Advertisers are ahead of publishers in one simple area…

Partner content: Machine learning can hand control back to publishers, writes Ezoic’s John Cole – what’s stopping them?

I’m at the airport in Cologne, having just visited the aircraft-hanger-sized bonanza of digital marketing that is Dmexco, Germany. Differentiation in among all the parties in this ecosystem is getting harder and harder (“it’s a DSP, SSP and a DMP all in one…!”)

After spending days talking with digital aficionados, I needed a break to find out what’s been going on in the real world.

Advertisers have more control to personalise than publishers

Getting out my phone (iphone 6+, 4G, 4 bars), I typed into the browser (Chrome) bar ‘hurricane florence’. Then, I saw the news results from Google for the inbound storm due to hit the Carolinas.

After about two seconds, I arrived on the landing page of a well known news site. The content loads quickly and I begin to read. I’m scrolling slowly, absorbing the predicted track of the storm from an image and noting the times that it will hit various cities.

Mid-way through my browsing experience, the whole page refreshes as an ad arrives and I have to re-scroll down the page from the top again to regain my reading point. I reach the midpoint in the page and begin to speed up my reading, now only scanning the content.

This article is in-depth, but it is too much information. Pretty soon, I’m spinning down the page to the summary paragraph (12”-24” of rain in 2 days, with small trees being uprooted and roofs and sidings likely to be lost, citizens are urged to evacuate the coastal areas immediately.”)

I then close my browser. My session time was around two minutes.

This is a fairly typical browsing experience on mobile. No doubt, product teams at the digital publisher who attempted to construct my ‘user experience’ (UX) will be focusing on their broad objectives of ‘fast load time’, ‘lower bounce rates’, ‘higher session duration’ and ‘higher page views per visit’.

The editorial team who produced the newspiece are no doubt content that the article was ranked high enough in Google to be easily selected and that I was able to satisfy my search with their information. It was good enough; I got what I wanted and then I left the site.
[advert position=”left”]
The ads that I encountered in my session were worked up by the publication’s ad ops team, offering advertisers the opportunity to serve me an ad (via header bidding or via a real time bidding exchange). There was possibly a Data Management Platform in the background selling my cookie (mid 40’s, C-Level, UK male).

There also could have been a Private Marketplace Auction that happened because I opted into the GDPR model for ‘interest based ads’ when I arrived on the site.

How will this team be graded? The RPM of that page? The eCPM of the ads sold? The average viewability percent?

This is all very ‘normal’ UX / monetisation and the way it happens now…

Machine Learning can hand control back to publishers

The trouble with the current method is not that it’s overtly deficient, but lacks optimisation power. The next visitor that comes to that same article is likely see things exactly as I saw them, regardless of whether or not they possess the same attributes or interests as me.

With all of the data available – the advertisers are given an extraordinary opportunity to treat visitors differently.

Meanwhile, most publishers are still treating every one of their visitors the same.

While advertisers are finding ever more powerful ways to adapt in the digital era, publishers are still holding fast to rules from older forms of publishing. This includes fixed content, fixed layouts, fixed ad combinations, and ultimately – fixed experiences.

Articles are no longer printed. They don’t have to look the same for everyone. This is where A.I. is going to actually benefit publishers long term. It will help them overcome the asymmetry of data and information between publishers and advertisers.

Machine learning personalisation offers publishers the chance to gather all their ‘average’ user data and turn it into something truly unique – for each visitor.

The browser, the connection speed, the landing page, the upstream traffic source, the keyword, the time it took for the page to become scrollable (aka DOM interactive – if you want to look it up), the word density, scroll depth, time on page, viewport size, time of day and day of week, postal or zip code. All of these are publisher data signals that can be used to deliver better browsing experiences for all visitors.

How do I know all this? Because we’re doing it at Ezoic. We’ve watched how machines can ‘learn’ from visitors over time and make small adjustments that greatly improve objective user experience metrics.

What else have we learned? Happier users also produce much more money from advertising. Greater than 50% lift. This is because longer sessions mean more ads are shown, and higher engagement levels mean higher ad rates over time. It’s a compounding force – in the publishers’ favour.

Every scrap of anonymous visitor data can be fed into a machine and help the computer ‘learn’ what users might like best.

Does this mean we don’t need product people or ad ops people to call the shots? No. Quite the opposite. We need people to direct and understand how these technologies are applied. Technology can help us all make better decisions – even if the machines are making some of them for us.

As I learned at Dmexco; there is a lot of technology in this space that is indistinguishable from other tech (and overly complex to understand).

Ultimately, publishers themselves (the humans in this equation) will be at the heart of this move towards personalisation. Machines simplify the data-driven decisions, while providing the control, transparency and optimisation to grow the business using computer logic.

Maybe next year’s Dmexco will be manned by more A.I. bots and fewer humans? Cologne brewers, hoteliers and booth manufacturers will certainly hope not.

John Cole is the CCO of A.I. technology platform and Google Partner, Ezoic. Ezoic uses machine learning to help publishers manage their user experiences at scale adapting each page to each visitor.

Ezoic was Mediatel’s partner for its 2018 Dmexco roundtable debate.