AI is a regular feature of news stories across nearly every industry. But, says Stephen Upstone, the headlines hide the fact there's a big human role in making AI work effectively.
From healthcare, to driverless cars, to cyber security, AI is increasingly seen as the Holy Grail for innovation, productivity and profitability. At the other end of the spectrum, nay-sayers preach the coming obsolescence of humans with the rise of the job-killing robots. But in reality, the use of AI – in marketing and advertising, at least – is not immediately as clear cut as the sensationalising headlines suggest. The truth is, it will take a human to get the best strategic use out of AI.
Ensuring that AI technology is a good fit for marketers
Building on advances such as programmatic advertising and data-driven marketing, advertisers can build further momentum by incorporating AI to streamline day-to-day campaign management and reporting. This is all well and good in theory – but doing this effectively in practice is another story. The reality is that high-level strategic thinking behind the application of AI for marketing remains the domain of the expert human marketer.
Aligning the AI with a specific objective should be your starting point. Are you pursuing brand awareness? Or are you looking to optimise to other brand outcomes further down the funnel, like brand affinity and consideration? Think about the KPI you are working towards, as AI is likely to have greatest impact on lower funnel metrics.
From our experience of working with clients, AI can be up to 19 times more effective at driving purchase intent than traditional advertising optimisation. Data scientists aspire to an AI that can ascertain where each potential customer is along the growth funnel, and optimise by the campaign objective that is most appropriate to drive them along the path to purchase. We are not there yet, so for the foreseeable future, human experience is still required to align AI capabilities with brand objectives.
AI softwares really add value when they use dozens of factors including time of day, weather, browsing behaviour, current and previous location, to build predictive models and optimise accordingly. The AI has to be “fed” with new campaign data to refine targeting over time and to enhance campaign performance from day to day. AI is the logical enhancement of data-driven marketing.
The best way to demonstrate the effective use of AI by marketers is through examples, so here are a couple of our own.
Case Study: driving purchase intent
A well-known tech brand successfully used AI solution to drive purchase intent within the scope of a new product release. With an embedded video and a rich media format, the aim was to see which of the two yielded the best results in driving intent among the target audience.
For these two formats, the AI solution was used to assess purchase intent, asking potential customers whether they would consider buying the new product. Potential customers’ responses were fed back into the AI engine to optimise towards positive responses. This was also combined with rich data collection of positive audience profiles – this helped build a custom audience segment for future campaigns.
With the AI solution, the rich media unit produced an uplift in purchase intent of 54%. The embedded video led to an uplift of 55%. The rich media actually recorded a higher purchase intent metric than previously, which meant that it was a more effective buy than embedded video.
Our verdict is that AI’s greatest value is obtained when it is used to feedback into campaign strategy, helping you recognise which channels and media formats are working hardest.
Case Study: driving footfall
A second example comes in the form of a luxury automotive brand. The brand set its sights on reaching their in-market shoppers to drive visits to dealerships. They were especially interested in getting high CTRs and VCRs, along with developing detailed audience insights to inform marketing activities more broadly.
Combined with externally-sourced location insights, the AI solution optimised towards those consumers with the highest propensity for footfall.
The campaign resulted in above-benchmark performance in terms of CTR and VCR. Footfall to dealerships rose by 42.4% – which translated into over a thousand incremental dealership visits. Putting this in context, their average vehicle is valued at over $28,000 – the ROI is clear.
Strategic use of AI necessarily requires thoughtful use of data to power the predictive models. Algorithms work their best when nurtured with rich data, and location can help bridge the gap between online and offline marketing – if this forms part of your strategy.
In the case of marketing, the kinds of decisions that are needed to really harness the power of AI are still the preserve of the human strategist. As the tech evolves, it will be able to make increasingly sophisticated decisions on its own. But until that time comes, it will take a human to get the best out of AI.