Building the measurement gap from digital to offline

Connecting the digital and physical worlds is becoming increasingly critical for retailers to measure the ROI on marketing accurately.

Online and offline worlds are fusing more and more by the day. Advertising budgets are increasingly focusing on marketing online, yet online sales only accounts for approximately 7% of retail with traditional bricks and mortar stores still dominating the sector. With more pressure than ever on retailers to provide accurate evaluation of campaigns and allocate valuable marketing dollars efficiently, online-to-offline attribution has found itself in the spotlight. Marketers are under pressure to understand the way different advertising channels conduct attribution and to get transparency about the procedures working behind the scenes.

Attribution measurements have always had two key components; the target of success and how this target is linked back to the initial advertisement.

The goal in an offline performance evaluation is to establish in-store revenue attributable to advertising. Simple right? Sadly not. Few large platforms offer POS data access or a link to aggregators of credit card data. While companies may be hesitant to share some of their most sensitive data with large data collectors, the indirect matching of online users to credit card data is often a black box for advertisers with little insight as to how the matching is conducted. Also, the enhanced security of payment card schemes, aggravates tracking individual customers through their payment details in-store.

As current industry conditions are great for large scale tracking, the alternative method of measurement is to track store visits. Australia continues to be one of the leading smartphone markets globally and as smartphone penetration approaches the 90% mark, nearly all consumers have a device that provides additional value to their offline shopping experience through location-based services. This location data can also be used to determine how many customers have visited a store after interacting with mobile ads.

With recent studies revealing 82% of smartphone users consulting their phones on purchases, the link becomes even easier. Of course, the ultimate goal is to not only determine store visit statistics but also how many consumers made that all important purchase. In our mobile wallet app Stocard, for example, we combine the store visit data with in-store loyalty card usage to determine whether a customer transacted in-store.

Very simplistic models for online attribution have been used for a long time. Last-click attribution is still widely adopted, and we have only recently seen a shift towards attribution models which distribute the generated value dynamically among several touchpoints. Traditional online attribution models do not work in an offline setting, as only a few interactions can be tracked. Additionally, most of them ignore the probability that a customer would have visited a store and purchased the advertised product without the advertorial interaction.

A successful attribution model must consider the base probability of a purchase and show an isolated effect from all other untracked touchpoints. The solution is A/B testing. In an A/B test all customers from the potential target group are randomly assigned to a treatment group and a control group, the only difference being the exposure or non-exposure to the advertisement. The difference in the conversion to a visit or in-store purchases between the two group shows the isolated effect of the advertising channel, as all other factors are equal for both groups.

Ultimately, no one wants to compare apples with pears and be left with a suboptimal allocation of marketing budget, so it’s important that the control group only is made up of customers that match the targeting criteria. Comparing the conversion rate to buy women’s fashion of a middle-aged housewife and a young male student would be biased and a wasted exercise. A/B testing allows for a precise Return On Ad Spend (ROAS) evaluation based on the true incremental uplift. However, it requires a large sample size and subsequently long observation periods to obtain statistically robust results. Therefore, the method is best suited to evaluating a whole marketing channel as opposed to individual campaigns.

Marketers should look for channels that offer offline attribution that truly measures incremental visits or purchases and at the same time be aware of the limitations of the different attribution methods. Mastering the offline performance evaluation will be one of the keys to successful marketing in a world that becomes increasingly digital but in which most customers still prefer to purchase offline.

Stocard is the leading wallet app in Europe and Australia and combines smart technology with maximum usability. A simple scan feature lets users scan every card and add it digitally to the app. Moreover, users can receive highly relevant offers from their favourite retailers. Learn more at stocardapp.com.