Time Series Uplift Modeling - how to test scenarios to forecast new media channels success

When planning on innovating in media, many marketers are stopped in their path by a simple question “Will it works? and “How high will the ROI be?”. Investing massive amounts of money in unknown territories may sound thrilling but risk-averse marketing board members and fragile business conditions have led many companies to refrain from innovating drastically.

How to test innovation securely, to envision results when scaled? Here is an extremely useful approach that data-driven marketing can bring to innovative marketers.

A time series uplift modeling - using regional test scenarios to forecast the success of new media channels - can help convince of an innovative strategy.

This approach provides a clear view of the effect of a particular media on your KPIs for business success and helps to decide if an investment on a larger scale would be promising.

How does Time Series Uplift Modeling work?

1. At first, the development of the interesting key indicator in different regions is analyzed
2. Findings are used to define the baseline on which a test can happen: Two similarly equipped regions are selected and both media activity and business success indicators can be measured in that defined region.
3. The innovative channel activity will be implemented exclusively in one of the two regions, leave the second as a control group.
4. At the end of the activity, the expected development without media activity can be compared to the actual results and the effectiveness of the tested channel can be calculated

Using this solution to test on a smaller scale is beneficial to both the advertisers and the market.

• it offers to test new trends without risking the entire media budget.
• It equips marketers with proven results when presenting innovations internally
• it permits flexible publishers and media vendors to prove their positive effect on business results to encourage diversity on the market

What is needed to make this work?

• In order to obtain a meaningful result, ideally, data should be available for 1 or 2 years. At a minimum, a few months of activity is necessary to build the correlation between advertising and the main business success indicator
• Two similar regions in terms of business development
• Clear business indicators that are responsive in the short term and can be followed with granularity at both regional level and over a period of time
• A budget to dedicate to the test.

Done!

Except if you do not have an internal analytics team. In this case, we can help! At MMT, we offer such Time Series Uplift Modeling within our MMT Scope environment. The entire process, from the creation of the test scenario to the modeling and presentation of the results, is accompanied and implemented by our experienced data intelligence team in an advisory capacity.