent at fifteen engineers reads inconsistent at fifty. The institutional memory of why a decision was made three years ago lives in commit messages and in the heads of two former employees. Every new engineer onboards into a system whose explanations are partially extinct.

The third marker is tech debt. Ward Cunningham coined the term technical debt in 1992 to describe exactly this — short-term decisions that buy productivity now and accrue interest later. Engineers understand it intuitively. Almost nobody else does. The board cares about the business growing. The customers care about features shipping. Tech debt is a balance sheet that does not show up on any actual balance sheet — and engineers pay it back with interest, every release. Decisions that were correct at eight engineers have aged badly at forty. The right move is a refactor. The refactor takes six months. The roadmap is full.

A word on the term bet, since I will keep using it. A bet is a strategic initiative pursued on imperfect information — without the full analysis a normal prioritization model would demand. It is a hypothesis, not a forecast. Good product development has always been about developing and testing hypotheses; bet is the colloquial term for that same work. The best leaders I have worked with do not gamble — they test hypotheses structurally and routinely, planning four to six quarters out. That is the discipline.

Inside that gravity, the bet — the strategic one, the one the next five years depend on — loses every prioritization conversation. Not because anyone said no. Because we'll get to it next quarter is what every leader says when the alternative is delaying a customer-promised release.

There is a name for what that costs. I call it the distraction tax: the price every product leader pays when the team that runs the core business is also asked to ship the bet.

The tax does not show up on a line item. It shows up distributed — the discovery session rescheduled three weeks running; the senior engineer pulled back to a production fire who never returns to the bet at full capacity; the bet that was on the Q1 strategy slide that is now on the Q3 slide, same description, same owner; the workspace still open eighteen months later, last update from a Tuesday in November.

Nobody adds them up, so nobody declares them. The tax compounds anyway. The bet that needed to ship in six months ships in eighteen. A competitor with a more contained engagement ships in three. The wedge closes.

There are companies that ship the mature product and the strategic bet without paying the distraction tax. They tend to share one of two structural conditions.

The first is being exceptionally well-funded — enough to run two, three, or four circuits in parallel. The shape varies: a core-product org plus a dedicated innovation team; Google-style twenty-percent time; multiple independent engineering functions. Every version requires sustained funding through quarters of unclear ROI, plus staffing capacity that does not rob the core. Most companies cannot or will not commit to that.

The second is having had extreme discipline from day one — codebase kept clean, tech debt addressed in flight, engineering culture that treats maintainability as a first-class deliverable. This is what every engineering blog post recommends and what almost no organization actually does at scale.

Most companies are in neither state.

The last twelve months made the tax worse. Every team has been told to experiment with the latest AI tools. Every engineer is being told they are behind the curve. Every executive has watched a competitor demo something flashy and walked back asking what the AI roadmap looks like. And the technology is moving so quickly that the world changes every three months — the model that was state of the art at the start of the quarter is mid-tier by the end; the workflow that was experimental in January is table stakes by April. The FOMO is industry-wide and structural.

Pushing AI experiments through the same team that runs the mature product does not relieve the distraction tax. It compounds it.

If the company does not figure out how to run a second circuit, the strategic choice narrows to two trades. Both are losing. Clayton Christensen named the pattern the innovator's dilemma in 1997, and the math has only sharpened since.

The first is innovating something new and letting your existing customers and brand presence drift — reallocating toward the new bet, watching support escalations rise and renewals get harder while the new bet is still proving itself. Customers do not wait.

The second is focusing on the existing business and letting the future market bypass you — protecting the core, watching the strategic bet sit in the backlog until a competitor ships it and takes the wedge. Markets do not wait either.

Both options assume you have to pick. You do not — but only if a second circuit exists.

I have watched a second circuit work in three shapes.

Acquisition. Buy a company, a team, or a skill set. The label on the deal varies — acquisition, acqui-hire, tuck-in — but the shape is the same. Works when a target is available, the price is sane, and the acquired leadership is willing to operate inside yours. The cost is integration: acquired engineers pulled into all-hands and architecture reviews; product leaders spending two quarters figuring out what to do with the roadmap they just bought. The bet sometimes survives. Often it does not.

A dedicated internal organization. Stand up an innovation team, ring-fence an engineering group, or structure twenty-percent time across the org. Dedicated staffing on payroll, protected from quarterly prioritization gravity. The constraints are real — bounded by who you can hire, the budget you can sustain through ambiguous quarters, the bandwidth to keep them from being absorbed back into core, and the people-management that comes with permanent headcount. When it works, it works for years. When it fails, it usually fails on the funding line.

Partner.Engage an outside organization to run a discrete leg on a contained clock, with the asset transferred back at the end. The mature-product team does not lose people. Leadership does not absorb integration work. The partner brings its own pattern for closing the loop on a bet, runs it outside the core organization's prioritization economy, and hands the asset back when the leg is done.

Which shape fits depends on the bet. A market position another company already holds — acquisition. A capability the company will need permanently and can sustain the funding profile for — internal org. A discrete strategic experiment on a fixed clock — partner.

Every executive I have worked with carries a backlog of bets they believe will define the next five, ten, or fifteen years of the company. Most have not been tested, because the team that would have tested them has been busy keeping the mature product alive.

The longer a bet sits in the backlog, the higher the probability someone else gets there first. There is no version of waiting that does not narrow the window.

The question is not whether the distraction tax is real. The question is which shape of second circuit fits the bet you have been deferring.