When Klaviyo launched Composer into public beta on June 30, 2026, it included a feature most consultants overlooked. Retention scoring is moving from enterprise feature to consultant baseline.
Klaviyo Composer, in public beta as of June 30, runs full audits of campaigns, flows, and segments to surface revenue opportunities. It identifies underperforming flows, lapsed high-value segments, welcome journey dropoffs. Manago AI, rebranded from SALESmanago on the same date, offers agentic analysis to identify strategic segments and next best actions from raw customer data.
The pattern is clear: tools that used to cost $50K annually and require a data engineering team are now accessible to any consultant who knows how to plug in a CRM.
This means one thing. You should have a retention score running.
The Case for Scoring Client Health
I've run a membership community since 1997. The members who stopped showing up to calls were always the ones who cancelled 60 days later. The signal was there. I just wasn't scoring it.
That's the opportunity on your desk right now. Your clients are leaving signals everywhere—in how often they schedule calls, in whether they pay invoices early or late, in whether they're asking for more scope or less, in whether they're referring others. Most consultants watch these signals disappear into a spreadsheet and never connect the dots.
A retention score solves this. It's not predictive magic. It's systematic observation.
Scoring works because it forces you to name what "healthy" looks like in your business. Once you've named it, you can measure it. Once you measure it, you can intervene. Once you intervene, you keep more clients.
The math is simple. The discipline is the hard part.
Building Your Retention Score
Here's a working model that works in a spreadsheet, in GHL, or handed to an AI agent to monitor continuously.
Each factor gets a raw score from 1 to 10. Sum them up. That's your composite.
Engagement frequency (0–10). How many substantive touchpoints happened this quarter? Meetings, deliverable reviews, email exchanges that represent actual project motion. Not admin emails. Not newsletters. Real engagement. Score 10 if 8+ touches per month. Score 1 if you haven't heard from them in 60 days.
Scope direction (0–10). Is the client asking for *more* work or *less*? Asking for more is 8–10. Asking for less, cancelling add-ons, or narrowing scope is 1–3. Flat is 5.
Payment behavior (0–10). Early payment: 9–10. On time: 7–8. 15 days late: 4–5. 30+ days late: 1–2. Late payments are flight risk signals. Cash flow clarity matters.
Referral activity (0–10). Has the client referred new business in the last 90 days? One referral that turned into a deal: 8–10. Verbal referrals with no closes: 5–6. No mentions: 1–2. Clients who sell you don't cancel.
Satisfaction signals (0–10). This is NPS if you have it, or sentiment from recent check-ins. Has the client complained about quality or cost? Have they said "we love this"? Three months of complaints: 1–3. Neutral: 5–6. Vocal satisfaction: 8–10.
Contract proximity (0–10). When does their contract end? More than 6 months out: 6–7. 90–180 days out: 5–6. 30 days or less: 3–4. Close contracts demand attention.
Add-on uptake (0–10). If you've offered them advanced services, tools, or packages, did they buy? Yes and they're using it: 8–10. Said maybe: 5. Said no: 2–3.
Client health score = sum all factors.
You get a number between 7 and 70. Call anything below 40 red zone. 40–70 is yellow. 70+ is green.
| Client | Engagement | Scope | Payment | Referrals | Satisfaction | Contract Proximity | Add-on | Score | Status | |--------|-----------|-------|---------|-----------|--------------|-------------------|--------|-------|--------| | Atlas Corp | 8 | 9 | 8 | 6 | 9 | 7 | 8 | 63 | Yellow | | Foundry Labs | 3 | 2 | 3 | 1 | 2 | 4 | 1 | 16 | Red | | Crescent Media | 9 | 8 | 9 | 9 | 10 | 8 | 9 | 62 | Yellow | | Nexus Group | 10 | 10 | 10 | 8 | 9 | 9 | 10 | 66 | Yellow | | Sterling Inc | 6 | 5 | 4 | 3 | 4 | 3 | 2 | 27 | Red |
Red clients need immediate intervention. Not a Slack message. A call. "We notice engagement has dropped. What's happening?" Honest conversation. Either you find out what's wrong and fix it, or you find out they're leaving and you get 60 days to replace that revenue.
Yellow clients are your expansion candidates. They're healthy enough to add to, not healthy enough to take for granted. A proactive check-in—"what's working, what would make this better?", often surfaces a $5K upsell or a contract extension.
Green clients you nurture for advocacy. They're your case studies, your referral partners, your annual expansion plays.
The System Requirement
Scoring only works if you run it monthly, not when you think about it.
