Re-engagement Emails: Duolingo's Playbook for User Retention

A re-engagement email is an automated message sent when a user stops doing the thing your product is for. The ones that work remind the user of something they have already built. Duolingo's streak reminders are the famous example: "keep your streak alive" is a far better reason to come back than "we miss you." Pair them with progress emails for active users, such as milestones, "you're close" nudges, and weekly summaries, and you have the core of an email program for user retention.

Below: what Duolingo has actually published about its reminders, which triggers to use and when, copy patterns with examples, and how to tell whether an email caused anyone to come back.

What has Duolingo actually published about its reminders?

Duolingo is the standard example because it has described its retention system in unusual detail. Here is what the company and its people have said:

• Retaining current users was the biggest lever. Jorge Mazal, Duolingo's former chief product officer, wrote that its growth model showed current user retention rate had "5 times the impact of the second-best metric" on daily active users. Acting on that, alongside other bets, grew daily active users 4.5x over four years.¹
• Email is part of the streak system. "We remind learners to keep their streak by sending personalized email reminders," Duolingo wrote in 2017. The same post reported that a "streak wager" feature improved day-7 retention by 14% in testing.²
• The first week decides the habit. Learners who reach a 7-day streak are 2.4 times more likely to use Duolingo the next day than learners without a streak.³ Duolingo also says they are 3.6 times more likely to complete their course.⁴ Both figures compare groups rather than prove cause, but they show why the first week gets so much attention.
• Reminder copy wears out. Duolingo picks the wording of its daily practice reminders with a bandit algorithm. Its researchers found that templates a learner hadn't seen recently performed better, so they penalized reusing a template. The system raised total daily active users by 0.5% and new-user retention by 2% "over a strong baseline."⁵
• Reminders stop after a week. "We actually stop sending notifications after seven days of inactivity," CEO Luis von Ahn said in his 2023 TED talk. The last one reads: "Hey, these reminders don't seem to be working. We'll stop sending them for now."⁶ Duolingo's retention lead, Jackson Shuttleworth, explained why: after seven days, the value of reaching out again "is significantly lower."⁷

Duolingo's welcome email shows the same thinking from day one. It gives a single first goal, a 7-day streak, and tells the learner why it matters.

A word on those numbers. They come from a heavily tested product with about 15 million daily active users on a streak of a year or longer at the end of 2025,⁸ and most blend push notifications, email, and product changes. None shows one email type multiplying retention tenfold. Treat them as evidence that the approach works, not as a benchmark to expect.

Re-engagement email vs. win-back email

The two terms get used interchangeably, and email tools draw the line in different places. The useful split is whether the customer relationship is still active:

• Re-engagement emails go to people who still have an account, and often still pay, but have stopped using the product. The lever is habit and progress: a streak, a half-finished project, data they would lose track of.
• Win-back emails go to customers who have lapsed: they canceled, let a subscription expire, or, for a store, haven't bought in longer than their usual buying cycle.⁹ The lever is change: what's new since they left, a fix for the reason they left, and sometimes an offer.

Both are lifecycle marketing emails, not receipts, so give them a clear unsubscribe path. Duolingo's reminder footer reads: "You are receiving these emails because you are subscribed to Duolingo practice reminders."

Which triggers should start a re-engagement email?

Every trigger should come from product data you already store. These are starting points to test, not benchmarks:

To pick your own timing, look at returning users. How many days usually pass between one session and the next? If most active users come back within three days, someone silent for four is drifting, and that is when the first email should go. Sending on a fixed day 3 means nothing for a tool people open once a month.

Stalled setup deserves special attention. A user who never reached first value has no streak to protect, so this email overlaps with onboarding. Our welcome email examples cover the first-value step in detail.

Copy patterns that bring users back

Every re-engagement email needs the same three parts: what the user has, what they are about to lose or leave unfinished, and one action to take.

1. Protect what they've built

This is Duolingo's streak pattern, and it works for anything that accumulates: a streak, savings, saved work, a reputation score.

Subject: Your 23-day budgeting streak ends tonight

You've logged your spending every day since March 2. That's $1,240 tracked and one savings goal hit.

One entry keeps the streak going. It takes about a minute.

[Log today's spending]

The email makes no threat and uses no guilt. It shows the user their own investment and makes the next step small.

2. Pick up where they left off

People dislike leaving things unfinished. If a user started something, name it and show how far along it is.

