Why artificial intelligence is key in Google’s battle with Amazon

John Giannandrea’s appointment as Google’s search chief is a tactical move in the new search wars.

Outgoing SVP for search Amit Singhal did a phenomenal job of seeing off the challenge from Microsoft, Yahoo and countless others who have been consigned to the history books, but the world’s biggest search engine knows the next war is not one of preference (see Scroogled – it’s about requirement.)

Former Google CEO Eric Schmidt famously said: “Many people think our main competition is Bing or Yahoo. But, really, our biggest search competitor is Amazon.”

If Amazon’s search results represent a bigger and better inventory of products – and a large number of users think they do – Google knows it will have to ensure it is providing the best possible information in every other area in order to remain competitive.

Recent infrastructure updates like Hummingbird mark a shift in the way Google processes its search results – not just the order in which they are returned – and the search engine is putting its faith in Giannandrea to take this even further.

Google has historically relied on the appearance of searchers’ keyword queries within the copy and meta data of a webpage to determine relevance – in other words, whether the webpage answers the user’s question. This parroting of queries led to websites like Yahoo Answers dominating uncompetitive search queries regardless of the quality of the information – simply because the page contained more or less the exact question plus some form of answer.

Google’s rich answers have made Yahoo Answers and similar sites less prevalent. If a searcher wants a simple, single sentence answer to their query it will be right there above the search results, negating the need to click through. If the searcher wanted something else they’re more than welcome to click on another result, starving Q&A sites of traffic and shuffling them down the search results.

What Giannandrea has delivered in RankBrain – and Singhal with Hummingbird in 2013 – is the ability to discern what a searcher is likely to prefer in the results, based not on the copy on the page, but on intent.

Hummingbird used associations between concepts to determine what the likely answer to a question will be instead of parroting back the question; RankBrain applied this principle to search queries that have never been asked before – amounting to more than 15 per cent of the queries entered each day.

The appointment of Giannandrea shows that Google is serious about catering to these searchers – a key group of users if the search engine is to see off the challenge from Amazon.

According to 2012 Forrester research 39 per cent of purchases began with a search on Amazon versus a paltry 11 per cent on the world’s biggest search engine. Google knows that its users look to its organic search results for information that Amazon can’t provide – but in a fight to the death between e-commerce engines, Amazon beats AdWords hands down.

2011’s ‘Winning the Zero Moment of Truth’ Google whitepaper demonstrated the new mental model for purchases, where consumers used an average of 10 touchpoints before buying. Tellingly this had doubled since the previous year, so it’s safe to assume that Google knows exactly how much research is involved before a purchase in 2016.

The study stressed the need for brands to be involved at every stage of the path to purchase if they’re going to be seriously considered and the search engine knows it’s not exempt from this: relying on AdWords to enable its advertisers to sell their products is not enough to tempt searchers away from Amazon and help them to make a decision.

Google’s investment in AI and infrastructure for its algorithms shows that it’s serious about playing by its own rules.

Stephen Kenwright is director of search for Branded3

Stephen Kenwright

All by Stephen