To reduce churn, find out why customers leave and act before they do. Track early warning signs such as low logins or short sessions, get new customers to their first win fast, stay in close touch for the first 90 days, segment customers and make relevant offers, run a feedback loop, and grow accounts through upgrades. Make cancelling easy and legal, and win people back with value, not friction.
Key takeaways
- Aim beyond low churn: net negative churn means expansion from existing customers outweighs what you lose.
- The four key tips: understand the formula, upsell and cross-sell, expand your resources, and use your customer data.
- Groove cut churn by 71% by spotting users with short early sessions and offering them help.
- Nine best practices: define success, dashboards, onboarding and offboarding, a fast first 'aha', high touch for 90 days, segmentation, a feedback loop, a 360-degree view and real-time alerts.
- By 2026, hard-to-cancel flows have cost Amazon $2.5 billion, and Australia's easy-cancellation rules start on 1 July 2027.
I'm going to start this chapter with three words.
If you can achieve what they describe, your business can build a moat the size of the Atlantic Ocean.
What are they?
Net. Negative. Churn.
Churn is one of the biggest killers of subscription businesses. It doesn't matter how good your sales and marketing are. If you can't keep your subscribers, your business will fail.
Good businesses work to lower churn as much as possible. Great businesses aim for net negative churn.
Shopify is a great example. When I wrote the book, it was consistently reporting monthly revenue retention above 100%. It was making more from the customers who stayed than it lost from the ones who left.
SendGrid, the email API company, is another. Its net dollar retention was 112% in 2015, 111% in 2016 and a staggering 117% in 2017.
That's what you're shooting for. Not just less churn, but retention so strong that the churn you do have doesn't dent your revenue.
How do you reduce churn? The short answer
If you only read one section, read this. To reduce churn:
- Measure it properly, including expansion, so you know whether you're winning.
- Find the early warning signs in your own data, like Groove did, and act on them fast.
- Get every new customer to their first win quickly, and stay close for the first 90 days.
- Grow the accounts you keep with upgrades, cross-sells and more of what they already use.
- Reward loyalty in a modern way: tiers, personalization and community, not a dusty points card.
- Listen, act and tell customers what you changed.
- Make leaving easy and win people back with value. Friction is now a legal risk.
The rest of this page shows you how, with the stories behind each one.
What is net negative churn, and why is it so appealing?
Net negative churn happens when the new revenue you earn from existing customers is more than the revenue you lose from cancellations and downgrades. You're making more than you're losing.
Expansion revenue is the money you make by selling more to existing customers: a move to a more expensive tier, extra usage, or a related product.
Put the two together and your business becomes far more valuable, because it grows even before you add a single new customer. My page on negative churn goes deeper.
The four key tips for reducing churn
Tip #1: Understand the net negative churn formula
You can't improve what you don't measure. Here's the formula, using monthly recurring revenue (MRR):
Net revenue churn = (Churned MRR + Contraction MRR − Expansion MRR) ÷ Starting MRR × 100
If the answer is negative, you have net negative churn. The flip side is net revenue retention (NRR), which is the same idea expressed as a percentage of what you started with. Above 100% is the goal. See my NRR guide.
Worked example (corrected from the book)
Your business has 1,000 customers paying $20 a month, so starting MRR is $20,000. During the month, 20 customers cancel: churned MRR is $400. Thirty customers upgrade to a $50 tier. In the book I counted their full $1,500 as expansion, but strictly, expansion is only the increase: 30 × $30 = $900. So net revenue churn is ($400 − $900) ÷ $20,000 = −2.5%. Your MRR grows to $20,500 and your NRR is 102.5%. Still net negative churn, and still a great month. Run this every month and watch both numbers.
Tip #2: Use upselling and cross-selling
"Would you like fries with that?"
Almost anyone knows where that line comes from. It may be the most famous upsell in history. Everyone buying a burger at McDonald's hears it, and plenty of them spend the extra dollar or two.
Amazon does it with "frequently bought together" and recommendations based on what you've bought before.
Spotify does it beautifully in its free tier. It doesn't hide premium features. It lets you discover them, then tells you you've found a Premium feature and offers the upgrade right there, at the moment you wanted it.
Do you have an extra service your customers would benefit from? Tell them. A higher tier with more value? Tell them about it. Not everyone will buy, but far more will than if you wait.
Tip #3: Expand your resources
This means offering more of what customers already use.
