{"id":23243,"date":"2026-08-04T06:55:31","date_gmt":"2026-08-04T06:55:31","guid":{"rendered":"https:\/\/scannn.com\/preventing-churn-in-set-and-forget-products\/"},"modified":"2026-08-04T06:55:31","modified_gmt":"2026-08-04T06:55:31","slug":"preventing-churn-in-set-and-forget-products","status":"publish","type":"post","link":"https:\/\/scannn.com\/lv\/preventing-churn-in-set-and-forget-products\/","title":{"rendered":"Preventing churn in set-and-forget products"},"content":{"rendered":"\n
\n
\n
<\/p>\n
<\/picture><\/div>\n

<\/a><\/figure>\n<\/div>\n

\n
<\/p>\n
\"\"<\/picture><\/div>\n

<\/a><\/figure>\n<\/div>\n

Learn how to deliver AI-powered, self-serve analytics customers will pay more for, backed by new research from the 2026 Embedded Analytics Opportunity Report. <\/p>\n

71% of teams embed analytics but 57% see no ROI, top teams hit 150%+ returns, and 91% plan to launch AI vs. just 14% in production.<\/p>\n

Watch Now<\/span><\/a><\/p>\n

\n
<\/p>\n
\"\"<\/picture><\/div>\n

<\/a><\/figure>\n<\/div>\n

A company I advised had a particular type of growth problem that isn\u2019t talked about much.<\/span><\/p>\n

Activation was strong. Retention was fine (but not trending in the right direction). They were sitting on a ton of usage data so they pointed a data scientist at the problem to find behavioural predictors of long-term retention. Which signals in the first 30 days predict who\u2019s still paying in 12, 18, 24 months?<\/span><\/p>\n

The answer: nothing. No feature, no usage pattern, no frequency threshold. With hundreds of thousands of customers, they couldn\u2019t find a single behavioural signal that predicted long-term retention.<\/span><\/p>\n

But the issue wasn\u2019t with the data. It was much more fundamental – the category they\u2019re in. Their product is effectively set-and-forget: you configure it once, and if it works, you never come back. They were searching for human-behaviour predictors in a product where humans don\u2019t behave.<\/span><\/p>\n

I\u2019ve written a lot about engagement and retention e.g. <\/span>user growth accounting<\/span><\/a>, <\/span>growth state models<\/span><\/a>, <\/span>engagement states<\/span><\/a>. Nearly all of it assumes that a healthy user is one who keeps showing up. For a whole category of products, that assumption isn\u2019t valid. And if you run growth for one of them, most written advice out there (my past content included) leaves you with a gap: what happens between activation and the renewal invoice?<\/span><\/p>\n

Take email and notification plumbing. Transactional email delivery, bounce handling, webhook routing, alert deduplication, escalation policies, notification fan-out. A developer signs up, grabs an API key, wires it into the codebase, tests that the password-reset email lands, and ships. Maybe they configure an escalation policy and a bounce webhook. That\u2019s the whole relationship. If the product does its job, that developer has very little reason to ever log in again.<\/span>
It\u2019s part of the value proposition. \u201cYou\u2019ll never think about email deliverability again\u201d is the pitch. Something is very broken if there are users who log in daily to check whether their emails are sending. It means those users don\u2019t trust you.<\/span><\/span><\/p>\n

The category is bigger than you\u2019d think: uptime monitoring, backups, DNS, cert management, dependency scanning, feature flags at steady state. If your roadmap has ever included the phrase \u201cgive users a reason to come back,\u201d and it felt like pushing water uphill, you\u2019re probably in it.<\/span><\/p>\n

No habit loop, no DAU worth measuring, no session-based engagement model. Do we just give up on engagement and wait for the renewal to tell us how we did?<\/span><\/p>\n

No. We change who (or what) we\u2019re measuring.<\/span><\/p>\n

No need for panic; your engagement frameworks still work, but you need to point them at a different actor.<\/span><\/p>\n

After that developer ships the integration and moves on, something is still using the product every day: the integration itself. Emails send, bounces get handled., webhooks route, and alerts get deduplicated at 3am. The deployment has a heartbeat, and it beats far more regularly than any human user ever would.<\/span><\/p>\n

So the simple solution is to track usage from two different perspectives, i.e. run two ledgers.<\/span><\/p>\n

The machine ledger tracks the deployment: is the integration live, is it processing volume, is that volume trending in line with the customer\u2019s business? This ledger is continuous, high-frequency and honest (seasonality and the odd staging environment aside). It can\u2019t be flattered by a well-timed email campaign.<\/span><\/p>\n

The human ledger tracks the people: logins, dashboard sessions, report opens. In this category it\u2019s sparse by design, and that\u2019s fine. It has a different job (more on that later).<\/span><\/p>\n

