Many people at one or more times in their life growing up may have been told to talk less. The phrase ‘Silence is Golden’ may have been used in tandem with that.

The same thinking and phrase holds true in analytics. Modeling data for the telecom industry is always a challenge and there are many scenarios that highlight success when it happens. We have been testing numerous approaches for telecom companies during recent years and have had good successes for various companies. The key to testing is to capture some winning combinations that make the marketing efforts pay off in a big way. Applying models across different files with different offers and price points has been a continuous exercise. Some methods show great response rates with weaker conversion rates. Then if the focus is too much on conversion rates, the response rate becomes too low. As sales managers often say, “If the phone doesn’t ring, it really doesn’t matter”.

Earlier this year, we had a detailed discussion around some telecom data we were receiving. We spent some time asking detailed questions and taking a deeper dive into the information. The data discovery process gleaned some learning not previously found. This led us to define a lot of noise in the data which watered down the result of modeling. Campaigns were marginal but we still needed a winner. Modeling resumed once the noise was removed and the right universe was selected. We looked at the results and there standing tall was a clear picture of the target. Removing the noisy data created change to the criteria for universe selection as well as for the model itself.

Results… We got a puzzling call from the client asking us what we did differently to the data. When asked to clarify the question, the client explained that they hadn’t seen results like this in many years and that in deed a homerun was at the plate. The results were now coming in at four times the previous rate and the next question was, “How much more of this data can you provide?” As a colleague of mine says, the devil is in the details. So before you dive too deep into modeling, make sure you spend adequate time in data discovery and that all of the right questions have been asked.

Once you remove the noise… Silence is Golden.