I-Com Summit Experience 2020: Artificial intelligence and machine learning for marketing decisions

Day 11 of the I-Com Summit Experience saw marketers and digital media experts addressing the talk of the town in digital – artificial intelligence (AI) and machine learning (ML) – for ‘decisioning in marketing’.

 

Here are excerpts from their talks:

 

Matt Andrew, managing director, Ekimetrics, UK

 

“Execute (campaigns) for how users use the platform. While <6-second videos are up to 15 times more efficient, vertical formats are 1.4x more efficient. Reach more than 10% of the target population. Large reach campaigns are up to 6.8x more efficient and automatic placement is 13x more efficient. Allow the algorithm to do the job."

 

Emily Hoffman, product marketing manager, Salesforce, USA

 

“The modern growth marketer needs an agile mindset to balance expectations across the customer and business. Their key goals are to give a customer experience and get efficiency and growth. In customer experience, they need trust, personalisation, engagement, loyalty and advocacy. They need to adapt quickly and data helps you navigate with confidence so you can always position your business for growth. Data and measurement connect the dots between every single campaign and how it resonates. Marketers are facing functional data challenges:

 

1: Lack of a unified view of marketing investment

2: Lack of real-time performance insight to optimise ROI and results

3: Lack of alignment and governance to drive collaboration across teams and regions.”

 

Lise Regnier, director, data science and engineering, Warner Bros., France

 

“Good data science is not the same as good data science for business. Data science is a tool that solves problems. Good data-science for business takes into consideration the context and can be used for all. A data scientist’s job is a trade-off between minimising loss functions and maximising usage and impact.”

 

Camille Bernard, agency measurement partner, Facebook, UK

 

“How do we use ML on Facebook? Advertising on Facebook’s platforms can be very granular and tactical. What we observed is that typically, multiple dimensions of media activation will affect campaign effectiveness.” 

 

John Pickles, digital marketing optimization, programme lead, FORD, UK

 

“The exciting bit about data is that it's the gift that keeps giving. The key idea is that marketers know that all interactions on a website are not equal. We have to understand and quantify the difference between a customer that wants a test drive or is just looking at a product. “

 

Paul Martin, VP - marketing science, Xaxis, UK

 

“Explore and take advantage of all data. You don't have to settle for only DSP data. Watch the trade-offs between more data and more features. Faster feature-rich data with small outcomes can be less useful than data with large outcome volumes, but slower and fewer features. Choose a metric which reflects business outcomes. You don't have to settle for the usual digital media outcomes.  We can modify programmatic with first-party data. Adapt data usage to reflect different contexts. Think carefully about onsite visitor behaviour and tune AI models to inform programmatic buying accordingly.”