Why your renewals keep surprising you

It is the forecast call at the start of the quarter. A $90k account has been green in the CRM all year. The account manager says the relationship is strong; they spoke to the champion three weeks ago and it was friendly.

Then an email arrives. A new VP of Operations has started, and they are "reviewing all vendors before renewals". Six weeks later the account is gone, and the post-mortem finds what always turns up: usage had been falling since spring, the same integration bug had been raised four times, and the champion had quietly moved to another team in July.

Nobody missed the signs on purpose. The signs were simply never in the same place.

Where the warning signs actually live

Churn almost never starts in the CRM. By the time it shows up there, as a closed-lost renewal or a "risk" flag typed in a hurry, the decision has usually been made. The early signals sit in four other places:

1. Product usage. Fewer weekly active users, a key feature nobody opens any more, a team that never finished onboarding, logins that fell off after the first 90 days.
2. Support. The same issue raised three times, a jump in urgent tickets, an escalation that took two weeks to close. Silence can be a signal too: an account that used to ask a lot and has stopped asking.
3. Billing. A late payment, a downgrade request, a seat count that stopped growing, a procurement contact asking for "the contract terms".
4. People. Your champion changed role or left, a new executive arrived, the team that bought you was reorganised, or the economic buyer on the contract no longer works there.

Each of these has an owner, and each owner sees one piece. Customer success sees tickets. Finance sees invoices. Product sees usage dashboards. Sales sees the CRM and the last call. Nobody is paid to read all four together, so nobody does until the renewal date forces it.

Why it keeps happening

It is tempting to blame the account manager. In practice, the problem is structural, and it has three parts.

The data is split by department. Usage lives in the product database or an analytics tool, tickets in Zendesk or Intercom, invoices in Stripe or the finance system, and contacts in Salesforce or HubSpot. They rarely share an account ID that matches cleanly, so even a willing analyst needs a day to join them for one account.

The timing is wrong. Most teams start renewal preparation 30 to 60 days before the date. By then a new executive has already formed a view, the budget has been set, and a competitor may already be in a trial. The window to fix a real problem, such as a stalled rollout, is months, not weeks.

Health scores are built to be reported, not acted on. Many teams have a score. It is a number on a dashboard, updated monthly, with no reasons attached. "Health: 42" tells an account manager nothing they can do this week. So they ignore it, and trust their last conversation instead.

What it really costs

Surprise churn is expensive in ways that do not show up in a single line of the P&L. Take a team that renews 120 accounts a year at an average of $40k, so $4.8M of annual recurring revenue up for renewal.

• If 12% of that churns, that is $576k lost each year.
• Suppose a third of those losses had visible warning signs four months out, and the team could save half of the ones it sees in time. That is about $96k a year kept, from accounts that were already yours.
• Then add the hidden costs: the forecast miss that makes the board question every number, the scramble hours in the last six weeks, and the expansion you never pitched because nobody knew a new team was onboarding.

Swap in your own numbers. The point is not the exact figure; it is that the cost of seeing churn late is usually far larger than the cost of building the view that sees it early.

What a renewal-ready team does differently

Teams that renew well do not have better account managers. They have a better view of each account, early enough to act on it. Specifically, they have four things:

• One health score per account, with the reasons shown. Not "42", but "Usage down 30% over 60 days, two open escalations, champion moved team in July." A reason is something an owner can act on. A number is not.
• A renewal brief 90 days out. One page: what the customer bought, what they actually use, the value delivered, open issues, who the stakeholders are now, and the risks to deal with before anyone talks about price.
• Alerts when something changes, sent to the account owner in Slack or email with the reason and a suggested next step, not buried in a dashboard nobody opens.
• Expansion signals in the same view. The data that shows risk also shows growth: a new department onboarding, usage at the edge of the plan, hiring in the team you serve.

How to fix it: a step-by-step playbook

You can build most of this with the systems you already have. The order matters more than the tooling.

1. Join the data on one account ID

Owner: RevOps. Done when: for any account, you can pull usage, tickets, billing and contacts in one query.

This is the unglamorous step that makes everything else possible. Pick the CRM account ID as the master key, and map the product workspace ID, the support organisation and the billing customer to it. Expect 5 to 15% of accounts to need manual matching the first time.

2. Agree what "at risk" means

Owner: Head of CS with the CRO. Done when: a written rubric that two people would apply the same way.

Start simple. A rubric with five signals beats a model with fifty that nobody trusts. A good first version:

One high signal makes an account "watch". Two makes it "at risk". Write down who acts on each.

Not every at-risk account deserves the same response. Plot risk against renewal value, and the weekly review writes itself: the top right gets an executive sponsor this week, the top left gets a repeatable save playbook, and the bottom right is where you look for expansion.

3. Send the right alert to the right person

Owner: RevOps. Done when: the account owner gets a message within a day of a change.

A good alert reads like a colleague tapping you on the shoulder: the account, the change, the reason and a suggested next step. Send it where the owner already works, and write the reason to the CRM so it is there in the forecast call.

4. Generate the renewal brief at day 90

Owner: the account owner, with the brief drafted for them. Done when: every renewal over a set value has a brief before the first renewal conversation.

The brief has six sections: what they bought, what they use, value delivered (in their numbers, not yours), open issues, current stakeholders, and risks to resolve before pricing. The account owner edits it; they do not start from a blank page.

5. Review it weekly, and tune it quarterly

Owner: Head of CS. Done when: a 20-minute weekly review of at-risk accounts, and a quarterly look at which signals actually predicted churn.

After two quarters you will know which signals matter for your customers, and the rubric will get sharper.

Where AI agents help, and where people decide

This is work AI agents are well suited to, because most of it is reading, combining and drafting. An agent can check usage, tickets, billing and public news every day, apply your rubric, explain the reasons in plain language, and draft the renewal brief for the account owner to check. It can notice that a champion's LinkedIn title changed, or that the same bug appeared in three tickets with different wording.

What it should not do is talk to the customer on its own. A renewal is a relationship, and the moment an automated email lands at the wrong time, trust drops. The agent prepares and flags; a person decides what to say, and when. Every alert and brief should be logged, with the sources it used, and every change the agent writes to the CRM should be visible to the owner.

Where to start this week

You do not need a customer success platform to begin. Three steps, with nothing to buy:

1. List your renewals for the next two quarters, with value, owner and date.
2. For the top 20 by value, pull the four signals into one sheet, even by hand. It will take an afternoon, and it will almost certainly surface at least one account you thought was safe.
3. Agree the "at risk" rubric above with your CS lead, and book a 20-minute weekly review.

Once you have done it by hand for a month, you will know exactly which parts are worth automating. That is the work our Core program, with a customer retention focus, covers: a health score with reasons, renewal briefs and churn-risk alerts, built in your CRM and tools, so the next renewal is a conversation you are ready for, not a surprise.