Skip to content

← Home
Startup

Nairobi's mentor experiment

A class, a neighbour or an AI: what two Kenyan experiments found about advice · 14 min

In 2014, 372 young businesswomen in Dandora, Nairobi, were given a university business class, a month of visits to an experienced local businesswoman, or nothing. Only one of them raised profits, and it was not the class. Nine years later, 640 Kenyan owners tested an AI mentor on WhatsApp, and it pushed them apart. Whose advice should you take?

Takeaways

  • In Dandora, Nairobi, in two thousand and fourteen, three hundred and seventy-two young businesswomen were split into three groups: a university business class, a month of weekly visits to an experienced local businesswoman in the same trade, or nothing. Over the next seventeen months the mentored women earned about twenty percent more profit on average; the class changed habits but not profit. The gain faded as the pairs stopped meeting.
  • In two thousand and twenty-three, six hundred and forty Kenyan entrepreneurs joined a trial; half were offered a GPT-four business mentor on WhatsApp. On average, nothing changed. Those already doing well may have gained more than fifteen percent; those doing badly did close to ten percent worse. The difference was which advice they chose to act on: the strugglers took the generic items, like discounts and adverts.
  • Something to try: before Friday, find one person who runs the kind of business you want to run, in your own market, and ask one specific question about costs or suppliers. No one to ask? Walk the market, or ask an AI with your specifics and refuse the generic answers.

Chapters

  1. Ten dollars a month0:17
  2. Find a mentor2:06
  3. Not wisdom. Addresses.3:33
  4. The mentor that never sleeps5:41
  5. Same question, different street9:00
  6. One question about costs10:30
  7. The answer12:23

References

  1. Brooks, W., Donovan, K. & Johnson, T. R. — Mentors or Teachers? Microenterprise Training in Kenya(2018)
  2. Brooks, W., Donovan, K. & Johnson, T. R. (NBER conference paper, full text) — Mentors or Teachers? Microenterprise Training in Kenya
  3. J-PAL (Abdul Latif Jameel Poverty Action Lab) — Local Mentorship and Business Training to Increase Profits among Microenterprises in Kenya
  4. Otis, N. G., Clarke, R., Delecourt, S., Holtz, D. & Koning, R. — The Uneven Impact of Generative Artificial Intelligence on Entrepreneurial Performance: Evidence from a Field Experiment in Kenya(2026)
  5. Otis, N. G., Clarke, R., Delecourt, S., Holtz, D. & Koning, R. (Management Science) — The Uneven Impact of Generative Artificial Intelligence on Entrepreneurial Performance: Evidence from a Field Experiment in Kenya(2026)
  6. Otis, N. G., Clarke, R., Delecourt, S., Holtz, D. & Koning, R. (Harvard Kennedy School CID, working-paper record) — The Uneven Impact of Generative AI on Entrepreneurial Performance(2023)
  7. Counts, L. (Berkeley Haas Newsroom) — Gen AI field experiment shows mixed results in helping small businesses grow(2024)
  8. Branson, R. (Virgin.com) — The importance of having a mentor in business(2014)
  9. Clarkson, N. (Virgin.com) — Richard Branson: How to find a mentor(2015)
  10. Branson, R. (Virgin.com) — Finding the right mentor(2022)
  11. McKenzie, D. — Small business training to improve management practices in developing countries: re-assessing the evidence for 'training doesn't work'(2021)
  12. McKenzie, D. (IDEAS/RePEc record with abstract) — Small business training to improve management practices in developing countries: re-assessing the evidence for 'training doesn't work'(2021)
  13. Cai, J. & Szeidl, A. (NBER Working Paper 22951) — Interfirm Relationships and Business Performance(2016)
  14. Cai, J. & Szeidl, A. (The Quarterly Journal of Economics) — Interfirm Relationships and Business Performance(2018)