Malaysian Media Awards Winners Showcase





2021 | |

Business Analytics Engine – Marketing Mix Modelling on steroids!

Advertiser: QSR Stores Sdn Bhd
Brand: KFC
Creative Agency: Reprise
Credits: -

Objective & Challenge

With food & beverage competition and competitors coming up fast, business was affected. Making the situation worse, Covid hit Malaysian just when we were recovering. With sales affected fast, media budgets also dropped fast but we still had the same target to hit. How on earth were we going to do it?”

Insight & Strategy
Naturally as the media agency, we anticipated extreme optimisation of investment. Clients would want to be assured that we buy into most efficient and effective media that drive sales. Most importantly how do we instil the confidence into KFC that our media plans have been optimised that would also promise return of sales to the business? Our strategy is to leverage Business Analytics Engine (BAE), media agency’s analytical tool to assess media performance, measuring them against business KPI Sales and to create a media strategy that drives sales.

Execution
Using the concept of Market Mix Modelling, BAE helps decompose marketing and nonmarketing factors role in sales contribution for the past 104 weeks. We aimed to : identify media channels that are the most effective and efficient in driving sales further optimize these mediums in future campaigns In building a robust model that can explain sales drivers, all factors that may affect sales in any way were considered namely marketing spends, competitor marketing activities, website visits, search trend, brand health metrics such as buzz/ talkability, consideration, recommendation, ads awareness etc, launch/ promo startdate, seasonality amongst others. With a robust modelling plot, agency studied the impact of media investments to sales by week, by medium by product. The final suggested media mix is a result of a few rounds of simulations and optimisations towards sales. Firstly, we input data of the recommended plan down to detailed such as weekly media investments by medium and get a predictive sales number. Next we optimised the plan by shifting budget amongst the recommended media mix with an ultimate aim to get the best sales result with same media investments. Most often, medium that contributes most to sales will likely see adjustment in investment at the expense of those lesser contributing ones. These results are then matched back to the actual sales received on a weekly basis. For weeks that overperform, we analysed the performance based on media allocation. For weeks that underperformed, we noted on the weekly behaviour other sales drivers. Agency would run the model and optimise the plan on a weekly basis realtime to drive effectiveness and sales. Our analysis also covered the effect of preCovid,during Covid and postCovid and how are the respective impact on KFC sales including  which medium that are more effective and efficient in each different phases. So let’s say, the BAE model needs to output a plan starting Nov for 8 weeks with a media budget of RM1million and a sales KPI of RM100 million. To achieve this, BAE has advised us to tweak the current mix to Digital 49% (+4%), TV 40% (+1%), OOH 11% (minus 5%). Today, agency has in hand hygiene level media spends, cost optimization and estimated sales return after rigorous study on the sales impact. We have been also running systemized test and learn models to further improvise and drive confidence amongst clients.

Effectiveness
Across a period of 60 weeks, our recalibrated exercise improved the model’s confidence by another 4.5% which is huge given KFC’s billion dollar business. The model accuracy against actual sales are within healthy +/ 10% range and as a result of this, we are able to advise client on approximately 8% to 10% savings for every campaign given our sales prediction. Through optimisation exercise, agency improved the ROI by +52% in Q4’20. (by means of spending lesser media dollar while enjoying better effectiveness of contribution to sales)