\n
<\/p>\n
\"\"<\/picture><\/div>\n

<\/a><\/figure>\n<\/div>\n

Growth accounting works on the machine ledger with one change: swap the unit. Instead of new, retained, resurrected and churned users, you count new live integrations, retained live integrations, resurrected integrations (volume came back) and churned integrations (volume went to zero).<\/span><\/p>\n

Use the same states as the growth state model, but a different actor. \u201cActive\u201d means the integration is live and processing, not that a human showed up.<\/span><\/p>\n

Do that and the mystery from the intro dissolves. The company I advised couldn\u2019t find behavioural predictors of retention because they were looking at the human ledger while the signal was in the machine ledger the whole time.<\/span><\/p>\n

The machine ledger is so important because monetary churn is a lagging indicator. By the time the cancellation hits your dashboard, the decision was often made months earlier. And in a set-and-forget product, that earlier moment usually leaves a trace.<\/span><\/p>\n

\n
<\/p>\n
\"\"<\/picture><\/div>\n

<\/a><\/figure>\n<\/div>\n

Think about how churn actually happens with email plumbing. It\u2019s rarely someone logging in, looking around and cancelling. More often: the customer replatforms and the integration doesn\u2019t survive the migration. Or a new engineering lead consolidates vendors and swaps your SDK for a competitor\u2019s. Or the customer\u2019s own product is declining and send volume decays with it.<\/span><\/p>\n

In each case the commercial event (cancellation, non-renewal) is just the paperwork catching up with an operational event that already happened: your product stopped doing work.<\/span><\/p>\n

Each of these is visible in the machine ledger long before it\u2019s visible in revenue. Send volume dropping to zero means the integration was ripped out. Volume decaying quarter over quarter means the customer\u2019s business is shrinking. An error rate spike followed by silence means a migration broke the integration and nobody came back to fix it (because nobody comes back, remember).<\/span><\/p>\n

Of these, the replatform deserves a special mention as the canonical silent churn event in dev tools. Nobody decided to leave you. Your integration just wasn\u2019t important enough to survive the rewrite.<\/span><\/p>\n

And volume going up isn\u2019t automatically good news either. Usage growth means invoice growth, and if you\u2019re on volume pricing that isn\u2019t competitive, a rising bill is often the trigger that puts you under evaluation. The customer\u2019s finance team notices the line item, someone gets asked if there\u2019s a cheaper option, and an integration that was invisible for two years is suddenly on a spreadsheet next to three alternatives. With price-sensitive customers, that growth looks like expansion revenue right up until it becomes the reason they leave.<\/span><\/p>\n

\n
<\/p>\n
\"\"<\/picture><\/div>\n

<\/a><\/figure>\n<\/div>\n

Something to be aware of – you won\u2019t know in advance which of these signals actually predict churn for your product. That\u2019s exactly why you capture them now. When the negative events happen (and they will), you can look back, correlate the signals with the outcomes, and build the early-warning system that gets you ahead of the next one. But only if the data is there to correlate.<\/span><\/p>\n

So the retention playbook here isn\u2019t a re-engagement campaign. It\u2019s pre-churn detection: alerting on the machine ledger the way your customers alert on their own infrastructure. An integration that went quiet three weeks ago is a renewal conversation you still have time to change. A cancellation notice is not.<\/span><\/p>\n

There\u2019s a second problem the machine ledger can\u2019t solve on its own. Even for customers whose integrations are perfectly healthy, value in this category is delivered continuously but felt never.<\/span><\/p>\n

This is the same structural problem insurance has. You pay every month, the product works silently in the background, and the value only becomes visible on the day something goes wrong. In the meantime, the one recurring touchpoint you reliably have with the customer is the invoice. The only time you show up is when you charge them, and that\u2019s a bad position to be in when renewal comes around.<\/span><\/p>\n

The fix isn\u2019t getting people to log in more. What works instead is engineering moments where the value gets felt, with zero effort required from the user. A quarterly digest: 2M sends delivered, 40k bounces caught before they damaged your sender reputation, 12 incidents deduplicated before they woke anyone up. A short note when you handled something automatically: \u201cwe rotated your DKIM records, nothing needed from you.\u201d The customer reads it in 20 seconds, thinks <\/span>\u201cglad that\u2019s handled\u201d,<\/span><\/em> and gets on with their day.<\/span><\/p>\n

I want to be clear about something though – these aren\u2019t marketing emails and you shouldn\u2019t treat them as such. They\u2019re receipts for work performed. The goal isn\u2019t to fight set-and-forget – it\u2019s to turn set-and-forget into <\/span>set-and-felt<\/span><\/strong>. And in this category, engagement with those receipts is a real metric rather than a vanity one; it\u2019s the measurable trace of value being felt, not just delivered. (Measure clicks and replies rather than opens given what Apple\u2019s Mail Privacy Protection did to open rates.) That trace is what you want sitting in the customer\u2019s recent memory when the renewal invoice lands.<\/span><\/p>\n

\u201cAny touchpoint invites a cancellation decision. Why would we remind people they\u2019re paying us?\u201d<\/span><\/em><\/p>\n

