In 2024, Marcus Webb’s automation agency had one client representing 60% of revenue. They paid $18,000/month. They demanded 24/7 Slack availability. They rewrote his team’s processes weekly. They were, objectively, killing the business.
Webb fired them in January 2025. By December, his agency had doubled revenue—to $47,000/month—with 12 healthier clients replacing the one tumor.
This is the story of learning that revenue concentration isn’t just a risk. It’s often a pathology.
The Whale Trap
Webb’s agency—FlowState—built n8n and Make.com automations for e-commerce companies. The problem client—let’s call them GiantRetail—came through a warm intro in 2023. They needed complex inventory management across 12 warehouses.
The initial project was $45,000. Successful. GiantRetail asked for a retainer: $18,000/month for “ongoing optimization and support.”
Webb should have said no. Here’s why he said yes:
The psychological traps:
- Anchoring bias: $18K/month made other clients’ $3K retainers feel small
- Loss aversion: Fear of losing the revenue outweighed fear of losing sanity
- Status association: “We work with GiantRetail” impressed prospects
- Sunk cost fallacy: Already invested in learning their complex systems
The operational reality by month 6:
| Metric | Healthy Clients | GiantRetail |
|---|---|---|
| Revenue per client | $3,200 avg | $18,000 |
| Hours required/month | 15 | 80 |
| Team members dedicated | 0.5 | 3 |
| Revision requests | 2/month | 15/month |
| Weekend emergencies | 0 | 4/month |
| Payment terms | Net 15 | Net 45 (often 60) |
GiantRetail was consuming 60% of capacity for 40% of effective revenue when accounting for payment delays and revision scope creep.
The Breaking Point
December 2024. Webb’s lead developer quit. The reason, delivered in an exit interview: “I can’t do another Christmas Eve deployment because GiantRetail ‘forgot’ to approve staging for 3 weeks.”
Webb reviewed the Slack history. 847 messages from GiantRetail in the past 30 days. 12 from their next-largest client. The ratio was obscene.
He did the math he should have done 12 months earlier:
If GiantRetail disappeared tomorrow:
- Immediate revenue loss: $18,000/month
- Freed capacity: 80 hours + 3 team members
- Potential new clients at $3,200/month: 25 (with existing pipeline)
- Actual realistic new clients in 90 days: 6–8
- Revenue in 90 days without GiantRetail: $19,200–$25,600 (from 6–8 new + existing smaller clients)
- Revenue in 90 days with GiantRetail: $18,000 (minus 1–2 team members who’d quit)
The “whale” was actually a boat anchor.
The Firing
Webb drafted a professional termination letter. 30 days notice. Offer to transition to another agency. No penalty for early contract end.
GiantRetail’s response: shock, then anger, then negotiation. They offered $25,000/month. Then $30,000. Then threatened legal action over “breach of implied exclusivity” (not in the contract).
Webb held firm. The transition was messy. GiantRetail badmouthed him to mutual contacts. The “status association” became status contamination.
Then, relief.
The Reallocation
January 2025: Webb’s team had 80 hours/month suddenly available. He’d expected a revenue crater. Instead, something unexpected happened.
The freed capacity created optionality:
- His developer who’d quit agreed to return (condition: no client over 30% of revenue)
- Two prospects who’d been “considering” for months suddenly got proposals within 48 hours
- Webb himself had time to build a productized service—“E-commerce Automation in a Box”—that sold for $5,000 flat fee
By March, he’d signed 5 new clients. By June, 8 more. By December, 12 total—none over $4,500/month, most at $3,200.
The new math:
| Metric | 2024 (With Whale) | 2025 (Distributed) |
|---|---|---|
| Total MRR | $30,000 | $47,000 |
| Largest client % | 60% | 12% |
| Team satisfaction | 4/10 | 8/10 |
| Webb’s hours/week | 55 | 35 |
| Payment delays | $12,000 outstanding | $0 |
| Weekend work | 3–4/month | 0 |
The Specific Lesson
Webb now teaches this framework to agency owners:
The Client Health Scorecard
Rate every client 1–5 on:
- Revenue per hour required
- Payment reliability
- Respect for boundaries
- Strategic value (learning, network, case study)
- Team morale impact
Any client scoring under 3.5 on average gets a 90-day improvement plan or termination. Any client over 40% of revenue gets automatic diversification mandate—find replacements before you need them.
“We don’t fire clients because we’re difficult,” Webb says. “We fire them because we’re building a sustainable business, not a hostage situation.”
The Current Reality
FlowState has 14 clients as of April 2026. Largest is 11% of revenue. Webb works 30 hours weekly, mostly on strategy and team development. He’s writing a book: The Anti-Whale: Building Agencies That Don’t Depend on Anyone.
He keeps a framed screenshot in his office: the Slack notification from GiantRetail’s CEO, sent at 11:47 PM on Christmas Eve 2024, demanding “URGENT: inventory sync broken.”
Beneath it, a sticky note in his handwriting: “Never again.”
Kivora documents agency economics and the founders fixing broken client models. Join the newsletter — Subscribe here.