How do you accurately measure effectiveness of some of the below mentioned promotional campaigns?
- Product advertisements in catalogs, flyers and other forms of print media
- A multi-touch loyalty program for high value customers
- Product markdowns to a customer segment
- Impact of a new digital marketing program
- Impact of increasing radio advertising spend
The obvious answer is some form of predictive modeling technique as it remains the popular choice across industries to measure ROI of campaigns and do budget optimization. There are many predictive modeling techniques that can help answer some of these questions. For example, marketing mix modeling is a predictive analytics technique which is used to measure the effect and forecast the impact of various marketing initiatives on sales and ROI. Using this approach, you shift your spending from low ROI marketing campaigns to the ones which yield more ROI.
In our opinion, the more effective way of measuring campaign effectiveness is using the collective power of predictive modeling and rapid in-market experimentation. For instance, techniques like marketing mix modeling can help generate hypothesis based on robust statistical analysis of historical data while experimentation or test & learn can help when it comes to identifying & isolating cause-effect relationship thereby providing the much-needed accuracy before taking critical budget decisions.
Most of the popular predictive modeling techniques focus on establishing correlation and then extrapolating results to come up with a forecast. For e.g. sales going up in a holiday season does not necessarily imply that it was due to the increase in promotional spending. It could very easily be because of the uplift in consumer mood which in turn increases purchases. Predictive modeling assumes that the models based on historical data will hold true in future as well and will continue to affect key metrics in the same/similar way. Any change in the internal environment such as change in product packaging, or in the external environment such as new competitors entering the market, will impact key metrics and will not figure in the analysis that is based only on historical data. Such internal and external changes are inevitable, and at times unknown, but they are critical to account for to establish causation.
Rapid experimentation adopts a test-control approach which helps in clearly isolating the cause-effect relationship. Experimentation techniques also help isolate the impact at the customer segment and market level. So, a retailer can make more informed decisions and can use tailored roll-outs. With Experimentation, analyses accounts for all changes, known and unknown, happening internally as well as externally. Depending upon results, you can optimize roll-out which will maximize your ROI.
When you combine predictive modeling with experimentation, you can experience compounded effect which provides strong hypotheses on campaigns, empirical evidence of consumer behavior, isolates cause-effect relationship, maximizes ROI, and promotes a culture of innovation and improvement across the organization.
About Trial Run: Trial Run is a cloud-based product that lets companies conduct business experiments on sites, markets and individuals before they implement. It helps companies scale their experimentation capability efficiently and affordably by providing insight into which decisions will work and which will not.