Daniel Winterstein, founder and CTO at AdTech firm Goodloop, argues that a nuanced, selective AI could help marketers through global crises
When you have a hammer, it is tempting to look for the nail in every situation. This temptation doubles when the tool is as powerful and multi-faceted as AI.
Much is currently being said about AI and how it’s helping with solutions during pandemic e.g. in contact tracing, maintaining social distancing, or the search for a vaccine. AI is also helping some companies to cope with commercial disruption, for example Walmart has turned to robot cleaners. Behind the scenes, smart delivery management has helped keep e-commerce cope with increased demand.
So what about marketing?
Along with many sectors, the marketing industry has been badly disrupted. Events based marketing has been hit hard. Nor has online been immune: Numerous campaigns, months in planning and preparation, have been cancelled as not appropriate for lockdown. Meanwhile on the supply side, publishers are seeing a jump in readers — but a fall in revenue.
However for the marketing industry, AI doesn’t have anything to offer. Yet.
Why not? Because AI is driven by modelling the past – it excels at analysing past data, preferably from regularly repeated situations. When there’s no historical data to go on, or when the world has shifted and the old data is no longer representative — then AI is lost and unreliable.
You can run an AI on poor data — and people do. But don’t expect good results. In a crisis, stock-market traders are advised to turn off their AI systems, and for good reason.
Recent years have seen phenomenal progress in AI. Yet for flexible intelligence, there’s no substitute for human understanding.
Still, there are lessons we can learn, and ways in which AI can help should we find ourselves in a similar crisis situation in future.
And unfortunately, it seems all too likely that the situation will reoccur. Covid-19 is not the first deadly coronavirus outbreak — there was SARS in 2003, MERS in 2012. No doubt there are other viruses out there which could unleash a pandemic.
Next time though, we will have some data to guide us. With a basis in comparable data, predictive analytics and AI tools could be invaluable. Let’s look at the challenges covid-19 and lockdown have raised within marketing – and how with preparation, AI could be used next time.
One of the first effects within advertising, was a widespread cancelling of adverts by many companies — better to be safe, than risk upsetting people at this time. A lot of email marketing was also switched off for the same reason. This was not unwise — there are lots of examples of automated email systems that sent out inappropriate messages, accidentally exhorting people to break lockdown rules.
AI could help us be more selective — filtering adverts to flag which ones are inappropriate for a crisis mood, and which ones might conflict with lockdown rules. Conversely, we also have data on the kind of adverts that actively appeal to people during a lockdown.
A perverse effect of the pandemic has been a growth in readers for quality journalism — but a fall in revenue. This is because many advertisers have blacklisted “coronavirus” and related keywords — which pretty much every article of the last few months mentions.
This is reminiscent of a keyword blacklisting issue that arose when ISIS were in the news. ISIS were also known as Islamic State, or IS for short. So “is” was (briefly) on some advertisers’ blacklists — who then found it hard to place any ads.
By using more advanced AI analysis of news articles, crude keyword blacklisting can be replaced with more nuanced and emotionally-intelligent screening. AI can help us understand whether or not it’s appropriate to place your brand advert next to an article that mentions coronavirus, or similar “bad” words. This would be good for both publishers and advertisers.
Looking behind the ads to the industry that makes them — much of what we do is digital, and can carry on regardless of lockdown. However working-from-home has disrupted workflows and staff. Automated scheduling and task-management tools can provide more dynamic workflows — allowing companies to handle such disruption smoothly.
Finally, let’s consider how advertising can actually help in these bad situations. Advertising is not just for shoes and soap — it is also a key tool for public information. Governments and health-services have been using advertising to distribute information on coronavirus. The marketing industry is not as crucial as medical staff and ventilators, but it does play an important part in supporting healthcare, Smarter and better advertising systems can help spread information when it is most needed.