When I wrote the book, Dropbox Basic gave you 2GB of storage. Need more? Plus gave you 2,000GB, Professional 3,000GB, and business tiers 5,000GB. One click and you're upgraded.
Did you notice I wrote 2,000GB, not 2TB? So did Dropbox. That's how its upgrade page showed it, because 2,000GB sounds bigger than 2TB, even though they're the same. Very clever, Dropbox.
If your service naturally scales with usage (storage, credits, users, locations, sessions), this is the easiest expansion lever you have.
Tip #4: Use your customer data
SaaS start-up Groove had a problem.
It was bringing in plenty of new users, but monthly churn was 4.5%. Growth like that is unsustainable. Groove was losing too much money because it couldn't keep the customers it had.
So what did it do?
It got funky with its user analytics.
...
Sorry, that was terrible.
How Groove reduced churn by 71 percent with a simple email
Groove dug through its data to find what it called "red flag" metrics: patterns that customers who later left tended to share.
Two stood out: login frequency and session length.
Session length was the big one. Customers who left tended to spend about thirty-five seconds per session. Customers who stayed averaged more than three minutes.
Short early sessions hinted that a customer hadn't finished setting up. So Groove emailed that group and offered a short call to help them get set up.
The customer got better service. Groove got a chance to show off the product. Win-win.
It made similar offers to customers who could get more out of Groove if they knew about features they weren't using.
The result? Groove reduced churn by 71%, because its data told it why people were leaving, and that gave it something to do about it.
Don't ignore your data.
What happened to Groove
Groove's founder has since rebuilt the company for the AI era. GrooveHQ.com now redirects to Helply, an "AI-native customer support platform" built around AI agents, with Groove described as Helply's original product. Plans start at $49 a month with unlimited users ("seats are free"), and AI work is paid for in credits priced by complexity: 1 credit for a draft, 4 for a lookup, 12 for a full investigation. Unlimited seats removes a classic churn trigger, because customers no longer cut users to save money. The Groove case study has the full update. The original lesson stands: find the early behavior that predicts staying, and help anyone who hasn't done it.
Implementing a modern loyalty program
Loyalty is the secret ingredient in net negative churn. Loyal customers stay longer, spend more and recommend you.
We've all seen the traditional scheme: a card that gets stamped or swiped until you earn a discount. It's simple and it works, but it's not the most effective approach for a subscription business.
A deep dive into Uber's modern spin on loyalty
In the book I took apart Uber Rewards, because it was a brilliant twist on points.
You earned points on every ride and Uber Eats order: one point per dollar on Eats, but three per dollar on premium rides. That was a gentle nudge towards the more expensive services.
Then there were four tiers: Blue, Gold (1,200 points, so close to $1,000 of spending in a period), Platinum (4,000 points) and Diamond (7,500 points). Gold got partner rewards and 10% off Comfort rides. Platinum got Price Confidence, a locked-in price on a regular route. Diamond added free upgrades and premium phone support.
Uber had built a membership inside its loyalty program. Just as the book went to print, Uber announced it was ending the program in favor of a subscription.
What happened next: paid membership won, and so did the lawyers
Uber Rewards ended on 1 November 2022. The paid Uber One membership (up to $9.99 a month in the US) bundles $0 delivery fees and member pricing across rides and food, and passed 50 million members by Q1 2026. But on 21 April 2025 the FTC sued Uber, alleging it enrolled people without consent, overstated savings and made cancelling hard. An amended complaint, joined by 21 states and DC in December 2025, alleged users could face up to 23 screens and 32 actions to cancel. The lesson: paid membership beats free points, but "easy to join, hard to leave" is now a liability. See the Uber One case study.
Here are five more tips to make your loyalty program work harder.
Tip 1: Focus on existing customers as much as new ones. In 2014 Lenovo bought IBM's x86 server division and inherited its reseller partners, who weren't clear on the new direction. Rather than let them drift, Lenovo built the Lenovo Expert Achievers Program (LEAP): training modules and a sales dashboard that earned points for rewards. Seven hundred firms signed up, collectively sold seven times as many products as the year before, and Lenovo beat its $170 million revenue target by 40%.
Tip 2: Personalize. Designer Shoe Warehouse (DSW) found its automatic points program was so easy that some customers forgot they were in it. So it sent personalized emails showing each customer how many points they needed for their next $10 certificate and which rewards they could claim now.
Tip 3: Curate your rewards. The North Face uses customer data to tailor rewards, so a member who hikes in cold weather sees boots and limited editions that suit them, not generic offers.
