Why does a capacity problem stay invisible for so long?
Because the early symptoms look like normal agency life rather than a warning sign. A senior person covering two roles for a sprint because someone’s on leave. A project running a week behind because the brief changed mid-flight. Nobody logs these individually as “capacity risk”: each one has a specific, reasonable explanation, and reasonable explanations don’t trigger alarms.
What actually accumulates underneath the individually-explainable delays is buffer: the informal slack that lets a team absorb one bad week without anyone downstream noticing. Every stretched sprint spends a little of that buffer, and it rarely gets replenished. Replenishing it means deliberately staffing below full utilisation, which looks like inefficiency on a spreadsheet even though it’s actually insurance.
What does the actual breaking point usually look like?
Two client deadlines landing in the same week that individually would have been fine, and jointly aren’t, because the buffer that would have absorbed one of them was already spent months earlier on unrelated small overruns. From the outside it looks sudden: “we just had a bad week.” From the inside, the bad week was the first time the accumulated deficit became visible rather than the first time it existed.
The tell in hindsight is almost always the same: someone senior had been saying “we’re a bit stretched” in a hallway conversation for weeks before it became a formal problem. The informal warning didn’t turn into a plan because there was no immediate deadline forcing the conversation.
What does it actually cost to fix this after it’s already broken, versus before?
Caught early, the fix is close to routine: bring in temporary overflow capacity for a defined stretch, at a planned cost, decided with enough lead time to brief the extra people properly. The client on the other end never sees the mechanism: the deadline holds, and nothing about the delivery looks different from the outside.
Caught late, the same fix costs more in every direction at once. The overflow capacity is now urgent rather than planned, which narrows who’s available and usually raises the price of getting them. The briefing happens under pressure, which is precisely when handoff mistakes are most likely: the new people are least equipped to catch a gap in an already-stressed project. And the client relationship absorbs a hit regardless of whether the deadline is ultimately saved, because the scramble itself is often visible even when the delivery technically lands on time.
The gap between those two costs is the entire argument for treating buffer as a planned expense rather than a discretionary one. It’s cheaper by every measure to build the insurance in before it’s needed than to buy it back, urgently and at a markup, once it’s already gone.
How much buffer is actually enough, without just padding every estimate?
There’s no universal percentage that’s correct for every team, but the wrong instinct is padding every individual task estimate slightly to create informal slack. That approach hides the buffer inside inflated numbers nobody can see or manage, and it erodes the first time a client or a project lead pushes back on a timeline that looks generous on paper.
The more useful approach treats buffer as a visible, named allocation at the team level rather than a hidden one at the task level: a defined percentage of total capacity deliberately held back from committed project work, reviewed and adjusted based on how often it actually gets used over a quarter. If the buffer is consistently untouched, it’s probably oversized and can be trimmed toward more billable capacity. If it’s consistently exhausted before the quarter ends, that’s a concrete, measurable signal that either the buffer needs to grow or something upstream is generating more unplanned work than the team is structured to absorb.
The reason this matters more than picking a specific number is that it turns a vague, uncomfortable feeling (“we’re stretched”) into a metric someone can actually track and act on before it becomes a missed deadline, rather than a hallway conversation that never quite becomes a decision. Vetting the overflow partner before the quarter that needs one is its own checklist.
