Here is the reading by Jack and Suri on the Economics of M-Pesa. This is their 2011 paper.
They also have a 2010 paper with the same title which you probably should not bother with ...
In the 2010 paper they had a random sample of 3000 households (survey was in April 2008).
In the 2011 paper they also report results from a follow up survey conducted in 2009. However, the follow up-survey has only 2016 of the original 3000 households. So when you compare the tables in the 2010 paper with Round 1 survey results in the 2011 paper, the numbers may not be identical due to the difference in the sample size. Some other details/pieces of information/observations made by the authors in the 2010 piece are omitted from the 2011 piece ... I don't think they are critical.