Guy Thornton, the head of search marketing at Found, discusses the impact of machine learning on search marketing.
The way people interact with search engines is constantly changing. In the early days, we quickly adapted our behaviour to get the best results using short strings of keywords rather than full sentences. We understood the logic and reacted appropriately. Today, it’s all change again. We’ve moved away from this keyword-based way of searching and, instead, have become comfortable with more semantic search, accelerated by our growing adoption of the likes of Siri, Cortana and Google Now, all driving the involvement of Natural Language Processing (NLP) algorithms.
This latest behavioural shift has once again given much for the search engines to respond to and, understandably, Google is already leading the machine learning revolution. By analysing the context and intent of a user’s query, they are able to respond with results so relevant that users might not even need to click through to a top-ranked site to find their answer. And that’s partially due to machine learning (ML) algorithms like Google’s RankBrain, which update themselves without any supervision as they consume more and more data.
As the data collected becomes vast and diverse, the way it’s presented is equally moving away from simple metrics and stats. It’s involving ML to benchmark against other websites in the industry and to pre-empt visits that are most likely to result in a conversion (Google Analytics’ Smart Goals, for example). It’s getting even more crazily clever out there and it will be interesting to read what others in the industry have to say about the latest advancements in this week’s Search feature. But one thing’s for sure, online advertisers need to be ready for what’s undoubtedly coming next.
Each day also threatens to present new changes for search marketers to adapt to, with no advanced warnings like the ‘Mobilegeddon’ tip-off, leaving behind even the most vigilant SEO experts. Take the recent launch of the Google-led Accelerated Mobile Pages. This caught many publishers off-guard and has created uncertainty around how adverts are going to display on the mobile web. Google disrupts and leaves it to the rest of us to evolve and adapt.
Since companies like Google and Facebook became such data-mining powerhouses, advertisers and publishers also have more information than ever at their disposal. However, in order to adapt and maximise the opportunities that this information brings, they must build out their capability in advanced analytics through data management, predictive analytics and machine learning.
With 58% of British consumers saying they’re not concerned about the amount of data they share with brands, the opportunities for marketers to utilise ML to make their activity better targeted and more accurate for the benefit of the consumer are huge. The challenge for marketers to adapt, to take advantage in-line with the technology, is equally as great.
Successful examples of how marketers are adapting include the fields of lead scoring and cross-device pairing. Machine learning is already enabling marketers to discover patterns in big data sets that they can use as lead scoring criteria; allowing them to cleverly analyse their lead databases to find the indicators of buying intent and predict the likelihood of lead conversion.
With cross-device matching, giants like Facebook have led the way through single ID log-ins, allowing them to track user interactions across multiple devices. In our multi-channel world, cross-device matching can be used to map user interactions across multiple devices and is the holy grail of understanding the true user journey. However, advertisers in the technology space need a suitable alternative to enable them to bring the same cross-device sophistication to their campaigns.
With so much going on, big data becoming ubiquitous, and with artificial intelligence (AI) tools becoming more easily available, there is, however, a problem of producing meaningful, actionable insights. This requires certain analytical and cognitive skills and these skills, at the moment, are thankfully still beyond the capabilities of the ML AI.
The increasing demand for these in every industry - so these tools can be used efficiently to mine big data - has created the need for the modern data analyst. With just 38% of global digital companies feeling that they have the analysts they need to make sense of their data, our technical advancements have given us much greater scope and efficiencies and also left us with a talent challenge. In other words, rather than replacing humans, machine learning is actually creating jobs. So Siri, hold onto my P45, at least for now…