Tip 4: Build in tiers. Sephora runs tiers from the free Insider level up to VIB and Rouge, which you reach by spending $1,000 or more. Tiers let you guarantee yourself a certain income before you start paying it back.
Tip 5: Build community through the program. The Body Shop's Love Your Body Club invites members to events where they meet other members, plus parties and birthday vouchers. Friendships are hard to walk away from.
The 2026 version of tiers is Starbucks. On 10 March 2026 it relaunched Rewards with three levels (Green, Gold and Reserve) after hitting 35.5 million members active in the previous 90 days. Green members' Stars expire after six months unless they make a qualifying purchase, redemption or reload each month. Gold Stars never expire. That's gamification doing churn reduction: expiring rewards at the bottom give lighter users a reason to come back each month.
What is churn really costing you?
Calculate customer and revenue churn, and see the dollars walking out the door each year, with the free Churn Rate Calculator.
Keeping customers happy: nine best practices
Great loyalty programs keep customers happy. Using your data to make relevant offers keeps them comfortable. If they're happy, they stay.
It's that simple. Here are the practices I recommend, seven from the book and two more I have added since.
Practice #1: Know how your customers define success
How do you define customer success in your business?
Trick question. You don't. Your customers do. Your job is to make sure your product delivers their version of success.
The simplest fix is the one most people never try: ask them. Surveys, questionnaires and, best of all, conversations.
Practice #2: Have a dashboard for your metrics and usage
Cloud storage company Box used Salesforce to work out why customers left, which led it to define "churn events" and build campaigns to counter them. But it struggled to measure whether the campaigns worked. So it added dashboards from Gainsight showing product adoption, usage and the effect of its countermeasures. One dashboard listed top clients at risk of not renewing in the next ninety days: a perfect shortlist for the customer success team.
You can start with a spreadsheet. The point is to see negative usage patterns early.
Practice #3: Develop onboarding and offboarding processes
Signing up a subscriber isn't finished after the first payment. Tell new customers what to expect, where to go for what they need, and who to contact for help.
Offboarding isn't saying goodbye. It's finding out why they're leaving, fixing it for the next customer, and, if the account is valuable enough, getting on a call to see if you can keep them. A short exit survey is the minimum.
Practice #4: Get to the first "aha!" moment fast
Help each subscriber reach their first "aha!" as quickly as possible.
- Wishpond's "aha" is when a user publishes a landing page or contest and realizes they have a live campaign. It found that from its analytics and by asking customers, then removed steps before the publish button, added templates and added live chat.
- Alpha Software's "aha" is when a customer first sees an app working offline, a feature so loved it was patented.
- Cloud Campaign's "aha" is when a marketing agency sees a white-labeled one-click report with its own logo, which lets it charge clients a higher retainer.
Find yours, and remove every step between sign-up and that moment.
Practice #5: High touch for the first 90 days
The last thing you want is new subscribers feeling forgotten. Contact them regularly with emails, calls and check-ins.
There's a balance. I'd get sick of a company messaging me every hour. Stay present without becoming noise.
Practice #6: Segment your customer base
When Monster.com pushed deeper into small business in 2017, it needed to cut churn in a market it knew less well. It improved onboarding, but the key was segmentation. Using Totango, it split small business customers into three groups, tracked each group's activity and built a health score for each client. When a customer ran low on job listings, a rep reached out to top them up.
The result was reduced churn, because offers were relevant.
Practice #7: Create a feedback loop
What are your clients telling you about your product?
What do you mean, you don't know? Let's fix that.
Give clients easy ways to reach you. Then, and this is the critical part, keep talking to them. Tell them you got their message, what you're doing about it, and ask them afterwards whether it worked.
At my previous software company, we created a weekly "Voice of the Customer" meeting. The customer success team runs through every reason customers gave for leaving that week, with someone from every department in the room, so the whole business knows.
Practice #8: Build a 360-degree view of each customer
Bring everything you know about a customer into one place: what they bought, how they use it, support tickets, survey answers and payment history. When anyone in your team speaks to them, they should see the whole picture, not one slice.
Practice #9: Set up real-time alerts
Don't wait for the monthly report. Set alerts for the behaviors that predict churn, such as no login for a week, a key feature never used, a failed payment or a downgrade. Then make sure a person acts on each alert within a day or two.
Voluntary vs involuntary churn: fix the easy half first
Not all churn is a customer deciding to leave.
- Voluntary churn is a customer actively cancelling. Everything above is about this.
