What "pay for improvement" actually looks like
A few months ago, I wrote that ed-tech should be judged by one thing: are students actually improving?
A few people asked whether we genuinely measure this at Efekta.
Fair question.
So here’s a chart from a team meeting this week.
It tracks tens of thousands of Rwandan teachers learning English with Efekta. (For context: Rwanda switched its national language of instruction from French to English. Every teacher now needs to be able to teach in English. And those that need to improve their English are doing so with Efekta.)
We measure proficiency continuously using EFSET, mapped to CEFR.
Five months in, the cohort average has already moved from A2 (Elementary) to B1 (Intermediate) across all four skills: Listening, Reading, Speaking & Writing.
A few things matter here:
- This is the daily average across the entire cohort — not a cherry-picked success story.
- All four skills improved together. Speaking and writing — usually the hardest skills to shift — moved as strongly as listening and reading.
- Our target is B2 proficiency (50+ EFSET). We are nearly there. The job now is finishing.
To me, this is what “pay for improvement” looks like.
Not logins.
Not hours.
Not seats.
Daily learning gains, by skill, at national scale.
Because Rwanda didn’t want to buy access to software.
They wanted a workforce capable of teaching in English.
That’s the bar.
So here’s my offer to governments: you guarantee adoption. We’ll guarantee outcomes.
Originally posted on LinkedIn.