How to use notifications and lifecycle messages

1 The bottom line

Send each message because of something the user did, and put something useful in it. Messages tied to behavior bring users back, while generic volume teaches them to tune you out.

2 Key takeaways

• Build the retention loop in the product first, then use messages to reinforce it. A message cannot rescue a loop that does not exist.
  • Fitness trial users who work out once are far less likely to cancel, so those emails push the first workout Steve P. Young, App Masters4:35.
• Trigger each message from something the user did. Behavior shows what the user cares about and when to reach them.
  • Duolingo tested many complex timing rules, and a reminder 23.5 hours after the last practice always won Jackson Shuttleworth, Duolingo56:05.
  • Zinio saw a 13x increase in conversions after moving cart abandonment to three pushes Steve P. Young, App Masters7:28.
• Ask for push permission with a clear reason, at a moment that gives it context. Apple's system prompt can only be shown once.
  • Behavior-based segments and messages doubled or tripled push opt-in in some cases Mirela Cialai, Zinio6:21.
• Spend your testing effort on the message, not the channel. The wording and psychological angle drive the result.
  • Changing only the subject line to "our most popular workouts" doubled workout completion rate Steve P. Young, App Masters6:05.
  • A low-effort email series of 15 second GIF tutorials raised fully onboarded users by 81% within a month Jordan Bryant, TimeClock9:05.
• Protect the channel with fewer, better messages. Each extra push makes every other push work less well.
  • The average US smartphone user receives 46 pushes a day Steve P. Young and Rosie, App Masters5:16.

you need to build the loop first and then you build the tactical things around it to kind of reinforce those loops.

Ben Gammon, Ladder

2.1 Pick the behavior to push

Messages work when they push users toward an action that already drives retention.

• Find the action that predicts retention and message people toward it. Source: Ben Gammon, Ladder13:04
  • Find the features that drive retention and drive usage of them inside the product first Ben Gammon, Ladder13:04.
  • Message only users who started a trial, and deep link them straight into the core action, such as a workout Steve P. Young, App Masters4:35.
  • Trial starters showed intent, so they are much easier to re-engage than people who never started Steve P. Young, App Masters4:35.
• Remind users of features they already have before you build new ones. Source: Steve P. Young and Chaim, App Masters5:07
  • Send event alerts two days, one day and half an hour before an event, so the app feels smart and users reopen it Steve P. Young and Chaim, App Masters5:07.
  • Send email and push that highlight popular and little-known features, tied to the one problem your app solves Steve P. Young and Chaim, App Masters6:45.
  • A customer asked for more features, and event reminders using the existing event feature were suggested instead Steve P. Young and Chaim, App Masters5:07.
• Let user research decide the content, timing and wording of lifecycle emails. Source: Jordan Bryant, TimeClock9:36
  • Users told the team which three steps to teach, when to send them and how to explain them Jordan Bryant, TimeClock9:36.
  • Plug a funnel hole with the easiest non-technical fix before rebuilding the app. The TimeClock series took about a week to implement Jordan Bryant, TimeClock9:05.
  • Copying an email series without that research will not reproduce the result Jordan Bryant, TimeClock9:36.
• Connect push notifications to any network feature you add. Source: Steve P. Young and an app growth consultant, App Masters2:36
  • Notify users when something happens, such as a new connection or message, so they can jump straight back in Steve P. Young and an app growth consultant, App Masters2:36.
  • Without the notification, the network value stays invisible Steve P. Young and an app growth consultant, App Masters2:36.

2.2 Ask for permission and capture email

Earn push permission with a reason tied to the user's context, and collect email as a second channel.