- Involuntary churn is a customer who meant to stay but whose payment failed: an expired card, insufficient funds or a bank decline.
Involuntary churn is often the quickest win, because the customer never wanted to go. Card updater services, smart retries, pre-expiry reminders and a friendly dunning sequence recover a lot of it. My guide to involuntary churn and failed payments shows how.
What is a good churn rate? Real numbers from the book's companies
Benchmarks vary hugely by industry, price and customer size, so I'd rather show you what real companies reported than invent an average. My page on churn benchmarks by industry has more.
| Company | What it reported | What it tells you |
|---|---|---|
| Peloton (FY26) | Average net monthly churn of 2.2% in Q4, up from 0.73% when I wrote the book | Sunk-cost retention fades as equipment ages |
| Zoom (FY26) | Online monthly churn of 2.9% in Q4 | Self-serve, card-paying customers churn fast without switching costs |
| Procore (2025) | Gross revenue retention of 95% | Mission-critical software with unlimited users keeps customers |
| Udemy Business (2025) | Net dollar retention of 93% | Team plans can still shrink without expansion |
| TechnologyOne (FY25) | Net revenue retention of 115% | Deep switching costs plus more modules equals net negative churn |
Measure two numbers separately: churn for small, self-serve customers and churn for larger contract customers. They behave very differently. Then calculate what your churn actually costs you with the churn rate calculator.
Free Subscription Profit Review
Australian business owners can book a free 30-minute review with a BPAG Accountants adviser. An honest conversation about:
- whether a subscription model could work for your business
- how recurring revenue could cover your fixed costs each month
- how it could lift what your business is worth
No cost, no obligation. General information only until we understand your situation.
Make cancelling easy: the 2026 rules on leaving
Here's the biggest change since the book.
Making it hard to leave used to be a dirty growth hack. It's now one of the most expensive mistakes a subscription business can make.
- Amazon: on 25 September 2025 the FTC announced a $2.5 billion settlement over allegations that Amazon enrolled people in Prime without consent and made cancelling deliberately hard. It included a $1 billion civil penalty and $1.5 billion in refunds, plus required changes such as letting customers cancel the same way they signed up. See the Amazon Prime case study.
- Uber: the FTC's 2025 lawsuit (above) alleged up to 23 screens and 32 actions to cancel Uber One.
- Australia: as at October 2026, the Competition and Consumer Amendment (Unfair Trading Practices) Act 2026 has passed, and its subscription rules start on 1 July 2027. According to Treasury and the Parliamentary Library, businesses will need to disclose key terms at sign-up, give notices during the subscription (expected to include reminders before a free or discounted trial ends and before renewals, with timing set by regulations not yet made), and remove unreasonable barriers to cancelling. Some protections extend to small-business subscribers. My guide to Australia's subscription trap laws has the detail. This is general information only, so get legal advice on your own sign-up and cancellation flow.
So how do you keep customers without trapping them? Offer value at the moment of cancellation:
- A pause for customers whose need is temporary.
- A downgrade to a cheaper plan instead of leaving altogether.
- A "here's what you got" summary that honestly shows the value they received.
- A one-click cancel that still lets them say why, through a one-question survey.
My guide to cancellation flows that save customers legally has examples.
When you don't need all of this
Everything on this page sounds like a lot of work.
I won't lie to you. It is.
But here's the final point I made in the book. I've used Slack for years and I love it. I've never had a Slack account manager. I've never done a Slack training session. I'm almost sure I never got a ninety-day check-in, and I'm not in any Slack loyalty program.
So why do I stick around?
Because it's so simple and useful that I don't need any of that.
An easy-to-use product with strong market fit needs far less customer success effort. If you find yourself needing every practice on this page just to keep people, the real problem may be product/market fit. But while you're building, these practices will nudge you ever closer to net negative churn.
2026 update: churn in the AI era
Here's what I'd add to the chapter today.
Win-back can beat acquisition. Duolingo's biggest growth lever in Q2 2026 was a "Streak Revival" event in June, when 15.4 million learners revived lost streaks, including nearly 8 million who had no active streak when it began. Lapsed customers already know you. Give them a reason to return: restore their status, their history or their perks if they come back this month.
Pricing changes are churn events. The AI companies that lost customer trust in 2025 and 2026 mostly did it by quietly shrinking what a plan included, not by raising the headline price. Cursor publicly apologized and refunded unexpected charges after its June 2025 change. Developers summed up one GitHub Copilot change as "you will get less, but pay the same price." Give notice, show customers a worked example of their own bill and grandfather loyal customers where you can.
