Inspired tactics, part 4: What makes subscribers churn – and how the tech industry will keep them
By Bihag Karnani
As a senior product manager at Google, Bihag Karnani oversees the development of Reader Revenue Manager – a digital subscription service designed specifically for media companies. On average, he meets with approximately 50 to 70 publishers during the course of a year.
In Part 1, we covered first-party data and the first 100 days. Part 2 discussed trials and paywalls. Part 3 was about expanding through bundles, family plans and frictionless checkout. This part is about the retain stage of Intercom’s C.A.R.E. Model – what happens once a habit is established and a subscriber has been with you for a while.
All publishers discuss “churn” as if it is one thing. It is two completely different issues with two completely different solutions.
1. Voluntary churn
Voluntary churn occurs when a subscriber decides to cancel. They go to their account page, click Cancel, and leave. This is what almost every retention program focuses on.
2. Involuntary churn
Involuntary churn occurs when a payment fails. A credit card expired, was replaced, or declined due to fraud, or there were insufficient funds. The subscriber did not want to leave; their payment just stopped working.
The second issue is the under-told story. Patrick Campbell, founder of ProfitWell, has publicly stated that roughly 30% of media subscription cancellations are due to payment failures and not any decision by the subscriber to leave. Also, it is the easiest type of churn to fix.
The unspoken third of all churn
The Washington Post was more open about this than any major publisher. At WAN-IFRA’s 2018 World News Media Congress, then-vice president Miki King walked through the work the Post had done on involuntary churn. Their findings were: “About a third of our churn was coming from what we like to think of as non-voluntary cancellations. These were subscribers who were not actively going into their accounts and cancelling but ended up lapsing because they had outdated information on file with us.”
The solution wasn’t complicated. The Post had been sending email notifications to subscribers whose cards had expired. Almost all of them either didn’t see them or ignored them.
The Post switched to displaying the notification within the reading experience itself – a prompt appeared on site linked to the user’s profile where they could update their card in one click. King said: “By notifying the subscriber while they are in the reading experience versus sending an email that gets buried has created huge changes for us. We launched this in mid-2017 and it has had tremendous impact.”
Just one change. One-third of churn fixed.
The broader technology playbook here has quietly developed over the last decade. The standard recovery sequence takes place over a period of 14 days with smart retries based on paycheck cycles, increasing banners in app over time and a final downgrade offer before access is cut off.
If done correctly, the majority of failed payments can be recovered without losing the subscriber. Major payment platforms such as Stripe, Recurly, Chargebee & Adyen all include this functionality by default.
I should mention one feature of Reader Revenue Manager since it’s relevant to this topic. RRM uses Google Pay for subscription payments which means those using it will get the benefit of automatic updates to cards without having to build that integration themselves. When a Google Pay user’s card is updated everywhere in Google’s ecosystem, the update automatically flows through to the subscription. That represents a meaningful portion of involuntary churn that effectively drops to zero, on infrastructure that publishers don’t need to maintain.
Voluntary churn: Treat the cancel flow as product
When a subscriber decides to leave, the cancellation flow is a product surface – not afterthoughts for customer service. Clean data published by Churnkey – which has analysed more than 3 million cancellation sessions in SaaS – states average save rate across all cancellation flows is 34 percent.
Every additional survey question decreases the save rate by 6.7 percent. Customers who accept a discount stay an average of five months longer on their subscription.
A typical pattern looks like this: Cancel button -> Short reason survey -> Adaptive offer tied to answer.
- Too expensive: Gets a discount or a downgrade to a cheaper tier.
- Not enough time: Gets a pause for one to three months.
- Wrong topics: Gets personalised newsletter or topic mix adjustment.
- Switching to competitor: Gets bundle add.
- Moving: Gets digital only conversion.
News media organisations use several retention strategies, which include using pauses, downgrades, predictive models and proactive retention layers. One of the key issues with news media is that these retention techniques are used too little.
Netflix’s one month pause (which can be extended up to 3 months) allows subscribers to retain their profiles, watch lists and recommendations. Pause is shown through Churnkey data to be the second best method for saving subscribers after discounts – those who accept pause remain subscribed 5.5 months longer than average, and 67% of consumers stated they preferred pausing instead of cancelling.
Le Monde has been utilising this technique; WAN-IFRA’s 2026 World Press Trends Outlook noted that pause functionality should become the norm among all publishers.
Downgrade options also deserve recognition. Click Funnels released a case study demonstrating that when the process of cancelling was preceded by a downgrade opportunity rather than a discount, save rates increased by 130%.
This is based on the fact that “I cannot pay for the full subscription” typically does not equate to “I do not want any type of subscription.” Therefore, a downgrade to an advertisement supported version, or weekend only, or newsletter only may keep a relationship intact that would normally cease.
Preventing churn before it happens
Advanced retention efforts have moved away from a reactive posture toward predictive. Rather than waiting for a subscriber to click the cancel button, create a predictive model to identify at risk subscribers before they make a decision.
These tools are getting incredibly good. No-code churn modelling (Pecan AI). Predictive Cohorts (Amplitude). Churn Management (Recurly). These tools achieve accuracy levels of 78-85 percent when attempting to predict churns occurring three to six months in advance.