• Give the reason before the ask, tied to what the user just did. Source: Mirela Cialai, Zinio6:21
  • After users pick a practice time, say you will help them stay consistent and then ask for push Steve P. Young, Breathwrk3:20.
  • Explain why reminders improve routine success, and show the reminder pop-up in front of the setup screen TP Young, Routinery2:40.
  • Tailor the ask to the action. Ask new subscribers about new issues, and ask magazine searchers about sales Mirela Cialai, Zinio6:21.
  • Calm asks again right after a completed meditation, praising the user and offering morning, afternoon or night reminders Steve P. Young, Calm2:41.
  • Calm also shows the most popular time among members, 8:00 AM in the morning, which makes the choice one tap Steve P. Young, App Masters2:55.
  • For a game, explain rewards, chests that unlock and new collector cards. Chuka guesses this could raise opt-in chances by about 50% Chuka Ikokwu, Exploding Kittens13:21.
• Do not show the system prompt cold at install. Source: Mirela Cialai, Zinio5:06
  • Fastic asks on first open, and Steve does not recommend it, even for a fasting app, because users have not seen why they would want reminders Steve P Young, Fastic4:41.
  • Zinio first showed the native form to everyone after install. Few said yes because the value was not yet proven Mirela Cialai, Zinio5:06.
  • Onboarding is the best moment, because it is when most buyers buy and users are most likely to opt in. Genie asks for the ATT prompt, then a review, then notifications Steve P. Young, App Masters1:13.
  • Routinery tests the ask on the trial screen with "we'll remind you" against the second screen TP Young, Routinery4:24.
• Guard your one shot at the system dialog. Source: Steve P. Young, App Masters2:11
  • Show your own screen first, and show the system dialog only when the user taps to enable Steve P. Young, App Masters1:17.
  • If the user taps Not now, do not trigger the system dialog. A Don't allow there cannot be re-asked, and the user must go to Settings Steve P. Young, App Masters2:11.
  • Retry later at a moment when users are happier with the app Mirela Cialai, Zinio5:32.
• Capture email inside the app so you own a channel Apple does not control. Source: Steve P. Young and Chaim, App Masters7:05
  • Productive asks on a "track your progress" screen after the paywall, framed as personalization, and lets the user X out Steve P. Young, Productive7:54.
  • Don't overuse it. Add SMS as an extra layer to verify the user has a real mobile number Steve P. Young and Chaim, App Masters7:05.
  • Steve calls missing email capture the biggest fail for app developers. His example is HotelTonight's follow-up email after he looked at a hotel Steve P. Young, App Masters11:26.
  • Social reach is small. Boselin says only about two percent of a business's content is seen by its audience Jillian Lee Stout, Melody Boselin and Sunny Dulay28:53.

2.3 Trigger messages from behavior

Send when the user's own actions say a message will help.

• Time reminders from the user's own pattern, not a stated preference. Source: Jackson Shuttleworth, Duolingo56:05
  • Send the practice reminder 23.5 hours after the last practice Jackson Shuttleworth, Duolingo56:05.
  • Send a last-chance streak saver at 10 PM to users who have not extended their streak Jackson Shuttleworth, Duolingo56:38.
  • Pair a morning reminder with a later loss-framed nudge, and link the stakes to another person, such as "Your partner is going to be sad" An (Anup), Flame12:02.
  • Duolingo dropped user-picked reminder times because life changes and stated times are unreliable. The best predictor was when the user practiced the day before Jackson Shuttleworth, Duolingo56:05.
  • The streak saver could seem spammy, but users attach positive feelings to their streak, so they see it as helpful Jackson Shuttleworth, Duolingo56:38.
  • Flame's day-30 retention went from 4.2% to roughly 30% and then up to 50%. TikTok is roughly 40%, Instagram 35% and X about 30% An (Anup), Flame12:02.
• Personalize to what the user does and shares. Source: Giordano Contestabile, Life36016:42
  • Runna notifies users to raise a pace target when they smash it, or to slow down, based on their run data Miranda Paine, Runna9:27.
  • Match the topics users talk about, such as relationships or finances, to in-app content and lifecycle touches Raj Singh, Purpose29:20.
  • Name a person from the user's circle. Life360 asks what John was up to yesterday Giordano Contestabile, Life36016:42.
  • Raj credits the personalization for above median retention Raj Singh, Purpose29:20.
  • Wrestle AI sends at least one notification a day tied to past actions, though it does not yet track which ones drive clicks George, Wrestle AI41:16.
• Catch people at the point where they drop off. Source: Georgiana Laudi, Forget The Funnel42:50
  • Land dropouts of long flows on the step they abandoned, not at the beginning Kristen Berman, Irrational Labs40:00.
  • If someone misses a milestone or falls out of regular engagement, trigger an in-app message, email or personal outreach. Do not keep sending generic emails to people who never found the valuable part Georgiana Laudi, Forget The Funnel42:50.
  • Send cart abandonment reminders at 1 hour, 1 day and about 1 week. Users who added an item have shown intent, so being aggressive pays off Steve P. Young, App Masters7:28.
  • If someone closes the app right after a search download, send an immediate use-case push. Add their city from IP, which needs no location permission Steve P. Young and a developer with a six figure app exit9:57.
  • The 13x Zinio result is told two ways. One telling says it came from moving from one push to three, the other from two to three Steve P. Young, App Masters13:08.
  • Another telling credits adding a one-week reminder to the 1-hour and 1-day ones. Steve was unsure of the exact timings Steve P. Young and Chaim, App Masters8:23.
  • A search download that goes unused signals to the App Store that you are not relevant, much like bounce rate on Google Steve P. Young and a developer with a six figure app exit9:57.
• Ask for a discount or a rating right after the great moment. Source: Steve P. Young and Chaim, App Masters11:38
  • Prequel shows a discount after the user exports and saves an edited photo Steve P. Young and Chaim, App Masters11:38.
  • The same trigger, used to ask for a rating, brought huge improvements in average rating, especially on Android Steve P. Young and Chaim, App Masters11:38.

