The Espresso Index

There’s a raging debate on whether Nairobi residents can sustain the mid to high-end businesses being setup in malls and other “middle class” neighbourhoods. What better way of tracking purchasing  power of Nairobians than monitoring  the conspicuous consumption of coffee 😉 .In the spirit of being humorous and factual, I give you The Espresso Index – a tracker for the number of customers indulging at Java House Africa outlets.

Doodle Art of Java House Logo by @MarkRenja

As a coffee addict, I kept most of my receipts from Java House. Within the receipts, there is a number succeeding the  letters CHK that indicate the number of customer you are at the time of purchase. Although the number sometimes overlaps between days, we can the estimate number of customers per day through data normalization and extrapolation.

Let’s get started with Java Junction which I do have a good number of receipts. The process would be to get the CHK number, estimate number of people served per hour then add CHK number to number of hours remaining in the day multiplied by no of customers per hour. The following graph (index) is born.

The graph looks ragged because of very little data used – with more data, it should smoothen out and show trends. However, we can still deduce that the outlet serves on average of 650 customers a day. If each customer spends on average Kshs 500, that translates to Kshs 325,000 a day and Kshs 9.75 million a month.

Take this analysis with a grain of salt, a lot of assumptions and estimates have been made. Example the outlet processes 20 customers an hour and customer arrival is uniform across the day (which isn’t usually the case).

Inspired by Caffe Con Leche Inflation Index


Are you an influencer looking to utilize your skills or a  brand seeking connection to the right influencer ? Blue Bear App matches businesses to online influencers and provides  an analytical platform to manage online campaigns.  Engage them today of their portal


  1. This could have been another great article from you, but I think it was rushed. Imo, it’s incomplete…unless there’s a part two or it’s apart if a mini series.



    1. Hi, thank you. Indeed it is incomplete. The idea is to introduce the idea and have other “series” once there’s enough data.



  2. Cant help but feel given the sample size, a T-distribution would have have given you a wider and more accurate berth to play in. But also safe to assume eventually a normal distribution of customers served can be assumed?



    1. Data from Foursquare checkins shows distribution changes by day of the week making it difficult to have a distribution to go by.

      Initially, I hypothesized customer arrival follows a normal distribution but it varies with the day of week. Still thinking on how to solve the problem.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: