Scenario Planning for an Agency: Three Versions of Next Quarter

Most agencies plan once a year, in one version, and that version is optimistic — because it was written in a good month. When reality diverges, the plan is not adjusted; it is quietly abandoned.

Scenario planning solves a narrower problem than forecasting. It does not try to predict which version happens. It forces you to decide, while you are calm, what you would do in each — so that when a client leaves in the middle of a quarter you are executing a decision rather than improvising one.

Build three versions, not one

Base. What happens if the current client list stays roughly as it is: existing retainers continue, one or two projects land, the team stays the same size. This is not your target — it is the honest continuation of today.

Bad. Your largest client leaves. For most agencies that is the realistic worst case, and it is far more likely than a general market collapse. If one client is 35% of revenue, that is the number the scenario removes.

Good. The pipeline converts better than expected and two large projects arrive at once. This sounds like the pleasant scenario, but it is the one that breaks agencies — growth consumes cash before it produces it.

The good case is a cash problem

Agencies rarely plan for success, and it is the version that most often causes trouble. Two new projects mean hiring or contractors, both paid before the client pays. If the client terms are 30 days after delivery and delivery takes six weeks, you are financing two and a half months of work.

So the good scenario needs one extra line that the base case does not: how much cash the growth consumes at its deepest point. That number decides whether you can accept the work at all, or need a prepayment to do it — see why prepayments are not profit.

Attach a decision to each scenario

A scenario without a pre-agreed action is a spreadsheet exercise. For each version write down, in advance:

• What we stop. Which spending is paused — hiring, tools, marketing, bonuses.
• What we protect. What does not get cut even in the bad case, because cutting it destroys the agency's ability to recover.
• What triggers it. The observable signal that tells you this scenario is now the one you are in.

The trigger matters most. «Revenue falls» is not a trigger — by the time revenue falls you are already three months into it. «Our largest client does not renew by the 15th» is a trigger, and it arrives early enough to act on.

A worked example

An agency bills 1,400,000 a quarter. Its largest client is 480,000 of that. Fixed costs — payroll, rent, tools — are 900,000 a quarter.

In the base case, 1,400,000 against 900,000 leaves 500,000 before contractor costs. In the bad case, losing the large client leaves 920,000 against the same 900,000: the agency is at break-even with no room for a delayed payment. In the good case, two projects add 600,000 of revenue but require 380,000 of contractor cost paid one to two months before the client pays — a cash dip of nearly 400,000 at its worst point, even though the quarter ends well.

Three different situations, three different decisions — and none of them visible in a single annual plan.

Keep the reserve tied to the bad case

The usual advice is to hold three months of expenses. Scenario planning gives a more precise answer: the reserve needs to cover the gap in the bad case for as long as it takes to replace the client. If replacing a large client takes four months and the bad-case gap is 120,000 a month, the reserve is 480,000 — not an abstract multiple. See how much reserve an agency should hold.

Where to start

Write the bad case first, because it is the one with consequences. Remove your largest client from the next quarter, subtract fixed costs and see what is left. Then write down the three decisions and the trigger that starts them. That single page is worth more than a twelve-month forecast built in one version.

In Finmap you see revenue by client and the real cost of each project, so the scenarios start from your actual numbers rather than from memory. Try it free for 7 days.