When to fire your first agency client (and the playbook for doing it)
Most agency founders carry one client too long who's destroying margin and morale. The signals that say it's time to fire them, and the script that keeps the relationship intact.
Every agency founder has the same realisation between months 12-24: there's one client they should have fired six months ago. The math reveals it before the relationship does — that client is at 10-15% gross margin while the rest of the book is at 35-45%. The discipline is recognising the signals early and executing the firing without damaging the brand.
The five signals it's time to fire
- Margin under 20% for two consecutive quarters. The client is either underpriced, scope-creeping, or both. Fix the price or fire the client; doing neither is the most expensive option.
- They consume 2x+ the support time of a comparable account. Their inbound emails, Slack pings, and weekly meetings displace work for higher-margin clients.
- They've fired every other agency before you. They'll fire you too; the question is just whether you control the timing.
- Your best people don't want to work on the account. Watch where senior team members volunteer and where they avoid. The avoidance signal is real.
- Strategic creep without a new SOW. You started doing one thing; you're now doing four. Each new thing was a "small favour." Total: half a senior FTE you're not billing for.
The script that keeps the door open
"We've grown into a different specialty and I don't think we're the right team for what you need anymore. Over the next 60 days we'll wind down the engagement professionally — full handoff doc, recommended replacement agencies, no charge for transition work. I want to make sure you land somewhere good."
What this script does:
- Avoids blaming the client (no "you're hard to work with").
- Frames it as your evolution, not their problem.
- Offers a generous offramp (60 days, handoff, recommendations).
- Closes with care, not animosity.
What this script doesn't do:
- Apologise excessively. Don't.
- Leave the door open to "maybe we can fix it." The decision is made.
Weekly digest
Get new resources like this, weekly.
One email a week: new hubs, new tools, and the editorial pieces worth reading. One click to unsubscribe.