Unlimited seats removes a churn trigger. When AI lets customers do more with fewer people, they cut seats. Products priced on work done rather than users (Helply's credits, Procore's construction volume) don't shrink every time a client restructures.
A free substitute can turn churn into collapse. Chegg's homework-help subscription lost half its revenue as free AI answers spread. Ask yourself what customers would still pay you for if AI did your core deliverable for free, and build that now.
AI makes the feedback loop cheaper. You can now paste a month of exit survey answers, support tickets and reviews into an AI tool and get the top reasons people leave in minutes. Try the Customer Pain-Point Analyser on your own data.
How will you keep and grow your customer base? What to do next
The main takeaway: you only get net negative churn with happy customers, and you only get happy customers by focusing on the people who've already signed up as intently as on the next ones.
This week:
- Calculate your churn and NRR for the last three months, with expansion separated out.
- Find one red-flag metric. Compare customers who left with customers who stayed. What did the leavers not do in their first two weeks?
- Send the Groove email. Offer help to everyone showing that red flag.
- Define your "aha" moment and remove one step between sign-up and reaching it.
- Book a weekly voice-of-the-customer meeting, even if it's fifteen minutes with yourself and a spreadsheet.
- Audit your cancellation flow against the 2027 Australian rules and the "cancel the same way you joined" standard.
Read the whole book free
The Subscription Playbook as a free ebook and audiobook. Australian business owners can claim a free hardcover, shipped free.
Claim the free hardcover Get the ebookAction exercise: What actions will you take to retain and grow your customers? Write down your red-flag metric, your "aha" moment and the save offer you'll show at cancellation. Then move on to the final chapter: how to know if you have product/market fit.
Frequently asked questions
What is the most common reason customers churn?
Most voluntary churn comes down to customers not getting enough value, or not noticing the value they get. That often starts in the first weeks, when a customer never finishes setup or never reaches their first result. The rest is usually price, a change in the customer's circumstances, or a better alternative. Failed card payments are a separate, often fixable, cause called involuntary churn.
How quickly should I act when a customer shows warning signs?
As soon as the signal appears. Groove found that customers who were about to leave showed short sessions from their first visits, so it reached out early with an offer of help. Set up automatic alerts for your own red-flag behaviors, such as no login in seven days or a missed setup step, and contact the customer within a day or two.
Does raising prices increase churn?
Often a little, sometimes not at all. Netflix and Spotify have raised prices repeatedly and kept growing, because their products are hard to replace. The risk is highest when the rise isn't tied to visible new value or when customers were already lukewarm. Give notice, explain why, consider grandfathering loyal customers briefly, and watch churn closely for the next two or three months.
Are save offers still allowed when a customer cancels?
Generally yes, as long as they don't become a barrier. Offering a pause, a downgrade or a discount is reasonable. Making customers sit through many screens, phone during limited hours or repeat their request is what regulators target. In Australia, rules starting 1 July 2027 will require businesses to remove unreasonable barriers to cancelling. Get advice on your own flow.
What is the difference between customer churn and revenue churn?
Customer (logo) churn counts how many customers left. Revenue churn measures how much recurring revenue you lost, including downgrades. They can tell different stories. Losing ten small customers is less painful than losing one large one. Net revenue churn also adds back expansion from customers who stayed, which is how it can turn negative.
Get both books free
The Subscription Playbook (ebook + audiobook) and Feed a Starving Crowd, instantly. In Australia? Claim the free hardcover.
Robert Coorey, MBA
Robert is the author of The Subscription Playbook and Feed a Starving Crowd and a Director at BPAG Accountants in Sydney. More about Robert.
Sources
- Helply (formerly Groove), homepage and pricing (accessed 27 Oct 2026)
- FTC, FTC takes action against Uber for deceptive billing and cancellation practices (21 Apr 2025)
- FTC, FTC secures historic $2.5 billion settlement against Amazon (25 Sep 2025)
- Parliamentary Library, Bills Digest: Competition and Consumer Amendment (Unfair Trading Practices) Bill 2026
- Treasury, Government targets hidden fees and subscription traps (9 Feb 2026)
- Starbucks, Reimagined Starbucks Rewards loyalty program launches (10 Mar 2026)
- Peloton, Q4 and FY2026 results (6 Aug 2026)
- Duolingo, Q2 2026 shareholder letter (Aug 2026)
Facts and prices checked 27 October 2026. Spotted something out of date? Tell us.