A customer who cancels rarely decided that morning. The cancellation is the visible end of a quiet process that started weeks or months earlier, a feature that never got adopted, a champion who left, an outcome that never quite materialised. Churn management is the work of catching that drift early; customer success is the discipline of preventing it by making sure each customer actually gets the result they signed up for. Treat the cancellation itself as the problem and you are forever too late.

The quick version

  • Churn is the rate at which customers (or their revenue) leave. Customer success is the proactive practice of driving customers to their desired outcome so they don't leave in the first place.
  • Churn is a lagging indicator, it tells you the damage is done. The job is to manage the leading indicators (onboarding completed, the product actually used, value visibly delivered) that predict it.
  • Two scoreboards keep you honest: retention rate (how many you keep) and net revenue retention (whether your existing base grows after churn and expansion).
  • The trap is reactive "saves" at renewal. Discounting a customer back from the brink treats the symptom; the cure is delivering the outcome months earlier, when there was still time.

The idea in depth: churn is a symptom, not the disease

Start with the economics, because they are what turn retention from a nicety into a priority. The financial case was made decades ago by Frederick Reichheld and W. Earl Sasser in "Zero Defections: Quality Comes to Services" (Harvard Business Review, 1990). Their argument, drawn from Bain & Company's analysis of customers over their full purchasing life, was that defection is the most telling quality measure a service business has, and that retaining just 5% more customers could lift profits dramatically, by Bain's account "by almost 100%". The exact multiplier swings hard by industry and is easy to overstate, so hold it as a direction of travel, not a constant: a retained customer is cheap to serve, buys more over time, and the savings compound.

So the move is to stop measuring quality only at the point of sale and start measuring it at the point of defection. When a customer leaves, the reason is data, not a closed door, Reichheld and Sasser's sharpest point was that defection rates are "a guide to achieving" quality, because the people walking out tell you precisely where you are falling short. Run a real exit conversation on every churned account and route what you learn back into onboarding and the product, not into a save offer.

The honest limitation: the "5% retention, ~100% profit" figure is a famous headline that has been stretched far beyond its original credit-card and service-industry context. Reichheld and Sasser were careful; a lot of the slides that quote them are not. Use the principle, retention compounds, and be sceptical of any precise number lifted from it without your own data behind it.

Why "customer success" exists: the success gap

Customer success as a named discipline is young. It crystallised inside subscription-software companies in the 2000s, when leaders noticed that customers extracting the most value from a product were the least likely to leave, and that nobody owned making that happen. The definitive practitioner text is Nick Mehta, Dan Steinman and Lincoln Murphy's Customer Success (Wiley, 2016), the field's foundational book. Its premise is plain: in a recurring-revenue business the sale is the beginning, not the end, and someone must be accountable for the customer actually succeeding afterwards.

The most useful idea inside that body of work is Lincoln Murphy's "success gap", the space between what your product does and the outcome the customer actually wanted. A scheduling tool sends shifts; the customer wanted fewer no-shows and calmer mornings. If using the product correctly still leaves them short of their goal, they churn even though "the software worked." Most preventable churn lives in that gap.

Customers don't buy your product. They buy a better version of themselves, and they churn when they don't get it.

Which points at the work itself. For each customer segment, define the outcome they are actually buying, then design onboarding and check-ins to close the gap toward it, not just to teach features. In a first 1:1 with a new account, the question is not "shall I show you how the dashboard works?" It is "what does success look like for you in 90 days, and how will we both know we got there?" Write that down and review against it. The limitation worth naming: customer success is widely practised but thinly evidenced as an academic field, much of its canon is well-argued practitioner experience, not peer-reviewed study, so treat its frameworks as sharp lenses rather than proven laws.

flowchart LR
  A(["Sale closed"]) --> B(["Onboarding
first value reached?"])
  B -->|"Yes"| C(["Adoption
used in real workflow"])
  B -->|"No"| X(["Success gap
silent drift toward churn"])
  C --> D(["Outcome delivered
customer hits their goal"])
  C -.->|"usage stalls"| X
  D --> E(["Renewal + expansion"])
  X --> F(["Churn
the lagging symptom"])
Churn is the last box, but it is decided in the early ones, the success gap opens long before the cancellation. Leaders Loop

The metrics that keep you honest

Churn management lives or dies on measuring the right things, and the temptation is to watch only the lagging number. Start with the basics. Customer churn is the share of customers who leave in a period; revenue churn is the share of recurring revenue lost, and the two diverge sharply, because losing one big account hurts revenue more than losing five small ones. Retention rate is simply the inverse: who stayed. And net revenue retention (NRR) measures what last year's customers are worth today after churn, downgrades and expansion, above 100% means your base grows on its own, which Gainsight's Nick Mehta, in his interview with McKinsey, calls one of the biggest drivers of value in subscription businesses.

