For a decade, the operating manual for a VC or PE-backed company had one instruction: grow.

Burn multiple be damned, CAC payback be damned.

Capital was cheap, markets were expanding, and the board slide that mattered was the one with the steepest line. If you worked in Growth during that era, you know exactly what got funded: acquisition. More channels, more pipeline, more logos.

That doctrine is on life support. The money that used to reward growth-at-all-costs now asks about efficient growth, burn multiples, and net revenue retention before it asks about anything else. The math flipped. When capital is expensive, the cheapest revenue in your company is the revenue you already have.

And yet, walk into most of those same companies and look at how they actually operate. Attribution gets a dedicated analyst, a six-figure tool stack, and a weekly meeting where people argue about whether a lead was influenced by a webinar.

Retention gets a quarterly business review and a churn number that everyone stares at like a weather report. Something already happened, nobody can change it, and the meeting moves on.

Here's the uncomfortable version of that observation: acquiring a customer costs somewhere between 5 and 25 times more than retaining one, depending on whose research you trust and what industry you're in. (The famous stat comes from Frederick Reichheld's work at Bain in 1990. More recent analyses of B2B SaaS put the range at 5–10x. Pick either number; the conclusion doesn't change.) We built entire org charts, budgets, and software categories around measuring the expensive thing, and we measure the cheap thing with a rearview mirror.

This issue is about fixing that. Specifically: what leading indicators actually predict retention, which existing frameworks are worth stealing instead of reinventing, and how to wire those signals into your revenue teams so that Sales knows a high-value account is drifting before the renewal call gets weird.

Let's be precise about the problem, because "we should care more about retention" is a bumper sticker, not a strategy.

Churn rate, logo retention, gross revenue retention, net revenue retention — these are the metrics most companies use to manage retention, and every single one of them is a lagging indicator. They describe what already happened. By the time a weak NRR number shows up in a quarterly report, the retention problem it describes is months old.

The customer who churned in March stopped logging in back in December. The expansion that didn't happen died when the champion left in Q2 and nobody noticed until the renewal conversation in Q4.

Reading NRR to manage retention is like reading an obituary to manage your health. Accurate, well-formatted, and useless for intervention.

The research on churn timing makes this brutally concrete. Most churn signals appear 60 to 90 days before a customer cancels. Usage decline specifically precedes cancellation by 30 to 60 days. That's the intervention window — the period where the customer is still persuadable, the relationship is still warm, and a well-timed conversation can change the outcome. A quarterly review cadence guarantees you'll spend most of that window not looking.

Nobody would run new business this way. Imagine a sales org that only looked at closed-won and closed-lost, with no pipeline stages, no forecast, no deal reviews, no early signals of whether an opportunity was progressing or stalling. You'd fire that VP of Sales in a quarter. Yet that's precisely how most companies run the revenue they already have.

The good news: you don't need to invent retention measurement from scratch. Three frameworks already exist, they're battle-tested, and each one gets something right. The honest news: each one also has the same blind spot, which we'll get to.

The Customer Health Score. The industry standard. A composite metric, usually 0–100, that blends product usage, engagement, support history, billing signals, and relationship data into a single number per account. Done well, it's the closest thing retention has to a pipeline stage. Done badly — and it's usually done badly — it's a vanity dashboard built from whatever data was easy to pull, weighted by whoever built the spreadsheet, trusted by no one. The tell is simple: ask whether the health score has ever changed how a specific account was handled in a specific week. If the answer is no, you have a decoration, not an instrument.

Pendo's Product Engagement Score (PES). The cleanest product-side formula I've seen: the average of Adoption (how many of your core events an account actually uses), Stickiness (DAU/WAU or DAU/MAU — how often users come back), and Growth (accounts gained versus dropped). Three numbers, averaged, tracked over time. Pendo's own data shows PES correlates tightly with outcomes: the highest-scoring accounts renew and expand, the middle renews flat, and the bottom of the distribution is where churn lives. If you instrument nothing else, instrument these three.

Gainsight's DEAR framework. Deployment, Engagement, Adoption, ROI. The most enterprise-shaped of the three, and the one that asks the question the others skip: is the customer actually getting business value, and can they prove it internally?

Deployment asks whether the product is even set up correctly (an embarrassing number of churns trace back to a botched implementation nobody revisited).

Engagement asks whether the right stakeholders — including executives, not just daily users — are still at the table.

Adoption covers usage depth.

And ROI asks whether the customer could defend the line item in their own budget review.

Gainsight built DEAR explicitly to forecast NRR, and the logic holds: an account can have healthy usage and still churn if the economic buyer can't articulate what they're paying for.

Steal liberally from all three. But notice what they have in common: they were all built by and for Customer Success and Product teams, they live inside CS and product analytics tools, and in most companies the people carrying revenue targets have never seen them.

That's the gap, and it's where the actual opportunity is.

Before we get to the system, let's inventory the raw material. Across the churn research and the frameworks above, the leading indicators that reliably fire inside that 60–90 day window fall into four families.

Usage signals

The workhorses. Week-over-week active usage trend (the trend matters more than the level — a small account trending up is healthier than a big account trending down).

License utilization: if a customer bought 100 seats and 41 are active, the renewal conversation has already started without you.

Stickiness (DAU/WAU) as a habit measure. And depth of adoption — how many of your genuinely core workflows the account uses. An account using one feature isn't a customer; it's a customer-shaped churn risk with a workaround waiting to be found.

