Adaptive mastery-streak nudges for learner retention
Khan Academy's mastery streak is one of the most loved features in online learning — and one of the easiest to break the moment a learner has a hard week. We built an adaptive nudge layer that personalizes lesson length to a learner's energy curve, lifting 14-day retention 19% for opted-in cohorts.
- 14-day retention
- +19%
- Learners enrolled
- ~1.2M
- Personalization latency
- <60ms
- Languages live
- 12
Good morning,
Maya
You're ~12 minutes from today's goal. Pick up where you left off.
Today's plan
Adapted to your energy curveSolving systems of equations
DNA replication review
Renaissance: finish unit
Subject mastery
7-day trendKhanmigo noticed something
Yesterday you got stuck on quadratic substitution. I queued a 4-minute warm-up before your Algebra lesson today.
Recent quiz scores
Linear systems
Today · 7:42 am
Cell respiration
Yesterday
Industrial Rev.
2 days ago
Polynomial factoring
3 days ago
Badges
12 of 48Class assignment
From Ms. Chen
Quadratics: word problems
Due Friday · 12 questions
3 of 12 complete
The brief
The mastery streak is Khan Academy's quiet superpower — a daily learning loop that millions of students use to stay on track. It is also fragile. A student has a busy week, the streak resets, and the system has just penalized exactly the kind of week that needed gentler scaffolding, not more pressure.
The bet: an adaptive nudge layer that personalizes lesson length and mix to each learner's recent energy curve would protect the psychological benefits of the streak without punishing the messy weeks that life actually involves.
What we built
A real-time personalization service that scores each learner's "energy curve" — recent session length, accuracy, breaks, time-of-day patterns — and recommends a daily plan calibrated to what they will probably complete, not what an idealized version of them would. The service returns a plan in under 60 milliseconds, well inside the page-load budget, and is updated continuously as the learner progresses.
The Khanmigo integration was small but important: when a learner is visibly struggling on a recommended item, Khanmigo proactively offers support instead of waiting for the learner to ask. The threshold for "struggling" is itself learned from the learner's own history.
We launched against a tightly-bounded A/B with strict guardrails: no suggested plan can be shorter than what the learner has demonstrated the ability to complete; no streak can be silently extended. The system recommends, the learner decides.
Outcome
14-day retention for opted-in cohorts lifted 19% versus the control. ~1.2 million learners are now enrolled. The personalization service is live in 12 languages.
The most quietly important outcome: cohorts with previously-broken streaks recovered them at a meaningfully higher rate after the launch — the system is doing what it was designed to do, which is to help learners come back after the hard week instead of giving up.
"We did not want to gamify the kid into doing more — we wanted the system to meet them where they were."
Keep exploring.
Book a callA modern policyholder portal for a 75-year-old insurance co-op
Co-operators is one of Canada's largest insurance co-operatives — 600+ locations, generations of customer trust, and a legacy self-service experience built around PDFs and phone trees. We rebuilt the policyholder portal around real-time claim status, digital ID cards, and a single-tap "file a new claim" flow that absorbs roughly a third of the inbound contact-centre volume.
- Inbound call reduction
- ~28%
- Time-to-claim filed
- 4 min
An "Ask your vault" AI layer for wealth-management advisors
FutureVault sits at the centre of how RIAs organise client documents — estate, tax, insurance, identity. We built a retrieval-grounded AI layer on top, letting an advisor ask plain-English questions across an entire household's document history and get cited answers in seconds instead of digging through folders.
- Document lookup
- 3.2× faster
- Citation accuracy
- ~96%