Imagine you could ask your data a question. And it answers you in real time!

The question

Our client – a top 5 consumer goods company, wanted to apply Machine Learning to deduce real time insights from multiple data sources to answer end user questions. Answering these questions today requires vast amount of person-hours to extract and analyse the data, as well as specialised analytical skills.

So, put quite simply, the question set by the client was:

Q – “How can we speed up the interrogation of multiple data sets, to be able to act on it in real time and answer end user questions?”

Q – “How can we ask natural language questions and get real time answers/insights around critical KPI’s?”


Our approach

We set OKRA’s machine learning algorithms to work on the data and within 2 weeks, OKRA had a deep understanding of the information.

A smart search/response user interface was created and deployed within weeks, that enabled users to ask natural language questions using text or voice inputs.

Here’s a few examples of how the data was interrogated:

“What is the overall performance of [Brand X]?”

“How is [Competitor X] performing, by each brand?”

“Why is [Brand X] losing (or gaining) market share?”

“What is the predicted market share for [Brand X]?”


The results

A deep understanding of brand and market performance  in real time was established, rather than weeks/months of work with a team of analysts.

Blind testing was performed by OKRA to demonstrate accurate predictions of the preceding quarter’s top line KPIs, including:

  • % penetration
  • Price positioning
  • Market share
  • Brand equity

Prior knowledge of these critical factors meant this global business could now implement speedy responses to market changes that could have a hugely positive impact on sales.