From ‘basic applications’ to ‘dynamic applications’

Predictions for 2017
This has been a seminal year for artificial intelligence (AI), with much of the media hype – and consequent ‘replacement anxiety’ which it generated – now quelled and replaced with a more practical understanding of AI’s potential role in = everyday life. This includes real-time language translation, improved product recommendations, advanced image search, as well as applications in industry such as data diagnostics, predictive analytics, and of course automation. As people place more importance on personalised and seamless shopping experiences, it pays to use AI to unlock the power and insight that big data has promised to deliver.

What happened in 2017?
The age of general AI is here, with increased adoption over the past year. According to the Vanson Bourne ‘State of AI for Enterprises’ study, 80% of enterprises now have some form of AI (such as machine learning or deep learning) in operational use today. Globally, 42% of enterprises see lots of room for further implementation and process integration, but 91% see
barriers ahead – lack of IT infrastructure (40%) and lack of talent (34%) being the most significant.

The media industry is embracing the potential of AI. The growth in programmatically traded media has allowed us to deploy AI techniques both internally and as part of wider client solutions. At Zenith, we have applied advanced machine-learning  techniques to Aviva’s digital campaigns, improving the cost per quote (CPQ) by 6% for search and 10% for display. In a world first, we are passing our unique conversion score back into DoubleClick Bid Manager (the demand-side platform used to make programmatic purchases from Google’s digital inventory) and have fully optimised our buying.

What’s next?
2018 will be the year that some companies begin more strategic implementation, and start realising more of AI’s benefits. But for most, 2018 should still be about exploring the more basic applications of AI as it advances and becomes easier to deploy using ‘off the shelf’ solutions. More customised deep-learning use-cases will begin to appear that demonstrate the transformative power of more specialised AI on business processes across a broader range of well-funded projects, as well as its cost-saving potential.

AI will be able to meet consumers’ growing demand for built-in ‘immediate’ service. Until now m-commerce has been regarded as standard shopping over the smartphone, but with AI we can expect automated commerce that offers dynamic experiences to fulfil the customer journey. In 2018, the customer journey will quickly encompass virtual-reality purchasing,
dynamic pricing, automated voice recognition and digital attribution, which can be managed within a single system and with a natural-language interface. The dynamic experience supported by AI will allow for a more ‘in-the-moment’ adaptive shopping experience.

What does this mean for marketers?
AI is a broad topic and has almost universal application across business in 2018. While on a macro level AI is the engine that drives tech giants, on a micro level ‘off the shelf’ solutions make AI actionable, accessible and applicable to many business tasks. For example, online German e-tailer Otto used Blue Yonder’s CERNdeveloped tech to crunch 3bn past transactions and 200 variables (such as past sales, searches and weather information) to predict what customers will buy a week before they order. With an accuracy of 90%, this technology allowed Otto to confidently pre-order items and reduce returns by 2 million items per year. This is the kind of magic AI can perform, but not only in logistics.

AI is not just having an impact on media, it is also helping to drive creativity. It was striking that at this year’s Cannes Lions Festival, not only were people talking about AI as the next big thing, it was already a big thing in the award entries & winners. AI in its many guises was the magic that made many campaigns exceptional – none more so than the one from Shanghai General Motors, which created a unique ID – “MyCarPulse” – for each of its 20 million cars. It used this ID to send hyper personalised messages to the car owners, based on an AI-powered profile including driving patterns, lifestyle and car condition, and databases from CRM, dealerships, OnStar and WeChat. It then used optimised dynamic messaging to more than double the number of service visits to improve car care.

Advertisers however need to recognise that for this magic to work they need powerful data, which needs to be set up correctly for machine learning. While there’s nothing sexy about naming conventions, data tagging and compliance, it is imperative that these become a business priority, otherwise everything will remain dumb.