Subject: "Fitness App" is 67% done

Last week you finished the login screen and the workout log. Two screens are left before you can share a preview link.

[Open Fitness App]

A progress bar does a lot of work here. Duolingo's early reminders showed exactly this: the learner's level, the next level, and "133 points away from level 7."

3. Give new users a reason when they have nothing to lose

A user who never activated has no progress to protect. Instead of "come back," remove the thing that stopped them: a template, sample data, a shorter path, or a person to help. If you add an incentive, tie it to reaching the first result, not just to logging in.

Subject: We set up a sample dashboard for you

You signed up on Tuesday but haven't connected any data yet. To show you what you'll get, we loaded a sample dashboard with demo sales numbers.

[See the sample dashboard]

When you're ready, connecting your own data takes about five minutes. Reply if you get stuck and a person will help.

4. The honest last email

When reminders stop working, say so and stop. Duolingo's practice reminder notifications end after seven days of inactivity, and the last one admits they aren't working. The company has also sent an email in the same spirit:

Subject: Should we stop sending these?

We've sent you a few reminders and haven't heard back, so we're pausing them. Your projects and settings are saved.

[Pick up where I left off]

Von Ahn's summary of what happens next: "They come back." Stopping also protects delivery. Gmail asks bulk senders to keep reported spam below 0.3%,¹⁰ and every reminder to someone who has stopped responding is a chance for a complaint. The email deliverability guide covers the rest.

Progress emails: keep active users moving

Progress emails keep active users from drifting in the first place. Milestone and "you're close" emails fire right after something the user did, so they arrive when motivation is highest. A weekly summary runs on a schedule instead, which lets it also reach people who have started to slip.

Research on workplace motivation by Teresa Amabile and Steven Kramer found that making progress in meaningful work was the single biggest boost to people's motivation during a workday.¹¹ Progress emails apply the same idea: show people they are getting somewhere.

Three progress emails worth building:

• Milestones. First report created, 10th invoice sent, 30-day streak. Send it right after the action.
• "You're close." The user is near the next milestone or level. This is the Duolingo 2014 email above.
• A weekly summary. What the user did this week compared with last week, plus one suggestion. Duolingo sends a weekly progress email built on that comparison. It also goes to learners in the first few weeks after they stop practicing, since an email has more room than a push notification to show their progress.⁷

Each should answer three questions: what did I just do, what does it mean, and what should I do next? The last matters most. "Congrats" alone ends the loop; a next step keeps it going.

Subject: You sent your 10th invoice

That's $8,400 billed through Ledgerly since June. Customers paid your last three invoices in an average of 6 days.

Next step: turn on automatic reminders so late invoices chase themselves.

[Turn on reminders]

How do you measure re-engagement emails?

Measure whether the email caused people to come back, not whether they opened it. Opens are unreliable because Apple Mail Privacy Protection loads email content, including tracking pixels, before the reader looks at it.¹²

1. Define "came back" as a key action. A lesson, a report, a sent invoice. Not a login and not an open. Duolingo's research counted a lesson completed within two hours of the reminder.⁵
2. Keep a holdout group. Randomly exclude 5-10% of eligible users from the email. Many users return on their own (Duolingo's researchers noted that a large share of lessons are organic), and the holdout shows what would have happened anyway.
3. Compare over a fixed window. Pick a window that suits the trigger, such as 7 days for "going quiet," and compare the share of each group that completed the key action.
4. Check the longer term. Are re-engaged users still active 30 days later? Did any upgrade?
5. Watch the costs. Track unsubscribes and spam complaints per send. If a reminder brings back 1% of users but annoys 3%, rewrite it or cut it.

A worked example with round numbers: 2,000 users go quiet this month. You hold out 200 and email 1,800. Within 7 days, 16% of the emailed group completes a key action, compared with 11% of the holdout. The email's real effect is the 5-point difference, about 90 extra users, not the 288 who returned.

A starter plan for this week

1. Define your key action and measure your natural return gap.
2. Ship one "going quiet" email with a 10% holdout.
3. Ship one milestone email for the first meaningful win.
4. Add a stop rule: pause reminders after a few sends with no response.
5. After 30 days, review lift, unsubscribes, and complaints, then add the next trigger.

If you want these as a timed sequence, the drip campaign guide shows how to lay out behavior-triggered steps and exit conditions.

