In the marketing world of late, there has been growing adoption of technology known as Artificial Intelligence (AI), or at least claims of it.

Defined as the “ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings”, the technology has been both touted as a new dawn in marketing efficiency and derided as a ‘flavour-of-the-month’ marketing ploy. In fact, few topics in the industry cause such a divided response at present.

AI in action

Away from the mainstream of AI-engineered chatbots and advances into voice search, 2017 saw the technology become the go-to product pitch in performance marketing. Countless companies laid claims to the tech, and with just a sampling of headlines from PerformanceIN including “iProspect Launches New AI-Powered Engine Core”, “Performance Horizon Makes Artificial Intelligence Debut” and “AppNexus Launches Industry’s First Programmable DSP with Machine Learning”, even we were reeled in by the claims.

When speaking to those on the ground, it becomes apparent that there is a wider opinion among experts that amid all the noise of claims to AI technology, it is important to make a distinction between those ubiquitous two letters and another familiar phrase; ‘machine learning’.

The phrase ‘AI’ itself is the “epitome of a buzzword”, says Ken Leren, founder of tracking and attribution company Tech Essence; “It’s a really simple thing to do, built from scratch or through the tools publicly available from IBM, Microsoft, AWS and Google. Pretty much anyone can claim it, it sounds 100 times more powerful than what it actually is and it’s very overused.”

The overblowing of the term AI is a sentiment echoed by Hanan Maayan, co-founder of data science company Trackonomics, who says “we have to be careful” about using the term to describe any work that is being done in the performance marketing industry. Maayan claims there has “scarcely been more hype with so little substance”, however, look a little deeper, and machine learning technology – the development of algorithms that process large data sets, evolve based on the data process and identify key trends and learnings from that data – has “real-world, real-life uses”.

Machine learning “predominantly” used

Machine learning is what companies laying claim to AI innovation are doing in the “majority of cases”, according to Leren, who describes the distinction in performance marketing terms as “like AI, but more like narrowing to voucher sites rather than affiliates”, adding that the technology can be “incredibly powerful, if trained properly, with enough data”.

At first glance, the real win behind machine learning for performance marketers appears to be the processing and analysing of data in larger quantities in real-time. Webgains is just one of several global affiliate networks working on a number of machine learning products; According to CEO Richard Dennys, examples of this use in practice include “identifying the most likely sales opportunities, to portfolio management and optimisations right through to rapid matching of almost 10 million product SKUs with available media units and campaigns across our affiliate landscape.

“From ‘zetabytes’ of data about advertising platforms, product preferences, time of day, the positioning of the product, AI can identify the marketing approach that is most likely to end up in a sale – a bit like buying every item in a supermarket and producing one delicious meal.”


But while headway is evidently being made “behind the scenes”, there are current limitations to the scale of adoption and the stature of those that can enter. The main factor is the ability of algorithms to run through “billions or trillions” of data points in little to no time at all, while the volumes collected can sometimes be isolated, which could prove problematic for smaller parties.

“The really big affiliates will have enough data to analyse and optimise their own offers, but because the traffic hitting individual advertisers is a fraction of that, even when all publishers are taken in to account, you don’t really get the amounts of data you need to optimise supply chains better than a human for example,” Leren explains.

Implications on the human element

In the long term – and perhaps giving the technology’s power more credit than it’s capable of at current scale – are concerns around the impact machine learning technology could have on the day-to-day roles of marketers. But while machine learning is there to make those decisions “easier, quicker and better informed”, says Maayan, “it is wrong to say the development of machine-learned technology in digital marketing will cost humans jobs or influence”.

On Webgains’ intentions, Dennys says the network is using the technology to enhance and support human decision making. Putting that into context, he explains the latest cognitive capabilities allow businesses to collect unstructured data into one body than almost instantaneously identify those patterns that humans might take weeks to see, or might not be able to find at all.

This symbiotic approach will see human roles evolve to become “campaign engineers”, says Dennys, drawing knowledge and experience to understanding the results of processed data acquired through the technology from multiple campaigns, and adapting them for continued performance.

Maayan shares the belief that widespread adoption of machine-learned technology, such as reporting and campaign management tasks, may change the “day-to-day reality” of every digital marketer.

“Instead of spreadsheets, graphs and assumptions, the digital marketer should be able to rely on machine learning to produce accurate, informed reports and forecasts without human intervention, freeing the human to be a decision-maker and strategist rather than a data-cruncher.

“This may, of course, have an impact on the skill sets that are valued in digital marketing, as well as the scope of entry-level roles. But these changes will take time.”


There is no ignoring that there is interest from companies within the performance marketing industry to using this subset of AI, and while the acronym itself may justifiably carry buzzword associations -machine learning itself has real value at the centre point of integration to product, services and strategies, and the potential to really impact the way companies work.

While the hype around AI looks set to continue for some time, behind the scenes machine-learning technology will evolve further; it’s here to stay, and as marketers, we need to begin harnessing it for better performance.

“Machine-learning should become a daily part of the job of every performance marketer,” says Maayan; “It has the potential to drive incredible efficiencies, change working practices and make the delivery of affiliate marketing more effective than it’s ever been before.”