Outdoor Ad Impact Forecaster
Awards: Grand Prix, Best Use of Research Budget (Winner), Brand Advertising Effectiveness (Highly Commended) and Research and Data Innovation (Winner)
MeMo² observed an important transformation within the domain of marketing analytics, namely: the transformation from ‘Descriptive Analytics (What happened?) to ‘Prescriptive Analytics’ (How to shape the future?). This transformation has become a motivation for the start of an innovative big data & machine learning project.
Since 2014, Exterion Media has invested significantly in offering advertisers data and learnings regarding their outdoor campaigns. Post-campaign results concerning the realised Impact and uplift in Awareness and Purchase Intent were shared with more than 275 advertisers. The market however, shows a strong desire to receive insights before their campaign starts.
This was the starting point to develop a powerful online tool to predict the quality and impact of (Digital)Outdoor campaigns. Based on machine learning and the underlying results of over 300 studies conducted since 2014, which included 68.000 respondents and a variety of brands and industries, MeMo² has created a state-of-the-art and easy-to-use tool to forecast the impact of (Digital)Outdoor campaigns.
After validation in 2017, the Outdoor Ad Impact Forecaster was launched in 2018, and successfully used for more than 100 cases. In May 2018 the tool was awarded with a prestigious AMMA Research Award. Tool is currently public for jury only (https://exteriontool.memo2.nl/ login:MeMo2 pass:Abrit00l2018)
In MeMo² & Exterion Media started an unique collaboration with the VU University Amsterdam, 300 studies (N=68.000) showed the strongest drivers of Brand Lift realised via Outdoor advertising (see Whitepaper). This journey proved to be a solid start for a new and innovative approach to predict and improve outcomes of Digital Outdoor campaigns. MeMo² and Exterion Media started an extensive modelling and machine learning project. Main goal was to develop a powerful online tool that supports advertisers with Digital Outdoor campaign optimization. The tool needs to deliver useful and reliable predictions and recommendations before the start of an (Digital)Outdoor campaign.
Input for the forecasts are Creative, Brand and Media characteristics.
Important methodology steps undertaken:
- Coding and controlling the research data.
- Data enrichment with campaign information, e.g. impressions, number of panels, timings and audience.
- Detect and select strongest predictors with a Python image processing script.
- Built an Extreme Gradient Boosting (XGBoost) predicting model
- Development of an interactive tool in which users can upload their 'own' ad and fill in their campaign characteristics. Based on this input, the model forecasts the effects of the campaign (realtime) and provides the user with recommendations.
3. Key results
The predictive validity of the tool is tested by calculating the correlation coefficient between the prediction of the Forecaster and the factual results measured post-campaign. The conducted verification showed a high correlation (0.75) between the forecast output of the Outdoor Ad Impact Forecaster™ and the results generated via control measurements. This test confirmed the power and predictive value of the tool’s output. In other words, the tool proved to be a solid and thorough instrument to successfully predict the impact of (digital)outdoor ads.
Important outcome: impact of (digital)outdoor is for 40% attributable to Creation, 30% Brand and 30% Media.
4. Impact and Application
• Exterion Media and MeMo² succeeded in developing and launching an intelligent and user friendly tool, that is able to predict the impact of an (Digital) Outdoor campaign. The Outdoor Ad Impact Forecaster™ is an online machine learning tool, that offers powerful validated predictions, taking into account the most important variables that could influence the effect of (Digital)Outdoor advertising.
• Currently the Outdoor Ad Impact Forecaster™ is implemented within the research products offered by Exterion Media. Since 2018 the tool is used for the (Digital)Outdoor campaigns of more than 100 advertisers. Digital Out-of-Home Experts of Exterion Media use the tool to provide clients with actionable, trustworthy and independent recommendations, before campaigns run.
• Forecasts and recommendations coming from the tool are highly appreciated by Media and Research experts working at launching partners and key clients such as Coca-Cola, Takeaway.com, Tele2 and ING.
• The Outdoor Ad Impact Forecaster™ is awarded for an AMMA Research Award in May 2018. Jury: “…The jury appreciates the cooperation between the different disciplines, the combination of research methodologies, the implementation of brand, campaign and creative data in order to predict campaign effects. Besides this, the jury admires the manner how the results are shared with the market, via a free accessible and use friendly online tool that delivers on-demand insights and recommendations.“ https://www.mwg.nl/foto/normaal-Hans%20du.jpg
• Independent global agency MeMo² stopped offering its 'traditional' Outdoor pretesting methodology and has become a frequent user of the Outdoor Ad Impact Forecaster since the launch.
• 48h delivery promise. After filling in an online form (https://www.outdooradimpact.com/home) and uploading the (D)OOH ads, forecasts and recommendations are delivered within 48 hours. This quick response offers marketeers a data solution for short-term optimization of their DOOH campaign strategy.
Locally and globally, there are lot of words and publications about Big Data Analytics and Machine Learning. However the number of successful initiatives within the domain of digital media are limited. MeMo² and Exterion Media experience that only a small number of initiatives or scientific publications focus on the factual Brand Lift of (Digital)Outdoor Advertising. MeMo² and Exterion Media started something new. This tool is called a 'game changing analytics solution'. The Outdoor Ad Impact Forecaster is a perfect example of how research and data can contribute to ‘Prescriptive Analytics’ that are useful for professionals working in Digital Marketing.