The benchmarks are worth stating plainly, because churn compounds in a way that fools intuition. Per ChartMogul's benchmark analysis of its subscription-business customer base, the best companies target under 2% customer churn per month, and a seemingly modest 5% monthly churn compounds to roughly 46% lost over a year, almost half the base gone if nothing intervenes. Churn also varies hugely by who you sell to: ChartMogul's data shows monthly churn far higher for low-priced, small-business accounts (above 6% for accounts under $25/month) than for high-value enterprise ones (around 2% above $500/month). One blended churn number flatters you. Segment it.

The practical shift is to stop steering by churn alone, it only confirms losses after they happen, and to instrument the leading indicators that predict it. A simple "health" view per account (Did they finish onboarding? Are they using the product weekly? Is the original champion still here? Have they hit a value milestone?) turns a renewal surprise into a problem you can see ninety days out, while you can still fix it. The limitation: health scores are only as good as the signals behind them, and a tidy green dashboard can hide a quietly unhappy customer, so pair the numbers with actual conversations, not instead of them.

A worked example

Take a mid-sized software firm, call it Cadence, selling a team-scheduling tool to hospitality groups. (Illustrative figures throughout; this is a teaching example, not real accounts.) A regional café chain signs up: 30 seats at, say, an illustrative £35/seat/month, roughly £12,600 a year. The deal closes, the salesperson moves on, and the account lands with whoever answers the support queue.

Eleven months later, the renewal is in doubt, and nobody saw it coming, because the only number anyone watched was "are they still paying?" Run the leading indicators backward and the story is obvious: onboarding stalled at week two, only one of four sites ever logged in regularly, and the operations manager who championed the purchase left in month five. The success gap never closed. The chain bought "fewer no-shows and calmer Monday mornings"; what they got was a login they mostly ignored.

flowchart TD
  A(["Renewal at risk
~£12.6k/yr (illustrative)"]) --> B{"Why? Check the
leading indicators"}
  B --> C(["Onboarding stalled
at week 2"])
  B --> D(["1 of 4 sites
actually using it"])
  B --> E(["Champion left
in month 5"])
  C --> F(["Fix: re-onboard to the
outcome, find a new champion"])
  D --> F
  E --> F
  F --> G(["Value visible →
renewal saved, room to expand"])
The renewal looked like a sudden risk; the indicators show it was decided months earlier. Leaders Loop

The fix is not a renewal discount. It is to re-onboard against the outcome the chain actually wanted, set up the no-show report, get all four sites live, recruit a new internal champion, and show the operations director the hours saved at month-end. Done early enough, that account renews and has somewhere to grow. The deeper point: had Cadence watched onboarding completion and weekly usage from week one, the intervention would have happened in month two, cost a fraction of the effort, and never felt like a rescue. Churn management is mostly the discipline of moving that intervention earlier.

Frequently asked questions

What's the difference between customer success and customer support?

Support is reactive: the customer has a problem, raises a ticket, you resolve it. Customer success is proactive: you reach out before there is a problem, because you can see from usage that the customer is drifting from their goal. Support keeps the product working; customer success keeps the customer succeeding. Many companies run both, and confusing the two is how proactive retention work quietly gets starved.

Isn't some churn just unavoidable?

Yes, and naming which kind matters. A customer who goes out of business or genuinely outgrows you is good churn you can't prevent. Involuntary churn (a failed credit-card payment) is a billing fix, not a relationship failure. The churn worth obsessing over is avoidable churn: customers who would have stayed if they had reached their outcome. Separate the three before you panic about a headline number.

How early can you actually predict churn?

Often within the first 30–90 days. The strongest early signals are whether a customer completes onboarding and reaches first value, and whether usage becomes a habit rather than a trial. An account that never embeds the product in a real workflow is at risk long before the contract ends, which is exactly why leading indicators beat waiting for the renewal date to arrive.

Should customer success carry a revenue target?

It's contested. Tying customer success to expansion revenue (renewals and upsell) aligns it with the business and is common practice; the risk is that a quota turns a trusted adviser into another salesperson and erodes the relationship that made them effective. A reasonable middle ground: hold customer success accountable for retention and outcomes, and let expansion follow as the earned result rather than the daily pressure.

What's a good churn rate to aim for?

It depends entirely on who you sell to. For subscription software, the best performers keep monthly customer churn under roughly 2%, but small-business-focused products churn far higher than enterprise ones simply because small customers fail and switch more. Compare yourself to companies with a similar customer profile and price point, not to a best-in-class figure from a different market, and watch the trend more than the absolute.

Related in the Toolkit

Churn is shaped long before anyone signs, by the motion you sell through (GTM strategy & motions, which determines whether onboarding is self-serve or high-touch) and by how well you read what the customer is really buying (customer needs & latent needs, the skill that lets you close the success gap).

Where to go next