An agency’s capacity problem is usually visible weeks before the deadline it breaks, in signals nobody is asked to track. This post gives a weekly check for those signals, and shows how capacity was added in one real case: a US agency whose backlog for a multi-location healthcare client had started to threaten the relationship.
Why the problem stays invisible
The early symptoms look like normal agency life. A senior person covers two roles for a sprint because someone is on leave. A project runs a week late because the brief changed. Each has a reasonable explanation, and reasonable explanations do not trigger alarms.
Underneath, the buffer is being spent: the informal slack that lets a team absorb one bad week without anyone downstream noticing. Every stretched sprint uses a little, and it is rarely replenished, because replenishing it means staffing below full use, which looks like waste on a spreadsheet.
The breaking point is usually two deadlines in the same week that would each have been fine alone. From outside it looks sudden. From inside, it is the first week the deficit became visible, not the first week it existed.
- 01A covered roleOne senior covers two roles for a sprint
- 02A changed briefA project runs a week late
- 03No alarmEach has a reasonable explanation
- 04Buffer spentEvery stretched sprint uses a little slack
- 05Never replenishedStaffing below full use looks like waste
- 06Two deadlinesThe same week, each fine alone
- The deficit showsSudden from outside, but not the week it began
A weekly check for the signals
The fix is to make the buffer a named number at team level, not slack hidden inside padded task estimates. Then count, every Friday, the five things that spend it.
| Signal | How to count it each week | What it tells you |
|---|---|---|
| Unplanned hours | Hours spent on work that was not in the plan on Monday | How fast the buffer is being used |
| Doubled-up roles | Number of people covering a second role this week | Where the next slip will come from |
| Queue in against queue out | Items added to the backlog against items closed | Whether the backlog is a queue or a growing risk |
| Unsized promises | Items given a date before anyone estimated them | Dates that capacity cannot back |
| Waiting on approval | Items finished but waiting on a client reviewer | Whether delivery or approval is the real limit |
None of these needs a target set by someone else. Record them for a quarter and your own normal emerges. If the buffer is never touched, it is oversized. If it is gone before the quarter ends, either it needs to grow or something upstream is creating more unplanned work than the team is built to absorb.
The last row is easy to miss. In regulated work especially, finished work waiting for sign-off looks like a capacity problem when it is an approval one, and adding people will not move it.
How capacity was added in one engagement
The agency in our healthcare delivery case study had the problem every growing agency eventually meets: it was winning work faster than it could deliver it. Deadlines were slipping, the client was asking harder questions each month, and the relationship that anchored its revenue was cooling.
The usual fixes were slow or risky. Hiring takes months and adds fixed cost the moment the backlog clears. Freelancers fragment quality and cannot be shown to a client as a team. So the agency handed us the queue, not a project: a dedicated pod under its brand, working in its tracker and templates and joining its standups.
The order of work is the part worth copying:
- Triage the whole backlog into quick wins, structural work, and things that should never have been promised.
- Rebuild the client-facing schedule around what can actually ship.
- Size work before it is promised, and run sprints at a rhythm the agency can sell against.
- Move quality review upstream, so the agency’s producers stop being the last check.
- Once the queue is under control, fix the intake and sequencing that let it build up.
We rebuilt the schedule after triage and before adding speed: credibility first, heroics second. The last step mattered as much as the first. Clearing the backlog and stopping there was what the brief literally asked for, and it would have set up the same conversation six months later.
The cost of adding capacity late
Caught early, adding capacity is routine: overflow help for a defined stretch, at a planned cost, briefed with time to spare. The end client sees nothing; the deadline holds.
Caught late, the same fix costs more in every direction. Help is urgent, so fewer people are available. Briefing happens under pressure, when handoff mistakes are most likely. And the relationship takes a hit even if the date is saved, because the scramble shows.
There is a ramp even when it goes well. In the healthcare engagement, the first fortnight was slower than either side wanted while the pod learned the agency’s tools and the client’s history.
That ramp is why agencies that use a partner bench tend to put the paperwork in place before the crunch. On our capacity and delivery page, teams with an NDA and rate card already on file go straight to staffing. The ramp still happens; the paperwork just stops adding to it.
When adding capacity is the wrong fix
More people is not always the answer, and a partner who says it is should be questioned. Look elsewhere when:
- The limit is approval, not delivery. Finished work waiting for a client’s compliance or legal reviewer will not move faster with a bigger team.
- The demand is permanent and predictable. Sustained growth is a hiring decision, made calmly, not an overflow arrangement.
- Nobody on your side can review what a partner ships. Extra output nobody checks moves the risk rather than removing it.
If you do bring a partner in, what to ask before signing covers the checks worth making first.
Written by
Parvessh Agarwal
Founder and Chief Executive, Vinove
Parvessh founded Vinove in 2004 with one idea: build focused technology companies that earn their place in real markets. His focus is on the long game, on companies that survive and serve real customers long after launch.
LinkedIn →