Relationship signals. The most predictive single event in the entire churn literature is champion departure — losing your internal advocate roughly triples an account's churn risk. And it's detectable: the champion's login goes quiet, their email bounces, a new admin you've never met contacts support.

Executive engagement belongs here too. If no one above the day-to-day user level has touched the relationship in two quarters, your renewal depends entirely on one person's goodwill and continued employment.

Friction signals. Support tickets are widely misread. Volume alone is ambiguous — engaged customers file tickets too. What predicts churn is the pattern: a sudden surge of similar issues, sentiment turning negative across conversations, or the same workflow failing repeatedly. A customer who files an annoyed ticket is still fighting for the product to work. The one who quietly stopped filing tickets six weeks ago already gave up.

Value signals. Time-to-value for new accounts — the longer the gap between contract signature and first real outcome, the higher the eventual churn odds, and this is measurable in week one of the relationship. And the DEAR-style ROI question: does this account have a number, a story, or an internal artifact that justifies the spend? If your product got mentioned in the customer's own board deck, you're safe. If nobody at the account could explain the line item, you're a cost-cutting exercise away from a churn.

None of these signals is exotic. Most companies already have the data. What they don't have is a system that turns the signals into action — which brings us to the framework.

Here's the doctrine: manage at-risk revenue exactly the way you manage new-business pipeline. Not "share a health score with Sales." Not "improve alignment." Run the same operating system on both sides of the revenue equation. I call it the Retention Pipeline, and it has five components — each one a deliberate mirror of something your sales org already does, which is exactly why it works. You're not asking anyone to learn a new religion. You're extending one they already practice.

1. Score every account, weighted by evidence. Build a composite score from the four signal families above. A starting weight that matches how predictive each family actually is: usage 40%, relationship 25%, friction 20%, value 15%. Then — and this is the step everyone skips — backtest it. Pull your last 20 churned accounts and check whether the score would have flagged them at 60+ days out. Tune the weights until it would have caught most of them. A health score that wouldn't have predicted your own historical churn is a guess wearing a costume.

2. Stage the risk, like a deal. Sales doesn't treat every opportunity identically, and you shouldn't treat every risk signal identically. Three stages are enough: Watch (score declining, no action beyond monitoring), Engage (score crossed a threshold or a tripwire fired — champion left, utilization dropped below 50%, sentiment turned — triggering a defined play within five business days), and Rescue (multiple signals stacked, exec-level intervention, treated with the urgency of a must-win deal).

3. Put a dollar value on every at-risk account. This is the single most effective political move in the whole system. "Account health declined" moves no one. "$340K of ARR entered Engage this week" moves everyone. Denominate risk in dollars and the conversation changes rooms — it stops being a CS metric and becomes a revenue line item that executives track.

4. Assign one owner per at-risk account. Not a team, not a dashboard everyone can see — a name. For high-value accounts, that name should frequently be the account executive or account manager who owns the commercial relationship, not just the CSM. The person who sold the expansion should feel the risk. If your comp plan pays for expansion but not for defense, your comp plan is quietly voting for churn.

5. Forecast retained revenue weekly. The keystone habit. A standing 30-minute review of at-risk ARR, run exactly like a pipeline review: what entered a risk stage, what moved between stages, what plays were run, what's the forecasted retained revenue for the quarter. Sales leadership sits in the room. This meeting is what makes retention a leading-indicator discipline instead of a quarterly autopsy — and it's how Sales finds out a high-value client is losing traction while there's still time to do something about it.

That last point deserves a beat.

The reason Sales is perpetually blindsided by churn isn't that salespeople don't care.

It's that the signals live in tools Sales never opens, denominated in units Sales doesn't think in, reviewed on a cadence Sales doesn't attend. The Retention Pipeline fixes all three at once: the score lives in the CRM next to the pipeline, the risk is denominated in dollars, and the review happens weekly with Sales in the room.

Five moves, in order of leverage.

1. Run the autopsy backtest. Pull your last 20 churned accounts and reconstruct their final six months: usage trend, champion status, ticket pattern, utilization. Write down which signals were visible 60+ days before cancellation. This is your evidence base — and your pitch deck for the rest of this list.

2. Instrument the three PES metrics this week. Core-event adoption, DAU/WAU stickiness, account quick ratio. Don't wait for the perfect composite score; three honest numbers tracked weekly beat a sophisticated model shipping next quarter.

3. Set two tripwires that page a human. Champion departure and license utilization below 50%. These two alone, wired to an alert with a named owner and a five-day SLA, will catch a meaningful share of preventable churn.

4. Put the health score in the CRM. Not in the CS tool, not in a BI dashboard — in the system Sales lives in, on the account record, next to the open opportunities. Visibility is the integration. Everything else is process.

5. Book the weekly at-risk revenue review. Thirty minutes, dollar-denominated, Sales leadership required. The first one will be awkward. By the fourth, someone will save an account in that meeting, and the habit will defend itself.

I've argued in this newsletter that the future of Growth is the revenue architect — the person who owns the full arc from stranger to expanding customer, not just the top of the funnel. Retention is where that claim gets tested. Anyone can own acquisition; the tooling is mature, the playbooks are public, and the results are legible. Owning the whole system means owning the part where the money already lives.

The era that's ending measured Growth by how fast you could pour water into the bucket. The era that's starting will measure whether you noticed the holes before the water level dropped — and whether you built a system where the people carrying the number found out in time to act.

Your churn rate is history. Your leading indicators are a choice. Make it this quarter, before the rearview mirror does it for you.