Every January, somewhere in a planning review, a version of the same exchange happens. Someone asks whether the team can actually carry the number. Someone else pulls up the capacity model, points to the headcount plan, and says yes: quotas are assigned, reqs are approved, the math closes. Everyone moves on to pipeline.
I have sat in that review more times than I can count, on both sides of the table, and the question that almost never gets asked is the one that decides the year: how much of this quota belongs to people who do not exist yet?
Phantom capacity is quota carried by seats that are empty or not yet productive. It's the unfilled req holding a full-year number. That start date that slipped from February to April while the quota stayed put. The mid-ramp hire counted at full productivity because the model assumes a ramp curve the team has never once achieved.
None of it looks like a problem when your new revenue plan goes live, because on deploy day the plan is a spreadsheet, and every seat in a spreadsheet is full.
Of the 23 failure modes we mapped in How Revenue Plans Break, phantom capacity is the one teams are most certain they do not have, and the certainty comes from a legitimate place. The hiring plan was approved. Finance signed it, recruiting is working it, and the dashboard shows the reqs are open. An approved hiring plan still has to survive sourcing, offer cycles, start dates, and ramp before it becomes revenue capacity, and every one of those steps leaks time that the quota model rarely gives back.
The research behind the field guide puts numbers on the leak. Across 403 B2B revenue plans, the benchmark for full AE ramp was 5.3 months, while plans routinely assumed less. And the plans in the strongest third on capacity realism averaged 80% quota attainment against 56% for the weakest third, a 24-point gap that was measurable before the year started.
The cruelest part is the timing. A capacity gap created in January surfaces in Q3, and by then it has changed costumes: it walks into the QBR looking like an execution problem. Reps miss, pipeline reviews get tense, and the autopsy lands on coaching and activity, while the arithmetic that guaranteed the miss was visible in the plan the day it was approved.
Capacity Realism covers five failure modes in the field guide, and they compound with each other.
Total assigned quota sits well above the operating plan to create cushion, and the cushion quietly teaches reps that targets are fiction while hiding an expected shortfall from day one.
New hires reach productivity slower than the ramp curve assumes, so planned capacity arrives light and late, even when every hire lands on schedule.
A departure with no backfill req behind it leaves quota structurally uncovered while the forecast keeps counting it, and every week of delay pushes the recovery further into next year.
Open reqs sit unworked because the managers who own them have no interviewing capacity, which means the hiring plan will slip regardless of what the recruiting dashboard says.
1. How much of our assigned quota is carried by people who are not hired or not ramped? Add the quota on unfilled seats to the quota on hires still inside the 5.3-month benchmark. Most leaders who run this number for the first time do not like it.
2. What was our last cohort's actual time to productivity, and does the plan use that number? The ramp assumption in the model and the ramp your last three cohorts actually achieved are frequently different numbers, and only one of them pays commissions.
3. For every departure this year, is there an open req with a dated start behind it? A backfill that exists as an intention rather than a dated req is phantom capacity in the making.
This is the first in a series walking through the six pillars of How Revenue Plans Break, our field guide to the 23 failure modes of B2B revenue plans. The full guide is free to read, with a designed PDF if you want it for your next planning review. If you would rather start with your own plan, the Plan Reality Check scores it against the four deploy-day drivers in about ten minutes, and the results are not gated.