Where Dreamlit fits

The re-engagement triggers above can use the activity your app already saves in Supabase or Postgres. Dreamlit, the company behind this blog, lets your AI agent turn those rules into email workflows, draft the messages, and show you a preview with real user data before you publish.

You still choose the return action and measure whether the reminders help; the workflow itself isn't evidence of a retention lift.

Start a 7-day free trial⁠ or connect your agent.

Footnotes

1. Jorge Mazal, "How Duolingo reignited user growth," Lenny's Newsletter (February 2023), https://www.lennysnewsletter.com/p/how-duolingo-reignited-user-growth⁠ ↩

2. Duolingo Blog, "How Streaks keep Duolingo learners committed to their language goals" (May 2017), https://blog.duolingo.com/how-streaks-keep-duolingo-learners-committed-to-their-language-goals/⁠ ↩

3. Duolingo Blog, "Improving the streak" (November 2020), https://blog.duolingo.com/improving-the-streak⁠ ↩

4. Duolingo Blog, "How Duolingo's streak builds habit," https://blog.duolingo.com/how-duolingo-streak-builds-habit/⁠ ↩

5. Kevin P. Yancey and Burr Settles, "A Sleeping, Recovering Bandit Algorithm for Optimizing Recurring Notifications," KDD 2020, https://research.duolingo.com/papers/yancey.kdd20.pdf⁠ ↩ ↩²

6. Luis von Ahn, "How to make learning as addictive as social media," TED2023, https://www.ted.com/talks/luis_von_ahn_how_to_make_learning_as_addictive_as_social_media⁠ ↩

7. Sub Club podcast, "How to time reactivation campaigns for maximum impact" with Jackson Shuttleworth of Duolingo, https://subclub.com/episode/how-to-time-reactivation-campaigns-for-maximum-impact-jackson-shuttleworth-duolingo⁠ ↩ ↩²

8. Duolingo, 2025 Annual Report on Form 10-K (filed February 2026), https://investors.duolingo.com/static-files/f19d76fb-dee4-4f13-96ae-138ebfd0f2d3⁠ ↩

9. Klaviyo Help Center, "How to create a winback flow," https://help.klaviyo.com/hc/en-us/articles/115002775192⁠ ↩

10. Google, "Email sender guidelines," https://support.google.com/mail/answer/81126⁠ ↩

11. Teresa M. Amabile and Steven J. Kramer, "The Power of Small Wins," Harvard Business Review (May 2011), https://hbr.org/2011/05/the-power-of-small-wins⁠ ↩

12. Apple, "Use Mail Privacy Protection on iPhone," https://support.apple.com/guide/iphone/use-mail-privacy-protection-iphf084865c7/ios⁠ ↩

---

Frequently asked questions

What is a re-engagement email?

A re-engagement email is an automated message sent when a user stops doing the main thing your product is for, such as logging in, finishing a lesson, or creating a report. The strongest ones name something specific the user already built, like a streak, a project, or saved data, and ask for one small action.

What is the difference between a re-engagement email and a win-back email?

The terms overlap, and tools draw the line in different places. A re-engagement email usually goes to someone who still has an active account but has gone quiet. A win-back email goes to a customer who has lapsed: they canceled, let a subscription expire, or, for a store, haven't bought in longer than their usual buying cycle. Re-engagement leans on habit and progress; win-back leans on what has changed since they left and, sometimes, an offer.

When should I send a re-engagement email?

Send it just after the gap in which an active user would normally have come back. Measure how many days usually pass between sessions for users who return. For a daily habit app that may be the same evening; for a weekly work tool it is closer to seven days.

How does Duolingo use email to keep learners coming back?

Duolingo sends personalized practice reminders that reference the learner's streak, and its welcome email sets a first goal of a 7-day streak. It picks reminder copy with a bandit algorithm that favors templates the learner hasn't seen recently, which its researchers reported raised daily active users by 0.5% and new-user retention by 2%. It stops practice reminder notifications after seven days of inactivity.

How many re-engagement emails should I send before stopping?

For a weekly product, two or three is a sensible start: a reminder, a second email with a different angle, and a final note saying you'll pause. Duolingo, a daily habit app, stops practice reminder notifications after seven days of inactivity, and the last one says the reminders don't seem to be working.

How do I know if re-engagement emails are working?

Hold back a random 5-10% of eligible users who get no email, then compare how many in each group complete a key action within a fixed window. The difference is the lift the email caused. Opens are not a reliable measure because Apple Mail Privacy Protection preloads images.