Churn prevention marketing

By the time a CSM calls about a shaky account, the decision has usually been made. The budget was reallocated, the champion moved teams, the tool stopped getting opened in week three and nobody noticed until the renewal calendar flagged it. Most of that is catchable by campaign, earlier, for a fraction of the cost.

This playbook is the campaign side of retention. Not the QBR, the trigger.

Which risk signals give you enough warning to act

Signals are not equal, and the practical difference is lead time. A failed invoice gives you days. A departed champion gives you a quarter. Building one generic at-risk campaign flattens that difference and wastes the signals with the most runway.

Signal	Typical lead time	Detection source	Who should own the response
Champion departure	60 to 90 days	LinkedIn job change, email bounce, CRM contact status	Marketing campaign plus AE for enterprise
Seat contraction	6 to 12 weeks	Billing system seat count month over month	CSM above $25K ACV, campaign below
Core action decay	4 to 8 weeks	Product analytics, weekly active on the key event	Automated campaign
Support sentiment shift	3 to 6 weeks	Ticket volume plus sentiment, escalation count	CSM with marketing assets
Onboarding silence	3 to 10 days	No activation event by day 3	Automated campaign only
Failed invoice	3 to 7 days	Stripe or billing webhook	Automated dunning

The two at the bottom are cheapest to fix and most often unowned. A failed card on a $99 monthly plan is not a retention problem, it’s a plumbing problem, and Stripe’s own recovery tooling plus a three-email dunning sequence recovers a large share of it. Do that before you build anything sophisticated.

Getting the signals into one place is the unglamorous prerequisite. If your product analytics, billing and CRM don’t share an account id, none of this fires. That wiring is covered properly in customer health scores for marketers, and it’s worth doing before you write a single email.

Why the usage decay save and the champion departure save share nothing

A usage decay save is a product problem dressed as a marketing problem. The account bought the thing and stopped using it, which means either they never reached the value or their workflow changed. The right message is instructional and specific: here is the one report your team stopped running, here is the 90-second Loom on the faster way to do it.

A champion departure save is a relationship problem. The person who understood why you were bought has gone. Nobody left in the building can defend the line item. The right message is a business case, addressed to whoever inherited the seat, containing the outcomes the account has already produced in your product.

Send the champion-departure account a usage tutorial and you’ve answered a question they didn’t ask. Send the usage-decay account a business case and you’ve reminded a bored user that they’re paying for something they don’t use.

The onboarding row deserves emphasis because first-week churn is concentrated and largely fixable. The whole sequence for that window sits in customer onboarding marketing for SaaS, and it’s the most influential thing on this list for a self-serve product.

The cancel page is a conversion surface, so treat it like one

Most cancel flows are a confirmation dialog built by an engineer in twenty minutes. Meanwhile the pricing page has had three rounds of testing. That asymmetry makes no sense, because the cancel page is the only place where a customer with clear intent tells you exactly why they’re leaving.

A good flow does four things in order. It captures the reason before offering anything. It offers a path matched to that reason. It makes the actual cancellation easy once the offer is declined. And it sets the win back clock.

The reason capture is worth more than the saves. Six months of cancel reasons, categorised and read out loud in a product review, will tell you more than any research program you could commission. Feed it into the voice of customer program rather than letting it sit in a billing table nobody queries.

What a pause offer actually does to your numbers

Pause offers recover roughly 15 to 30 percent of intended cancellations in most practitioner reports. The more interesting effect is downstream: customers who paused and resumed tend to retain better afterwards than customers who took a discount, because the pause selected for a timing problem rather than a value problem.

There’s a cost. Paused accounts sit off your MRR for the pause window, which makes the current month look worse in exchange for a better next year. If your board reports on net new MRR monthly, expect an argument. Have the cohort maths ready before you ship it, and run the numbers through the cost of churn calculator so the conversation is about lifetime value rather than this month’s chart.

Two rules that keep pause from being abused. Cap it at three months and one pause per twelve months, and require the card to stay on file. Uncapped pause becomes a free tier you didn’t design.

Win back: wait longer than feels comfortable

The instinct is to send a win back email within the week. It performs badly. The customer just told you why they were leaving, and a message three days later reads as a company that filed the reason and ignored it.

Thirty to sixty days is the window where win back starts working. By then they’ve implemented the alternative and hit its edges. The message that converts is not a discount, it’s a specific change tied to their stated reason: the integration shipped, the report exists now, the plan they wanted is live.

• Day 0 to 7: data export confirmation and a genuine thank you. No offer.
• Day 30 to 60: first win back, led by what changed since they left.
• Quarter 2: a second touch only if something material shipped against their reason code.
• Annually: a single relevant touch, usually a major release or a pricing change.
• Never: a monthly newsletter to churned accounts, which trains them to filter you.

Win back rates in the 5 to 12 percent band over twelve months are realistic for a product with active development. If your product hasn’t changed since they left, there is nothing to say and you should not send anything, which is the part most teams refuse to accept.

Wiring it into a standing operating rhythm

None of this survives without an owner and a weekly review. The workable shape is a single retention standup: signals that fired last week, saves attempted, saves landed, cancel reasons by category, and one product ticket opened from the reason data.

Keep it separate from the acquisition standup. Retention campaigns lose every prioritisation argument they share a room with pipeline, which is the structural reason post-sale content is orphaned at most SaaS companies. The wider set of programs this sits inside is mapped in SaaS customer marketing, and the specific sequences live in churn prevention email campaigns.

Once the defensive work is running, the same signal infrastructure powers the offensive side. Expansion triggers come from the identical event stream, just read in the other direction, which is the subject of how marketing moves NRR. And if you want more ideas for the quieter touches between saves, customer marketing campaign ideas has a working list.

The honest limit of campaign-driven retention

Campaigns cannot save an account that bought the wrong thing. If your churn is concentrated in a segment your product doesn’t serve well, every save play here will produce a modest lift over a bad baseline, and the real fix is in qualification or in the roadmap.

Diagnose that first. Run a cohort split by acquisition source and segment, and if one source produces double the churn of the others, fix the source. Then build the campaigns.

What to do in the next two weeks

Audit the dunning sequence and the cancel page, in that order. Both are cheap, both are usually broken, and both produce measurable recovery inside a month. The churn risk audit checklist walks the full sweep if you’d rather do it systematically, but if you only have one afternoon, spend it on the cancel page and its reason capture.