KG Labs × PEAK × DFID · E-commerce Acceleration · 2022
The Russian-language project summary lists the numbers and the cohort. This is the witness-voice English version — why the model worked, who it was for, and what scaling it would actually require.
2022-08-15 · Aziz Soltobaev · KG Labs Foundation
The Russian-language project summary from August 2022 — at /2022/08/08/second-ecommerce-accelerator/ — covers the structural results of the second KG Labs × PEAK Kyrgyzstan e-commerce acceleration programme that closed that summer. Six weeks of programme, twenty-two participants, aggregate sales growth of six million som — roughly seventy-four thousand three hundred US dollars at the time — through e-commerce channels alone.
This post is the English-language witness-voice complement. The numbers are the same. The reading of them is from inside the design and operation of the cohort.
What the headline number does not capture
The headline number is an aggregate. What the aggregate hides is the texture of where the growth came from. Six categories of participant produced the bulk of the six-million-som lift:
- Honey producers who shipped their first two hundred kilograms of trial honey to European buyers during the cohort. Not “agreed to” or “negotiated about.” Shipped and delivered, with the customs paperwork, the food-safety certification, and the cold-chain logistics all worked through inside the six weeks.
- Sewing producers who raised their minimum production-batch size from three hundred units per item to one thousand. A structural step up — that scale change opens the wholesale relationships that an individual atelier cannot enter.
- Marketplace sellers (on Wildberries and Ozon) who lifted their margins meaningfully without changing production volumes, by working through pricing strategy and listing-quality optimisation. Same product, same factory, more revenue.
- The Altyn Rysbekova oratorical-skills training company, which built out its lead-generation funnel during the cohort and stood up the skeleton of an online-course product. The cohort that month was a mix of physical-goods exporters and one services business; the services business benefited from being in the same room as the goods exporters.
- Sumolok producers — the traditional fermented spring drink, a product category that almost never appears in Bishkek startup-event materials. They lifted their export reach during the cohort.
- Ecoff.kg, the vegan sausage brand, which used the cohort to develop brand recognition that has carried them into the years since.
The cohort list in full appears in the Russian post above. Camelmilk.kg, Kurdash Workshop, Mir Instrumentov, CNC.kg, xpat4kg.pro, Gulum.kg, Lukoyono, and CPS Bishkek were the other named participants.
Why the format worked
What separated this cohort from a typical Bishkek acceleration was a deliberate design choice: pair the marketplace mechanics with real export-readiness work. Listing optimisation, image strategy, customer-response timing — those are the topics every accelerator covers. We added the layer most do not: customs paperwork, currency settlement, the economics of trial-sample shipping, the licensed-broker conversation for food products going to the EU, and the operational mechanics of using a regional logistics partner to consolidate shipments.
The cohort that could not handle the export-readiness side was filtered out by week two. The cohort that could turned the marketplace skills into actual shipped revenue inside the six-week window. The filtering matters: an accelerator that admits everyone produces an unfocused six weeks of sales-training that the strong participants outgrow in week one.
The institutional structure that made it economically possible was a combined PEAK + DFID funding stack. PEAK Kyrgyzstan brought the operational delivery and the existing accelerator infrastructure. DFID funded the programme. KG Labs Public Foundation co-led the cohort design and the mentor pool.
What this is the prototype for
The KG Labs view on what this cohort produced, three years on, is that this is the model that should scale in the Kyrgyz acceleration landscape — and it has not, structurally, because the funding stack that made it possible is unusual.
Six weeks is the right length: short enough that participants do not lose their day jobs; sharp enough that unserious applicants self-select out by week two; structured enough that the sales-growth metric is concrete and defensible at funder review.
The constraints to replicating it are not technical and not curricular. They are funding-stack constraints. PEAK + DFID was a specific moment of donor coordination that does not repeat by itself. The conversation with the country’s donor community about the next replication is the next conversation worth having — and is the kind of policy work I would rather participate in writing than read about not having happened.
What the official summary captures, and what it does not
The Russian-language post from August 2022 captures the cohort list, the headline number, and the photo set from the Demo Day. It was written in the closeout phase of the project, when the cohort was still fresh and the photos were freshly developed.
What it does not capture, by editorial choice at the time, is the filtering that produced the cohort, the format that produced the six-week sales lift, and the funding-stack that made the programme economically viable. Those are the parts of the story that travel — that someone in a different country, looking at a similar set of producers, can use to design the equivalent programme in their context.
This post is that part of the story. It is the witness-voice version. The structural summary lives at the Russian-language URL above.
— Aziz Soltobaev, KG Labs Public Foundation, May 2026.