Web users are impatient. We know what we want, and we want it now, which explains why so many of us would rather click back from a slow-loading page than wait to see what it has to offer. According to CDN provider Cloudflare, pages that load in 2.4 seconds have a 1.9% conversion rate, yet for those that take twice as long, that rate falls by more than half.
Page load times have a bearing on your search performance, as Google, Bing and their rivals strive to narrow the gap between query and answer – but it’s far from the only consideration. No search engine is keen on revealing precisely what goes into its algorithm, but their public-facing operations reveal a fair amount about what’s going on behind the scenes.
It’s good to talk
Google search delivers results for both Apple’s Siri and Google’s own voice assistants, whether hardware- or software-based. Alexa and Cortana rely on Bing and, according to Juniper Research, competition from Baidu and iFlyTek is likely to increase the number of voice-controlled assistants in use by 90%, come 2024. As consumers rely less on the keyboard and more on voice, optimizing content for spoken delivery becomes more pressing.
“That’s something you can help us with by using structured data [to] tell us a bit more about what this page is about,” said Google web trends analyst John Mueller, before warning against content optimized specifically to answer common questions. “Some other kinds of voice assistants… try to match the question more directly, so they’re looking for web pages that say ‘what is the tallest mountain’ as the title and they read the first paragraph. I think for Google, that’s probably overdoing it and quickly ends up in a situation where you create a doorway site with all of these question variations and a short answer, and these pages themselves have really low value.”
Forget the algorithm
Around 15% of the queries Google receives each day have never been entered before. What does this tell you about your audience? That it’s more diverse than you might have imagined. It also explains why, in the words of Google VP of Search, Pandu Nayak, “with the latest advancements from our research team in the science of language understanding – made possible by machine learning – we’re making a significant improvement to how we understand queries, representing the biggest leap forward in the past five years, and one of the biggest leaps forward in the history of Search.” That improvement is BERT, which we’ll come to in a moment, but it’s equally important to talk about RankBrain which, according to Google, is 10% better than its own engineers at identifying the most relevant content.
As Logic Based Marketing explains, RankBrain is an iterative process, which tests minor algorithm changes against set keywords and, if the results are better than it was already getting, writes those changes into the algorithm itself. It’s good news for marketers because, in the past, they’ve “had the awful experience of making superior quality content and getting lower rankings because their technical SEO wasn’t perfect”. Now, technical SEO is less important, as Google is no longer relying on such rigid rules: the algorithm is changing in real time and, the higher the quality of your content, the more it appears likely it will favor your site over rivals.
Meet BERT
Short for Bidirectional Encoder Representations from Transformers, BERT is a neural network-based natural language processing (NLP) system that better understands natural language search queries and has the potential to turn search on its head. Publishers who might have crafted their content with a view to satisfying web crawlers, rather than human readers, in the hope of attracting clicks and serving ads, will have to reorient towards delivering truly meaningful content with value.
For scrupulous publishers, who are already using content intelligence tools to define clear audience personas and forecast the kinds of questions they’re likely to ask, BERT will likely be a boon. Their content, both existing and in development, will be naturally optimized, thanks to the AI-based analysis that went into its creation in an effort to maximize its relevance to the potential audience.
Thanks to BERT, Google will be able to understand more complex search queries and, when they notice this, your audience will be less inclined to try and second-guess the algorithm by typing strings of keywords. This will make it easier for them to surface long-tail content that answers specific questions, rather than linking to more generic content and piecing together their own answer by visiting several sites in sequence. The brands that will benefit here are those that use content intelligence tools to forecast the questions their audience is going to ask, and to formulate answers in advance.
More intelligent search results
Machine learning is playing an ever-larger role in ranking and delivering search results. It’s also speeding up the evolution of the algorithms that sit behind each service which suggests that, in the not too distant future, even the search giants themselves may be largely unaware how each variable is weighted when calculating which links to serve.
Thus, while search is evolving, the advice for brands that want to maximize engagement remains the same: generate worthwhile content, put your audience first, and don’t try to game the system. While gaming may have had some success in the past, more intelligent analysis of what you produce is inevitable going forward, and such trickery will be found out.
Most importantly, though, keep in mind how you can help the search engines achieve their goal of narrowing the gap between a query and the most relevant result. Help them succeed in this respect, and you’ll benefit in return.