In our experience, most agency churn is effectively decided 60 to 90 days before the contract end date — the cancellation email just makes it official. The fix isn’t hiring more charming account managers. It’s treating retention as an operations workflow with three parts: a simple client health score tracked in your CRM, quarterly business reviews triggered automatically by revenue tier, and a renewal sequence that opens 60 days before every contract ends. If you systematized onboarding for the first 30 days, this is the sequel.

Why renewals fail quietly

We see the same pattern across agencies: onboarding is a documented checklist with owners and deadlines, but renewal is a vibe. Someone notices a contract ends next month, the account lead “feels good about it,” and then the client says they’re “consolidating vendors” — which usually means they stopped seeing the connection between your invoice and their results months ago.

By the time a client tells you they’re unhappy, you’re negotiating from behind. The clients who churn rarely complain first. They go quiet: slower email replies, skipped calls, a champion who stops forwarding your reports internally. All of that is observable long before the end date — if you’re actually looking. Retention, done well, is mostly about making that looking systematic instead of intuitive.

The client health score: five signals, one number

A client health score for agencies doesn’t need to be sophisticated. It needs to be honest and cheap to maintain. If updating it takes more than ten minutes a week across your whole book, nobody will do it, and a stale score is worse than no score because it manufactures false confidence.

What to track

Score each client 1–5 on five signals, then average:

Results vs. goals. Are you hitting the outcomes agreed at kickoff — not activity, outcomes. Deliverables shipped on time with declining performance is a 2, not a 4.
Engagement. Do they show up to calls, respond within a couple of days, and give feedback on work? Silence is not satisfaction; it’s the most common precursor to churn we see.
Relationship breadth. How many people at the client know and value your work? Single-threaded accounts are fragile — one champion leaves and the account leaves with them.
Commercial behavior. Invoices paid on time, no drip of scope disputes, no sudden requests to “revisit the budget.”
Trajectory. Is scope growing, flat, or shrinking? A client who quietly cut their retainer 20% two quarters ago is telling you something.

Make it operational, not decorative

Put the score as a field in your CRM, on the client record, updated weekly by whoever owns the account. Then attach rules: anything at 3.5 or above is green; 2.5–3.4 gets discussed at your weekly ops meeting with one concrete action assigned; below 2.5 triggers a save plan with an owner and a deadline — this week, not “when things calm down.”

The score isn’t the point. The point is that a falling number forces a conversation in week 2 of a problem instead of week 20. That’s how churn actually goes down: earlier detection, earlier intervention, smaller fixes.

QBRs triggered by revenue tier, not by mood

Quarterly business reviews die in agencies for two reasons: they’re scheduled ad hoc (so they slip), and they’re report readouts (so clients stop attending). Fix both.

First, tier your book by revenue and let the tier set the cadence automatically:

Top tier (roughly your top 20% of clients): a full QBR every quarter, live, with the economic buyer in the room — not just your day-to-day contact.
Mid tier: a lighter review every six months, with a short async video summary in the off quarters.
Small retainers: a quarterly async summary — five minutes of recorded walkthrough beats a meeting nobody wanted.

Create these as recurring tasks in your CRM the day a client signs, tied to the tier field. If a QBR happens only when an account manager remembers, it happens for the loud clients and the pleasant ones, and skips exactly the quiet accounts that need it most.

Second, change what the QBR actually is. Reporting on last quarter is the first ten minutes, framed against the goals set at onboarding — dollars and business outcomes, not impressions and hours. The remaining time is forward-looking: what’s changing in their business next quarter, what that means for the engagement, and what you recommend doing differently. A QBR where the client talks more than you do is working. One where you narrate slides is a report with better production values, and it will get delegated down until it dies.

Every QBR ends with the health score conversation made explicit: “On a scale of one to five, how are we doing — and what would make it a five?” Asking directly feels uncomfortable exactly once. Not asking costs you the account.

The 60-day renewal workflow

Here’s the agency renewal process we recommend, opened automatically 60 days before every contract end date. It’s five steps, and the first one has nothing to do with the client.

Day 60: internal account review

Before any client conversation, decide what you want. Pull the health score history, margin on the account, and scope trajectory, then pick a lane: renew as-is, renew and expand, renew and reprice, or let it sunset. Agencies skip this step constantly and end up re-signing unprofitable, low-health accounts out of momentum — which is its own kind of churn problem, just slower.

Day 50: the value recap

Build a one-page recap of the engagement: goals from kickoff, what was delivered, results in the client’s business terms, and what you’d prioritize next. This document exists because your champion has to re-sell you internally, and “they’re great to work with” doesn’t survive a budget meeting. Give them ammunition.

Day 45: the renewal conversation

Book a dedicated call — not bolted onto a status meeting, and not disguised as a QBR. Walk the recap, then propose next term’s plan. If the health score is below 3, this call happens even earlier and starts with listening, not proposing.

Day 30: proposal in writing

Send the renewal proposal with any pricing or scope changes clearly flagged. A price increase surprised into a client at day 10 reads as a squeeze; the same increase at day 30, justified by the recap, reads as a business decision.

Day 14: escalation

Unsigned at two weeks out is a yellow flag regardless of what anyone “feels.” It escalates to a founder or director touch — a direct, human conversation about what’s holding things up. Sometimes it’s procurement. Sometimes it’s the real objection nobody voiced on the day-45 call. You want to hear it while you can still respond.

All five steps live as a task sequence in your CRM, generated from the contract end date. If your stack can’t trigger a workflow off a date field, that’s a tooling gap worth closing — it’s precisely the kind of sequence we built OpenAva to run alongside client onboarding and delivery.

The mistakes that undo all of this

A few failure modes we see repeatedly: health scores with twelve weighted inputs that nobody updates after week three — five signals, weekly, or don’t bother. QBRs that are backward-looking slide decks. Renewal conversations that start when the client raises them, which means you’ve already lost control of the timeline. And treating a quiet client as a happy one, when quiet is, in our experience, the most reliable churn signal in an agency’s book.

None of this requires new headcount. It requires deciding that retention is a process with owners and triggers, the same way you decided onboarding was.

Start with the score

You don’t need to build all three pieces this quarter. Start by scoring every current client 1–5 on the five signals this week — it takes an hour and it will immediately surface two or three accounts that need attention before their renewal date does. Then wire the 60-day sequence into your CRM for the next contract coming up. The QBR cadence can follow once the first two are running. Retention compounds; so does starting.