Can data analytics improve record-keeping for African small businesses?

Small and medium enterprises (SMEs) are the backbone of African economies, accounting for over 80% of employment in many countries in sub-Saharan Africa. Yet most operate without systematic records of their sales or inventory. This limits their ability to decide what to stock, when to reorder, and which customers to extend credit to.

The record-keeping puzzle: Collecting data, not making sense of it

The conventional explanation is that businesses don’t see the value of keeping records. Training programmes that nudge entrepreneurs to maintain records and improve overall business practices have been tried across the developing countries, often with limited success. 

The puzzle is that most entrepreneurs know that record-keeping matters – in our study of retail SMEs in Nairobi, over 85% said they considered it at least moderately useful. Despite this, nearly 40% kept records less than daily, if at all.

We propose a different explanation: the problem may not be collecting data but making sense of it. A spreadsheet full of transactions tells you very little unless you can analyse it. A small business owner managing a shop, serving customers, and dealing with multiple suppliers at once may hardly have the time or ability to extract patterns from raw data (several studies suggest that small firms in developing countries are indeed labour-constrained). If entrepreneurs cannot learn from their records, the returns to keeping them are low, and so they do not bother.

Providing a point-of-sale app to businesses in Nairobi 

To test this, we partnered with Technoserve Kenya to run a randomised controlled trial among over 2,000 retail SMEs in Nairobi. We provided businesses with a free smartphone-based point-of-sale application, mSpark Basic. It allowed users to record sales and inventory digitally, view the entries in a spreadsheet format, and view total sales revenue.

After a two-month warm-up period, we randomly upgraded half of the app users to an enhanced version, mSpark Analytics. This version required no changes in data entry, but changed what the app did with the data: it automatically generated simple charts and plain-language summaries of sales patterns and popular products.

The key question was whether reducing the cost of learning from data would change how much businesses used the app and, ultimately, how they ran their businesses.

Easier insights resulted in better record-keeping

We found evidence supporting our hypothesis. Businesses with access to the analytics version of the app used it around 25% more intensively than those with the basic version. They were also 7-9 percentage points more likely to record sales and inventory on a daily basis.

The effect was largest for businesses that had previously not been keeping records at all. Among non-recorders, access to analytics raised the probability of recording sales in the app by 20 percentage points.

Figure 1: Does the app improve record-keeping? 

Crucially, the gains were concentrated among entrepreneurs who had cited time or difficulty as their main barriers to record-keeping. These are precisely the businesses that stand to gain most from automated analytics, as generating insights themselves is too costly. For this group, the analytics app roughly doubled the rate of regular in-app record keeping.

Early signs of change in business strategies

We also tracked whether the app changed how businesses operate. When we compared users of either version of the app to a control group that had no access to the app, we found that app users showed markedly better inventory knowledge, diversified their inventory portfolio (without increasing their inventory investment), and reported significantly higher daily sales.

However, we did not find significant short-run improvements in sales when comparing analytics and basic app users. This is unsurprising: the upgrade was rolled out only 6-8 weeks before our final survey, leaving little time for new insights to affect performance. It is plausible that improvements in business performance may accrue over time.

What does this mean for policies supporting SMEs?

Our findings carry a clear message for policymakers and development organisations supporting SME growth: training entrepreneurs to keep records is not enough. 

What matters is helping them learn from those records. Providing tools that automatically translate raw data into simple summaries can meaningfully increase record-keeping, especially among those facing the greatest time and capacity constraints.

This does not require expensive infrastructure or highly educated users – the analytics app worked on standard smartphones, and the analytical work was done for the entrepreneur. As mobile technology becomes increasingly accessible across sub-Saharan Africa, there is a real opportunity to embed automated learning support into the digital tools SMEs are already adopting. The barrier to learning from data, it turns out, is not the data itself – it is the cost of making sense of it.

This research was funded by the International Growth Centre (IGC) and Private Enterprise Development in Low-Income Countries (PEDL).

This blog is part of a series highlighting research supported by the Small and Growing Businesses (SGB) Evidence Fund, which generates evidence on the policies, programmes, and investments that can help firms grow, create jobs, and contribute to economic development.