Pilot II · 6–27 July 2026

We ran the study before we made the claim.

85 newcomers in Sweden practised speaking with Kompis every day for three weeks. Here is what we measured, what we did not, and where the evidence is thin.

85

participants baselined

46

median conversations per person

46

median conversations per participant

51%

still active in the final week

What the numbers say

Daily use held up for three weeks

Language tools usually lose people in week one. The interesting result was not day one. It was day twenty-one.

Active in week 1100%
Active in week 272%
Active in week 3 (final)51%
Returned after going quiet12%
Reported lower speaking anxiety−40%
Matched pre/post participants29
Median conversations per participant46
Study length21 days

Self-reported measures come from 29 participants who completed both the pre and post survey. This is a small, self-selected cohort with no control group. The numbers describe what happened here, not a proven effect.

The full report

The complete public report: design, findings, participant voices, limitations and what we do next. The PDF is the same document.

Download the PDF