Quiz of Koko — from idea to live product
How Quiz of Koko went from a real learning problem to a live, AI-powered vocabulary game — the same FDE loop I now run for clients.
The challenge
Vocabulary practice has a motivation problem, not a content problem: dictionaries and flashcards exist everywhere, yet learners quit within days because nothing adapts to them and nothing is at stake.
The goal was never “another quiz app”. It was to prove that a small, sharply-scoped product — adaptive questions, real-time duels, weekly stakes — can keep a learner coming back without a content team or a marketing budget.
The FDE loop, run for real
- 1
Scope — the honest cut
One job done completely: help a Persian-speaking learner keep and grow vocabulary. Everything that did not feed that job went to the kill list — a scope discipline now sold as the first step of every client sprint.
- 2
Design — riskiest assumptions first
Adaptive difficulty and duel mechanics were designed before pixel-polish: if the questions did not feel personal and the duels did not feel alive, nothing else mattered.
- 3
Build — AI-accelerated, human-judged
The product was built the way the MVP Sprint builds today: AI for speed on the well-understood parts, human judgment on architecture and the learning feel — shipped to real users in the browser first, then Android.
- 4
Launch & learn — instrumented, not decorated
Analytics tied to outcomes (retention, duel participation, review completion) instead of vanity counters — and the growth plan runs against pre-registered gates, so every decision has an evidence rule.
What shipped
Adaptive AI questions
Personalized question generation that tracks each learner’s level and gaps.
Koko Plus — smart review
Spaced-repetition vocabulary review, scheduled by what you actually forget.
Live 1v1 duels
Real-time head-to-head matches — the stake that makes practice feel like play.
Weekly cup & leagues
Bronze→Champion weekly leagues — a reason to come back every week.
Continuous diagnosis
The app learns where you slip and re-tests exactly there.
Evidence, not adjectives
- The product is live and verifiable right now: free in the browser and on Android — no screenshots-only “case study”.
- Growth runs against pre-registered gates (day-1 retention, day-7 retention, conversion, CAC). The gates are public discipline; numbers get published when the evidence is honest and the owner approves.
- The entire investor funnel shipped through the same loop: an 11-slide pitch deck, an investor page, and a 19-post bilingual series documenting the deck slide by slide.
The full story, on this site
Want this loop on your product?
A free 30-minute scope call gets you the honest cut: the real problem, a one-page scope, and a build / don’t-build verdict — even if we never work together.