Audit Your Current Client Roster to Find Your Real ICP

You are an agency P&L analyst. You read between the lines of a client roster to find the ICP the agency is actually winning with — which is often different from the ICP they think they have.

Agency: [AGENCY_NAME]
Services: [SERVICES]
Full client roster with these fields per client: industry, size, services purchased, MRR, gross margin, tenure in months, NPS or sentiment, source (referral/outbound/inbound/partner): [CURRENT_CLIENT_LIST]
Stated ICP today: [STATED_ICP]

1. Segment the roster into clusters based on industry, size, and services purchased.
2. For each cluster, compute average MRR, average margin, average tenure, and a fit-score.
3. Identify the top 2 "hidden winners" — clusters that quietly outperform the stated ICP on margin x tenure.
4. Identify the top 2 "hidden losers" — clusters that look like wins on revenue but lose on margin or tenure.
5. Recommend a revised ICP statement based on the hidden winners.
6. Flag any single client whose departure would meaningfully change the picture (concentration risk).

- Margin and tenure trump revenue. A $10k/mo client at 15% margin who churns in 6 months loses to a $4k/mo client at 60% margin who stays 30 months.
- Don't romanticize the founder's favorite clients. Go where the numbers go.
- Call out where the data is too thin to draw conclusions.

Return:
1. Cluster table (cluster | n | avg MRR | avg margin | avg tenure | fit score)
2. Top 2 hidden winners (with the pattern that defines them)
3. Top 2 hidden losers (with the pattern + recommendation: fire, reprice, or downsize)
4. Revised ICP statement (1 paragraph)
5. Concentration risk callout
6. 3 changes to make in the next 30 days

• [AGENCY_NAME] — Your agency's name
• [SERVICES] — Services you sell
• [CURRENT_CLIENT_LIST] — Paste client data with industry, size, services, MRR, margin, tenure, NPS, source
• [STATED_ICP] — Your current public/internal ICP statement

AGENCY_NAME: Forge Media. SERVICES: SEO, content, paid search. STATED_ICP: B2B SaaS Series A-B, $5-20M ARR. CURRENT_CLIENT_LIST: 14 clients — 6 are B2B SaaS ($4-9k MRR, 30-45% margin, 8-14mo tenure), 4 are professional services firms (law/accounting, $3-5k MRR, 65% margin, 22-36mo tenure), 3 are e-com DTC ($6-12k MRR, 20% margin, 5-9mo tenure), 1 is a marketplace ($14k MRR, 50% margin, 28mo tenure).

| Cluster | n | Avg MRR | Avg Margin | Avg Tenure | Fit |
|---|---|---|---|---|---|
| B2B SaaS | 6 | $6.2k | 38% | 11mo | 6/10 |
| Pro services | 4 | $4.1k | 65% | 28mo | 9/10 |
| E-com DTC | 3 | $9k | 20% | 7mo | 3/10 |
| Marketplace | 1 | $14k | 50% | 28mo | n/a |

**Hidden winners:** Professional services firms. Lower revenue, but 65% margin and 28mo tenure means LTV is ~2x your stated-ICP B2B SaaS clients. Pattern: 10-40 employee law/accounting firms outside top-50 metros.

**Hidden losers:** E-com DTC. Looks shiny on MRR but 20% margins and churns under a year. Recommendation: stop selling to them; refer out.

**Revised ICP:** Professional services firms (legal, accounting, consulting), 10-40 employees, in secondary US metros, who need consistent organic lead flow.

**Concentration risk:** Marketplace client = 16% of revenue alone. Diversify.

**30-day changes:** (1) Rewrite homepage for pro services. (2) Sunset DTC pitches. (3) Ask each pro services client for 2 referrals.

• Pull the data from your accounting tool, not your CRM — your CRM lies about margin.
• Include NPS or a 1-5 sentiment guess. Tenure alone misses unhappy clients who just haven't churned yet.
• If your hidden winner is in a niche you'd be embarrassed to specialize in, that's still a signal. Specialize anyway.