Il existe un défaut de conception dans presque tous les dispositifs d’activation SaaS, et il est tellement évident qu’on ne le voit plus : le message in-app ne touche que les utilisateurs qui sont revenus dans le produit.
Autrement dit, vous adressez vos plus beaux tooltips, vos checklists d’onboarding et vos tours guidés exactement aux gens qui n’en avaient pas besoin. Ceux qui ont créé un compte mardi et ne sont jamais revenus ne verront jamais rien. Ils sont pourtant la majorité, et ce sont eux qui décident de votre taux de conversion.
Le chiffre qui devrait faire mal : selon l’analyse de 200 produits B2B menée par Kyle Poyar avec ProductLed et ChartMogul début 2026, l’écart de conversion entre le premier et le dernier quintile des produits en self-serve atteint un facteur 10. Même produit, même marché, même prix. La différence se joue sur ce qui se passe entre l’inscription et le retour.
The real bottleneck is not conversion, it is activation
The 2026 data leaves no room for doubt. The median B2B SaaS activation rate sits around 52 %, while the top quartile reaches 65 to 75 %. More importantly: users who complete the product’s key actions convert three to five times better than the rest. Activation alone is said to explain 60 to 75 % of the variation in trial conversion rates.
Operational translation: you do not have a pricing problem, you have a first-use problem. And you have very little time to fix it. Analyses converge on one brutal point: most conversions are decided before day seven, and beyond day fourteen the rate falls to residual levels.
Seven days to bring someone to their tipping point, on a channel that can only speak to them if they have already come back. That is the problem stated properly.
Why email alone no longer closes the gap
The classic answer is to stack an email sequence. It remains necessary, but it suffers from two limits every growth lead knows without always putting them into words.
The first is congestion. A work inbox receives dozens of messages more urgent than your onboarding nudge every day. The second is subtler: email is read at a moment that has nothing to do with the moment the user could act. A reminder checked on a phone during a commute produces nothing, because the product lives on the office computer.
The web push notification solves exactly that mismatch. It appears on the workstation, where B2B SaaS is genuinely used, and it reaches the user whether or not they are inside your application. No install, no phone number, no personal address to obtain. One click on a browser prompt and the channel is open. The full mechanics are set out on our Web Push page.
The 6 use cases that change an account’s trajectory
1. Recovering abandoned activation at day one
A user creates an account, connects two integrations, stops before the third and closes the tab. That precise moment is the most profitable in your entire customer lifecycle, and the one most often wasted.
The notification goes out the next day, late morning, and above all does not say “come back to the platform”. It states what is left to do and what it costs: “One integration left to connect before your first report runs. Two minutes.” The effort is quantified, so it looks surmountable.
2. Saving the end of the free trial, with real numbers
Three days before expiry, most SaaS companies send everyone the same message. That is waste, because by then you know exactly what each account has done.
A user who genuinely used the product should receive their own summary: reports generated, hours estimated saved, colleagues invited. A user who never got past the welcome screen should not receive the same message, but an offer of hands-on help or a conditional extension. The same send for two opposite situations guarantees losing both.
3. Spotting disengagement before it becomes churn
On a paying account, churn is never announced by a cancellation email. It is announced by a drop in login frequency three to six weeks earlier, often invisible on dashboards because the account remains active in billing terms.
A notification triggered on that behavioural signal, rather than on a calendar, arrives while the relationship is still recoverable. This is exactly the logic described in our article on how to reduce churn: what makes a customer leave is never a single event, it is a slow disaffection nobody interrupted.
4. Driving adoption of a feature nobody has seen
You shipped a feature after six weeks of development. Three months later, 8 % of the base uses it. The problem is almost never the feature, it is that the announcement went into a changelog nobody reads and an email opened by a third of recipients.
A notification targeted at the segment that genuinely needs that feature, with a link opening the relevant screen directly rather than the home page, turns an announcement into usage. The direct-link detail matters more than people think: every extra click between message and action drives adoption down.
5. Alerting in real time on what is blocking a user
Quota nearly reached, export finished after forty minutes of processing, integration disconnected overnight, payment declined. These messages are not marketing, they are operational information whose usefulness expires fast.
This is the most underrated use case, because it makes the channel legitimate. A user warned in time that a sync had dropped is far more willing to keep receiving your other notifications. The channel’s perceived value is built on those messages, not on promotions.
6. Triggering expansion when the ceiling starts to bite
The best moment to propose an upgrade is not the end of the quarter, it is the instant the user hits their plan’s ceiling. Nine seats out of ten filled, ninety percent of the API quota consumed, a report blocked by an export limit.
A notification sent at that moment does not read as a sales nudge, it reads as an answer. It is also the use case that weighs most economically, since expansion on the installed base costs a fraction of acquiring a new account.
What makes these programmes fail
The first trap is wiring all six scenarios up on launch day. A user receiving four notifications in the first week unsubscribes, and unlike an email address, a lost push subscription cannot be won back. Two weekly sends is a reasonable ceiling, transactional scenarios excluded.
The second is thinking in calendars rather than behaviour. A nudge that goes out because it is Thursday ignores that the user logged in on Wednesday evening. These scenarios only work when wired into real product events, in a marketing automation logic driven by usage data.
The third is measuring the wrong indicator. Click-through rate on a notification tells you whether the message was well written. It does not tell you whether the channel creates value. The only honest measurement is to deliberately exclude one cohort from the notifications and compare its activation, conversion and retention with the exposed group. On building that calculation, our analysis of the ROI of your push campaigns sets out the method.
Where to start
If you could only switch on one, take the first on the list. Day-one activation recovery is the simplest scenario to put in place, the one whose effect is measurable fastest, and the one acting on the point in the funnel where money leaks quickest.
The other five will follow, in the order of your actual leaks. A product that activates well but retains badly will not start at the same end as one converting its trials poorly. The right sequence is not universal, it is read in your own numbers.
Frequently asked questions
Is Web Push relevant for a B2B or SaaS product?
Yes, and more so than for many consumer products. B2B SaaS is mostly consumed on desktop, and the web notification appears precisely on the workstation, without going through a mobile app. It reaches the user whether or not they are logged into the product, which solves the structural limit of the in-app message, which only reaches users who already came back.
How does it differ from in-app messages and onboarding emails?
The in-app message only displays if the user returns to the product, which mechanically excludes those you most need to win back. Email reaches everyone, but it is read at a moment often disconnected from when action is possible. The web notification triggers on real behaviour and appears where the product is used. The three channels complement rather than replace each other.
How many notifications can you send a SaaS user?
Two sends a week is a reasonable ceiling on relationship scenarios, on top of transactional alerts, which are better tolerated because they are useful. The decisive criterion remains the value of each message: a lost push subscription cannot be recovered, unlike an email address.
Does wiring these scenarios up require heavy development?
No, but you need to be able to surface the product events that trigger the notifications: account created, onboarding step completed, quota reached, last login. The complexity does not come from the notification itself, but from the quality of behavioural tracking on the product side.
How do you measure the real effect on activation and retention?
Click-through rate is not enough because it measures message quality, not programme quality. The reliable method is to deliberately exclude a cohort of users from the notifications, then compare its activation rate, trial conversion and retention at thirty and ninety days against the exposed cohort.