2.4 Write messages that add value

Each message should teach, show or help, not just ask.

• Lead with what is popular. Source: Steve P. Young, App Masters6:05
  • Test "our most popular workouts" against a benefits-led subject. It works because people feel FOMO and want to know what everyone else uses Steve P. Young, App Masters6:05.
  • Email new trial users the app's most popular workouts to push them into using the product Steve P. Young, Gamezop4:16.
  • That email reduced trial cancellations as well as doubling the workout rate Steve P. Young, Gamezop4:16.
• Give the reader something real, not show-and-tell. Source: Melissa Cash, Pok Pok50:00
  • Pok Pok is testing emails that compare a child's drawing from six months ago to now, or share parenting podcast episodes Melissa Cash, Pok Pok50:00.
  • Apt wants its mentor to spot below-market pay and build a 6 to 8 week plan to ask for a raise Tomas Hoyos, Apt26:22.
  • SuperMe's agent pings the network about what the user works on and reports back only if it learned something valuable Casey Winters, SuperMe27:40.
  • Trust is earned gradually, since no one gives it on day one Casey Winters, SuperMe27:40.
• Educate first and hold the sales pitch. Source: Steve P. Young, Grammarly15:01
  • Grammarly's first emails list its products, the Pro features the user is missing, and tips for the free account, with no ask for payment Steve P. Young, Grammarly15:01.
  • The full sequence is a welcome email, a premium features email, a free-account tips email, a 40% offer, an upgrade email and a last-chance email Steve P. Young, Grammarly15:18.
• Treat email as a series and keep testing the copy. Source: Elena Verna1:12:20
  • Tie email to product communication and test the whole sequence together Elena Verna1:12:20.
  • Email is cheap to send, so the value is in the writing. Experiment with the copy all the time Steve P. Young and Chaim, App Masters7:30.
  • Average open rate is about 25%, and at best you get 40 to 50% Elena Verna1:12:20.
  • Fintech subject lines that worked use popularity, social proof and numbers, such as "do this one thing to eliminate 10 years off your mortgage" Steve P. Young, App Masters7:29.

2.5 Make offers while intent is hot

Time offers to the user's intent, and change them so they do not turn into noise.

• Send the first offer fast, to people who showed intent but did not pay. Source: Steve P. Young, App Masters8:54
  • Treat a user who saw the paywall and skipped the trial as an abandoned cart. Send a push or email offer within the first hour Steve P. Young, App Masters8:54.
  • Send a 20% offer that leads to Stripe checkout right after the welcome email Steve P. Young and Chaim, App Masters6:46.
  • Show cart abandoners a 15% discount with a countdown timer Kiss My Keto, abtest.design.
  • People who buy generally buy within the first 20 minutes to an hour of opening the app Steve P. Young, App Masters8:54.
  • The countdown test produced a 15.05% increase in conversion rate Kiss My Keto, abtest.design.
  • Steve does not know the results of the 20% offer he saw, but likes "hammering while the iron is hot" Steve P. Young and Chaim, App Masters6:46.
• Run a winback sequence that escalates the discount and varies the message. Source: Steve P. Young and Igor Lyubimov, Headway34:22
  • Collect the email before the paywall. Send the first email within hours, then at one day and two days, with no more than two or three days between emails while the user is warm Steve P. Young and Igor Lyubimov, Headway34:22.
  • Give each email one button and one discount message. Later emails use a plain, person-to-person format Steve P. Young and Igor Lyubimov, Headway32:52.
  • Tie each discount to a specific message, such as asking whether the user is burnt out, and raise the value over time Steve P. Young and Rosie, App Masters3:25.
  • Headway sends eight emails with bigger discounts, up to 90%. A typical app gets 10% or less of sales from email, and 20% is a miracle Steve P. Young and Igor Lyubimov, Headway34:22.
  • Calm's repeated "last chance for 50 off" is easy to ignore. One user got four such emails in a month Steve P. Young and Rosie, App Masters3:25.
  • Grammarly's deadline offer is 40% off for 48 hours, followed by a short last-chance email. It forces users to decide whether to stay or leave Steve P. Young, Grammarly16:16.
• Link the email straight to the offer and track it. Source: Steve P. Young, Airbridge3:05
  • Deep link users into the app and onto the offer, then track sales per link in one dashboard Steve P. Young, Airbridge3:05.
  • Send users who never converted in-app to a Stripe checkout. It is much easier to test pricing and promos there than in Apple product configuration Jillian Lee Stout, Melody Boselin and Sunny Dulay7:17.
  • Zinio's top revenue and engagement campaign is a quarterly promo-code email. Heavy discounts performed best Mirela Cialai, Zinio8:48.
  • Runafy emailed its waitlist a video with a half-price referral code at launch. Most launch day revenue came from those codes Caleb Dean, Runafy27:30.

