In this interview, we sat down with Michael Wamberg, head of AI at Nexta, to discuss how marketers can boost ROAS (return on ad spend) in retail media. He expands on the benefits of self-service ad management, as well as how retail media campaigns can be simplified and improved. Wamberg also explains the importance of considering consumers’ online research behaviours before they make in-store purchases. 

What are some of the benefits of self-serve ad management? 

Self-serve ad management can be an empowering tool for marketers. By eliminating the need for third-party interaction, campaign management can become a more efficient and streamlined process. Direct control over campaigns gives marketers much greater control and autonomy over campaigns, while cutting costs and saving time. 

For retailers, saving time is also a significant benefit: the time taken to launch campaigns can be reduced considerably, as well as minimising operational hurdles. As a result, greater opportunities for efficiency and scalability are opened up. 

How important is ROPO (Research Online, Purchase Offline) in the retail media landscape?  

ROPO – which refers to a pattern where consumers research products online but make their purchases in physical stores – is essential for both brands/marketers and retailers to consider. As we know, many consumers conduct online research before making a purchase. Generally, the more expensive the purchase, the more likely a consumer is to spend time online investigating brands, contemplating models, comparing prices and analysing reviews. 

Delving into the more technical side, tools for collecting offline sales data, such as loyalty programmes, can make it possible to integrate the effects of ROPO into retail media attribution models. This not only ensures a holistic view of customer behaviour, but also facilitates fairer ROAS benchmarking when comparing retail media performance with platforms like Google or Meta. 

How can first-party data be leveraged best in retail media? 

When it comes to retail media, first-party data is a game changer. Again, both brands and retailers stand to benefit. Unique identifiers derived from first-party data allow brands and retailers to leverage advanced attribution models to compute ROAS with precision. First-party data allows the customer journey to be mapped across various touchpoints, which makes campaign performance much easier to understand. As a result, brands can be more effective when it comes to decision-making.

How can campaign management for retail media campaigns be simplified? 

The predictive modelling approach can certainly provide clarity. Predictive modelling harnesses historical data and statistical techniques to build models which are able to predict future customer behaviour such as purchase likelihood, campaign response, and churn rate. The latter, for example, uses queues such as how many times a product is used or the number of complaints made to a company’s support team in order to predict the percentage of customers who will stop doing business with it. By identifying customers who are at risk of churn, brands can be more proactive about their customer retention, targeting these specifically in certain personalised marketing efforts. For other customers, tailored marketing messages based on individual preferences can also be generated. 

Having reliable predictions on customer behaviour allows marketers to make data-driven decisions. Models which forecast impressions, clicks, and conversions, also enable smarter budget allocation and maximise campaign impact.

Additionally, the use of unified APIs can simplify campaign management significantly. With connections to the global advertising ecosystem – including platforms like Meta, Google Ads and TikTok – unified APIs enable live updates to budgets, creatives, and targeting, while providing immediate insights into campaign performance. 

How can the efficiency of retail media campaigns be improved?  

Automated budget optimisation can play a significant role in driving efficiency for retail media campaigns. At Nexta, we utilise predictive models within a Bayesian reinforcement learning framework to automate budget allocation. A Bayesian reinforcement learning framework revolves around the notion that by leveraging prior knowledge and continuously updating ideas with new observations, learning efficiency can be optimised and performance improved. Implementing this framework is particularly useful in the marketing industry, which is highly dynamic and uncertain. 

Using this framework allows us to see how budget adjustments ripple through the entire conversion funnel, ultimately determining the optimal campaign budgets across channels completely automatically. Consequently, we ensure that resources are allocated in the most impactful way possible.