All of this falls under what Stripe refers to as Negative Churn – the scenario where revenue generated by current customers exceeds revenue lost due to cancellations. Stripe and Chart Mogul’s Joint SaaS Benchmarks report indicates that expansion revenue accounted for 28.8 percent of total revenue in 2020 and rose to 32.3 percent by 2023 throughout the SaaS industry.
The context in which this takes place is important. Those publishers who view growth as “add more subscribers” will lose to those publishers who see growth as “the revenue from each of our existing subscribers grows year-over-year and we catch those at risk of churning before they leave.”
Proactive retention layer
On the flipside of retention is the proactive work that occurs prior to a subscriber even considering leaving – the proactive engagement that helps maintain the habitual nature of the subscription.
Technology has created a well-defined playbook regarding how to effectively engage subscribers proactively. The most iconic example of this mechanism is Spotify Wrapped, which provides a personalised yearly summary of activity and demonstrates the value provided through the subscription while encouraging sharing and creating anticipation for the subsequent year.
During the December 2024 holiday season Wrapped attracted more than 200 million active users in a 24-hour span. Strava, Apple Music Replay, Duolingo, Letterboxd and many others have followed suit.
The mechanism is straightforward – at predetermined time periods provide the subscriber with a reminder of the value they have previously received from the subscription so that when making the decision to renew there is value being referenced versus the monthly cost.
Amedia, a Norwegian publisher operating approximately 90 regional publications, has developed one of the strongest proactive-engagement/retention systems within news publishing. Their north-star metric is relatively simple: How many subscribers consume at least one article daily for at least 10 seconds?
As reported during a recent WAN-IFRA Newsroom Summit, Janne Rygh, Editorial Developer at Amedia defined it as follows: “It’s a formula. We know that the more you read, the less likely you’ll churn.” (WAN-IFRA)
The data supporting Amedia’s methodology is compelling. Subscriber(s) who consumed an article every three days or less churn at a monthly rate ranging from 7-12 percent. Subscriber(s) consuming articles every two days churn at 5-6 percent. Subscriber(s) consuming articles every day churn at 3-4 percent.
Amedia views their entire organisation as a retention engine – every editorial decision, every product feature and every marketing e-mail has implications related to increasing the reading frequency among subscribers.
Since implementing this strategy Amedia has seen their number of subscribers increase from 430K to 850K across their various titles with 87% of page views attributed to logged in subscribers. (WAN-IFRA)
A key takeaway from Amedia is not necessarily any specific tactics, but rather their organisational structure. Retention is not owned by a marketing group nor a customer success group – it is owned by everybody – newsroom, product, marketing/customer support – since every area influences whether subscribers return tomorrow. (WAN-IFRA)
What to do
This week: Pull your voluntary churn rate. There’s no way to understand voluntary churn if you can’t identify involuntary churn. Involuntary churn is defined as when a customer cancels their subscription due to some other issue (e.g., billing error) rather than because they’re unhappy with the product or service.
While many publishers claim they’ve made significant strides in identifying which customers cancel due to involuntary reasons, most still lack solid methods to measure and quantify involuntary churn.
This quarter: Build the cancellation flow you should have. There are four components necessary to build a proper cancellation flow:
- Survey question: Allows the publisher to ask why the customer is cancelling.
- Adaptive offers: Incentives tailored to the survey response.
- Pause option: Allows customers to temporarily stop billing instead of quitting.
- Downgrade path: Lower cost options to maintain the relationship.
Measure save rate against baseline. Track how successful your cancellation flow was by comparing your save rates before and after implementing the new flow.
This year: Find the engagement metric that predicts retention in your data. At Amedia, they’ve found that subscribers who spend 10 seconds per day reading articles are much more likely to retain their subscriptions.
The Financial Times uses an RFV (recency, frequency, monetary) score. Whatever the metrics may be in your business, use those metrics to create targeted messaging and offers across all of your touchpoints – including but not limited to onboarding messages, promotional push notifications, newsletters, save offer flows – to encourage subscribers to engage at a level that will help increase their retention.
Retention is an org-wide responsibility. No matter how good your “save” offers are, if you’re losing customers at a high rate, then you’re going to struggle with retaining existing customers.
Create a culture within your organisation where everyone understands that retention is everybody’s job – from engineering teams who implement solutions that reduce involuntary churn; to marketing teams who design campaigns that encourage users to stay longer; to sales teams who ensure that all interactions with customers result in increased satisfaction and usage.
Run all subscriber decisions as experiments. Use experimentation techniques such as a/b testing to determine which subscriber decisions (i.e., onboarding messaging, save offers, etc.) produce higher retention rates.
That’s really the core message of this entire series. Subscriptions are built upon numerous small ideas that are individually tested, learned from and ultimately combined into larger initiatives over time.
The tech industry has been experimenting with these ideas for nearly two decades. News is just now starting to follow suit – and it is the news organisations that move the fastest will be the ones we’ll hear about in 2030.
About the author: Bihag Karnani is a Senior Product Manager at Google, where he leads Reader Revenue Manager (RRM), Google’s subscription platform for news publishers, and drives user personalisation strategy for AI Mode and AI Overviews. He works with publishers globally on subscription strategy.