2.6 Protect the channel

Limit volume, skip gimmicks and use other surfaces so the channel keeps working.

• Cap frequency and keep quality high. Source: Steve P. Young and Rosie, App Masters5:16
  • Personalize, use emojis and get the timing right, because every extra push reduces how effective the others are Steve P. Young and Rosie, App Masters5:16.
  • Do not rely on pushes that grow with network size. Houseparty and Clubhouse sent more notifications as users followed more people, until users ignored them all Sarah Tavel, Pinterest18:05.
  • For agent-sent recommendations, use open rate as a signal. Below 20% opened is spammy, and above 80% means do far more Casey Winters, SuperMe29:17.
  • Apple makes it easy to turn off notifications, and a user who has turned them off is very hard to retarget Steve P. Young and Rosie, App Masters5:16.
  • Before cutting a channel, test it. Mindful Mamas moms complained when daily pushes were turned off Jillian Lee Stout, Melody Boselin and Sunny Dulay27:40.
• Skip emotional tricks that manufacture engagement. Source: Ajay Mehta, Tolan45:26
  • Tolan could show a sad character after a few days away, but chooses taste and craft aimed at a helpful, supportive companion Ajay Mehta, Tolan45:26.
  • A friend who texts guilt trips is not one you would keep three months later Ajay Mehta, Tolan45:26.
• Reach people through every surface, including the app itself. Source: Steve P. Young and a developer with a six figure app exit13:27
  • Run the same automations across push, email, SMS and in-app screens. For users opted out of everything, show a blocker screen in the app Steve P. Young and a developer with a six figure app exit13:27.
  • Use in-app events to reach existing users. They end on a visible date, which adds urgency Pelin Culha, Mobile Action5:30.
  • Standard prompts become noise. Life360 shows an animated dog on app open that asks to be petted, and tapping it opens the pet profile Giordano Contestabile, Life36019:30.

2.7 Where they disagree

Sources split on how to ask for push, when to sell, and how much emotion to use.

• Custom pre-prompt or direct system prompt.
  • Most sources show a custom screen first and the system dialog only after a yes Steve P. Young, App Masters1:17, Mirela Cialai, Zinio5:32.
  • William at Wager Lab tested a pre-screen against the system prompt and the direct ask won, with a 4.8 increase in opt-ins. Steve says he was wrong to recommend the pre-screen Steve P. Young, App Masters1:56.
  • An audience member also found higher opt-in with the default prompt than with a double opt-in Steve P. Young, App Masters6:03.
  • Test both in your app, since the two results point opposite ways.
• Offer first or educate first.
  • One app sends a 20% offer within minutes of the welcome email, and another within the first hour for users who skipped the trial Steve P. Young and Chaim, App Masters6:46, Steve P. Young, App Masters8:54.
  • Grammarly holds the offer until the fourth email, after three emails of education Steve P. Young, Grammarly15:01.
  • The fast offer fits users who already saw the paywall. The slow one fits free users who still need to learn the app.
• Guilt and loss framing.
  • Flame uses a loss-framed nudge, and Duolingo's streak saver is seen as positive because users value the streak An (Anup), Flame12:02, Jackson Shuttleworth, Duolingo56:38.
  • Tolan avoids "I miss you" tricks because they would not hold up three months later Ajay Mehta, Tolan45:26.
  • Duolingo's passive-aggressive inactivity message and crying Duo email became memes, and the team ran with them because the voice gave them permission Gina Gotthilf, Duolingo52:53.
• Win back lapsed users now or later.
  • At Chess.com about 80% of daily or weekly actives are existing users, so returning users matter as much as new ones. Make the returning experience excellent Albert Cheng, Chess.com33:30.
  • Faire advises waiting on resurrection until earlier funnel efforts are exhausted, because churned users already decided against the product Dan Hockenmaier, Faire24:32.
  • Georgiana Laudi says it blows her mind how many companies have no win-back flow at all Georgiana Laudi, Forget The Funnel42:50.
• Never test a single email, yet one subject line test doubled results.
  • Elena Verna says a one-off email rarely causes lift because too few people open and click it Elena Verna1:12:20.
  • The same email with only the subject line changed produced a 2x difference in workout completion Steve P. Young, App Masters6:13.
  • Both agree the message matters. The difference may come from how large your audience is and how big the change is.