Pick a calendar date. First Monday of the month. Last Friday of the month. Doesn't matter. What matters is consistency.
Spreadsheet scoring takes 30 minutes if you have clean data. If you have 50 clients, you're looking at a two-person-day investment the first month, then maintenance every month after. Worth it.
Or hand it to an AI agent. Give the agent read access to your CRM. Write a prompt that pulls recent emails, call notes, invoice dates, scope change requests. Have the agent recalculate scores weekly or biweekly. Output a CSV. Route red-flag clients to Slack. Move yellow clients into a specific action sequence.
Klaviyo and Manago are building this into their platforms. But you don't need a platform. You need a process.
The consultant businesses that will dominate the next two years are the ones that treat client retention as an engineering problem, not a hope problem.
When Scoring Changes the Conversation
Once you have a score, everything shifts.
Right now, you probably have a mental model of which clients are "at risk." It's based on feeling. Maybe gut instinct. Maybe that one awkward call last month. The feeling is usually right, but it's also usually late.
A scoring system catches the decline *before* the awkward call. It tells you that a usually-engaged client has gone silent for 40 days. That a client who always paid early is now 25 days late. That a contract is 120 days from renewal and no new scope has been signed.
These are system signals. They're not opinions. They're not intuitions. They're facts.
When you present a score to your team, you can say: "This client is at 34. Here's why. Here's what we need to do." You can measure whether your intervention worked. You can build a playbook: "Red clients who get contacted within 48 hours have an 80% retention rate. Red clients who wait 10 days have a 40% retention rate."
That's how you scale retention without burning out.
Due diligence is non-negotiable. That doctrine applies to client relationships too. You can't keep what you don't measure.
FAQ (See also: Manago AI agentic platform.)
Q: What if my client is a one-person shop that doesn't have many touchpoints?
One-person shops need *more* frequent touchpoints, not fewer. Score them on what matters: Did they take the meeting? Did they review the deliverable? Did they renew? A silent small-business client is a high-risk client. Track attendance, not response time.
Q: Can I score clients by revenue instead of health factors?
You can, but you shouldn't. Revenue is a lagging indicator. Health factors are leading indicators. A $100K client with a score of 25 is a higher priority than a $20K client with a score of 65. Revenue tells you what happened. Score tells you what's about to happen.
Q: How do I handle clients who are seasonal or project-based?
Adjust the engagement window. A project-based client in the off-season might naturally have zero engagement. Score them on their last project cycle and their payment history. Seasonal clients get scored during their active season. The framework works; the inputs change based on your business model.
Q: What score should I aim for as a baseline?
Aim for zero red clients. It's achievable. A 50–60 average across your book means you have healthy clients and some attention needed. If your average is below 45, you have a retention problem that predates your scoring system.
Q: Should I share my retention score with clients?
No. Use it internally only. It's a diagnostic tool for your team. Sharing it with a client becomes a negotiation. Keeping it internal lets you intervene proactively without signaling panic.
The Next Step
Your retention data is already in your CRM. Your call notes are there. Your invoice dates are there. Your contract renewals are there. You're not missing data. You're missing the model.
Build the model. Run it monthly. Take action on red clients within 48 hours. That's it.
Consultants who systematize this will have better retention. They'll know which clients to fight for and which ones are already walking out. They'll stop guessing.
Systems over slogans. Measurement over hope. That's how you keep clients.
Further reading:
- Klaviyo's Composer feature is helping brands surface revenue opportunities in their existing flows. Klaviyo Launches Composer Beta: AI-Powered Campaign Auditing shows how systematic auditing surfaces hidden churn signals.
- Manago's rebranding signals the shift from CRM-as-transaction-log to CRM-as-intelligence-engine. Manago AI Rebranding Reflects Growing Demand for Agentic Customer Intelligence breaks down what agentic analysis means for client health monitoring.
- For consultants building AI-driven client intelligence, AI Marketing Agents vs. Retainers: Consultant [consultant positioning H2 2026 H2 2026](/blog/ai-marketing-agents-vs-retainers-consultant-positioning-h2-2026/) covers how to position these capabilities with prospects.
- Client retention is part of building a Open-Source AI Marketing Suites: Consultant [consultant packaging playbook Playbook](/blog/open-source-ai-marketing-suites-consultant-packaging-playbook/) framework that scales without vendor lock-in.
*Jeff Barnes is the founder of DEMG.ai and Digital Evolution Marketing Group. He has no financial relationship with any tool, platform, or company mentioned in this article unless explicitly disclosed. DEMG.ai provides marketing education and systems for owner-operators, not investment advice. Results vary. Past performance does not guarantee future results.*