I\u2019ve had this conversation more times than you might think, because it\u2019s a fair concern. And it\u2019s half right. But it conflates two very different kinds of touchpoint. An activity nudge (<\/span>\u201ccome back and check out your dashboard!\u201d<\/span><\/em>) asks the user to do work, and implicitly invites them to ask why they\u2019re paying for something they never use. Value reinforcement (<\/span>\u201chere\u2019s what happened while you weren\u2019t looking\u201d<\/span><\/em>) does the opposite: it makes the invisible work visible and lowers the perceived case for cancelling. And if you\u2019re still nervous, this is a testable claim, not a matter of opinion: hold out a segment, send the digest to everyone else, and compare renewal rates instead of debating hypotheticals.<\/span><\/p>\n

The two ledgers also tell you which accounts are which. A healthy integration with a dormant human is fine: reinforce quietly, ask for nothing. A dead integration with a dormant human is not a bear you might wake. They\u2019re already gone; the only question is whether you find out now (when a win-back conversation is still possible) or at renewal (when it isn\u2019t). The bear to worry about isn\u2019t the one you might wake. It\u2019s the one that already left the cave.<\/span><\/p>\n

The strongest reason to care about any of this only shows up when you need to grow revenue from your existing base.<\/span><\/p>\n

Most set-and-forget companies eventually face NRR (Net Revenue Retention) pressure. The core product is priced per integration or per volume tier, natural expansion is limited, and the answer is usually a second product: the email plumbing company adds SMS, push notifications, an alerting product. At that point the engagement question flips. It stops being <\/span>\u201cdoes engagement predict retention?\u201d<\/span><\/em> (in my experience, only if you\u2019re measuring the right things), and becomes <\/span>\u201cdo we have any surface on which to sell the next thing?\u201d<\/span><\/em> And that\u2019s the human ledger\u2019s different job I promised earlier: it measures whether expansion is even possible.<\/span><\/p>\n

Cross-selling push notifications to an account that hasn\u2019t touched your dashboard in 18 months is possible, but it\u2019s a cold outbound motion with none of the product-led advantages you built the business on. The value digest becomes the road back to the product, and the product is where expansion conversations happen. In set-and-forget, engagement is more than just a retention lever. It\u2019s an expansion prerequisite.<\/span><\/p>\n

Set-and-forget is the customer\u2019s privilege. It can\u2019t be yours as a vendor. So, pick your machine heartbeat metric, define \u201cactive\u201d at the integration level, rebuild your growth accounting on it, alert when volume goes quiet (or spikes beyond what the customer\u2019s pricing tier makes comfortable), and put a first receipt-for-work-performed in front of your dormant-but-healthy accounts. <\/span><\/p>\n

The humans may not come back but the signal is still there. Make it felt.<\/span><\/p>\n

\n
<\/p>\n
\"\"<\/picture><\/div>\n

<\/a><\/figure>\n<\/div>\n

\n
<\/p>\n
\"\"<\/picture><\/div>\n

<\/a><\/figure>\n<\/div>\n

Book a free 1:1 consultation call with me<\/a> – I keep a handful of slots open each week for founders and product growth leaders to explore working together and get some free advice along the way. Book a call.<\/span><\/p>\n

View the State of AI Search for Dev Tools 2026 benchmark report<\/a> – 126,814 citation records across 1,075 representative tool-evaluation scenarios, 43 developer-tool categories, and five AI search platforms<\/span><\/p>\n

View your free public Dev Tool AI Market Presence Report<\/a> – 500+ dev tools across 47+ verticals and growing, all in the Dev Tool AI Search Landscape.<\/span><\/p>\n

Sponsor this newsletter<\/a> – Reach over 11,000 founders, leaders and operators working in product and growth at some of the world\u2019s best tech companies including Paypal, Adobe, Canva, Miro, Amplitude, Google, Meta, Tailscale, Twilio and Salesforce.<\/span><\/p>\n

Thanks again to this weeks sponsor: <\/span>ThoughtSpot<\/a><\/p>\n

\n
<\/p>\n
\"\"<\/picture><\/div>\n

<\/a><\/figure>\n<\/div>\n<\/div>\n

Source link <\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

Learn how to deliver AI-powered, self-serve analytics customers will pay more for, backed by new research from the 2026 Embedded Analytics Opportunity Report. 71% of teams embed analytics but 57% see no ROI, top teams hit 150%+ returns, and 91% plan to launch AI vs. just 14% in production. Watch Now A company I advised […]<\/p>\n","protected":false},"author":16,"featured_media":23244,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[143],"tags":[],"class_list":["post-23243","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai"],"_links":{"self":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/posts\/23243","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/users\/16"}],"replies":[{"embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/comments?post=23243"}],"version-history":[{"count":0,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/posts\/23243\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/media\/23244"}],"wp:attachment":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/media?parent=23243"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/categories?post=23243"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/tags?post=23243"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}