IBM’s ground-breaking Watson artificial intelligence software is now, for the very first time, being used to create better semantic targeting for native advertising.
The new integration is the latest innovation from the ADYOULIKE Group, a native Technology Platform and Network, which allows brand advertisers to scale their native advertising campaigns across premium and niche publishers with ease from a single point of entry. Julien Verdier, CEO and co-founder of ADYOULIKE, discusses the new integration, AI and the potential it has to change the way audiences interact with ads in the future.
To those that say that AI will make many jobs redundant, what is your reply?
I think that any kind of technical innovation makes many jobs redundant, but at the same time it creates new opportunities to create brand new jobs and brand new industries. As far as Adyoulike is concerned, AI will enable us to do what a human brain is able to do, bringing an unrivalled level of targeting on each and every ad campaign we run.
Watson has been used in native advertising. How does that work and how far does it go?
That’s a good question, I would say it’s stronger than Google Ad Words and its potential can go even further. AI in native advertising, in my view, is a game changer. It has the ability to give you a big picture analysis for each and every page where we've generated an ad. Watson gives us these details in real time at huge scale, giving us a relevancy percentage on keywords, entities, IAB categories (which are the industry’s guidelines and categorisations), concepts, sentiment of the page, whether it’s positive, negative or neutral and emotions such as anger, disgust, fear, joy or sadness.
This analysis of thousands of pages a second, means we perfectly know the context of each and every page of our network. This means we can serve the right ad in the right context. When we create a native campaign, we agree with the advertiser on the relevant keywords it wants to be associated with, and target those keywords when we run the ad. For instance, if a brand is launching a new lipstick, we can guarantee that we can target our publisher pages where the keywords “lipstick” “makeup” “lips” and “beauty” will appear.
We can also target our publisher websites, which are part of the Adyoulike network, matching the IAB category fashion/beauty/cosmetics/lipstick and picking only the pages that have a positive sentiment score above 70 per cent.
Is a commercial rollout of any sort planned? Have other companies been approached?
The commercial rollout is going to be massive. We want to empower every advertiser and marketer in the industry with this tool because we think that users deserve better, more relevant ads that they actually want to see and interact with. This will in turn bring better return on every dollar or pound invested by the advertisers because of this sophisticated targeting.
In the long run, we think that the Adyoulike-Watson partnership will be the norm in an industry looking for increasingly deeper contextual targeting. Companies unable to integrate that level of AI in their platform will be marginalised and finally die.
Are AI systems like Watson more suitable for native than for other forms of digital advertising?
Native advertising is all about being in the middle of a content feed and highly visible. The power comes from the editorial content and the native ad endorsing each other. In other types of digital ads like banners, the ad placement is next to the editorial content and not inside. Content and ads stand at a defined place and don’t endorse each other, which makes contextualisation less critical. In classic display advertising such as banners, pre-rolls etc, contextualisation is a nice to have. For native ads it’s a must have, and advertisers have been asking for that for a long time.
What does Watson do that regular targeting cannot?
Watson obviously wasn’t meant for advertising originally. However, its potential is beginning to be realised across the marketing industry. It has an unrivalled ability to understand the content of a page in a very deep way, from emotions to sentiment, categories to keywords.
This goes beyond regular advertising targeting, which is more about device, capping and website and socio-demographic categories that are often poorly relevant. Everyone does regular targeting, but its limitations mean irrelevant ads and bad user experiences. This can be actually part of the reason that ad blocking is on the rise.
What impact will AI-driven advertising have on publishers?
It’s going to be very positive. Currently, publishers are facing an uphill battle. The quality of ads that come from external partners that run on their pages must be high, as publishers know that their value lies in their editorial content and they don’t want it to be wasted by poorly targeted ads. Additionally, they can’t let ads that negatively impact on their brand to run, which is an issue that has been known to happen.
AI integration will mean publishers can set a blacklist to prevent any ads on pages that aren’t appropriate – such as dealing with sensitive issues about politics or war. And they can be sure that external ads will be targeted correctly, adding real value to the user’s journey.
Are there any risks involved in going this way (determinism, no serendipity etc)?
I would say that there are risks not going this way! For too long, ad targeting has been poor at best. We believe that every brand has a story to tell, and the Adyoulike-Watson partnership enables us to tell that story to the right people and in the right context.
Because native ads are integrated with the look and feel of each and every page and user journey, we cannot rely on just being lucky to run our advertisers' ads: they have to be super targeted, or they will just not work. Ad tech distribution is now a science. We have so many data points, billions of ad impressions and hundreds of campaigns run each month, that we cannot make the most of the targeting opportunities if we don’t use Watson.
What does the advent of AI mean for the programmatic market?
Programmatic and native led to native 2.0, which meant we could serve creative ads at a huge scale. What Watson’s integration means is that these creative ads will be served at scale, but will also be highly targeted to each individual user. This is phenomenal.
For example, we’re able to understand the context of each and every page that a user browses, then for each page we give a semantic relevancy score that we aggregate along the user’s browsing journey. In the long run, based on a person’s web browsing, we know what they care about and like, and we can make sure that the ads served to them are pieces of content and information that they will genuinely enjoy seeing and find useful. This data is currently aggregated in our DMP, which we give to our advertisers and publishers. However, soon we will open up these data clusters to other DMPs, and every advertiser will be able to access our semantic data cluster even if they don’t run their ads with Adyoulike. This will mean that even banner and pre-roll video campaigns will be better targeted and offer an improved experience.
In the long run we are expecting a big uplift for the whole ad industry, not just native campaigns. We hope to finally see the light at the end of the tunnel, and it will be the end of 20 years of poorly targeted ad experiences in our industry.
Image source: Shutterstock/Aysezgicmeli