Tracking graduates of job training programs is increasingly important, given high youth unemployment and the need to understand labour market outcomes to inform policy and program decisions. We describe the strengths and weaknesses of using text messaging to survey 2180 graduates of Vocational Training Centres in three Kenyan counties. The response rate was 18 percent. The short message service survey over-represented urban graduates and those from private Vocational Training Centres, and 39 percent were answered by someone other than the intended respondentseriously jeopardizing data quality. We report lessons learned and offer recommendations for designing graduate tracking surveys.