With search engine algorithms increasingly incorporating machine learning to determine results for search engine queries, how can SEOs develop strategies that aim to please both humans and AIs?
This subject was the primary focus of the first round of SEO talks at SMX London. Nagu Rangan, Senior Program Manager at Bing, and Stephanie Wallace, Vice President of Owned Media at Nebo, delivered talks on how digital marketers should approach the changing search engine landscape.
As Nagu Rangan explained, SEOs are often torn between adhering to algorithm changes or specific guidelines set out by search engines. The secret to ranking organically in 2019 is to bear both in mind, which ultimately goes back to the goal of search engines in the first place.
The goals of a search engine are to:
- Discover, crawl and index everything
- Create ‘the’ knowledge graph
- Provide super fast access to information
Well, at least in part.
Nagu reiterated that, ultimately, the main goal of a search engine is user satisfaction. Users want timely, relevant results that they can trust, which will lead to them continuing to use the search engine in the future. There is no mathematical formula for user satisfaction, which means that building a search engine is both an art and a science.
Nagu outlined the following process of building and improving a search engine:
Search engine teams like Bing and Google will set out their expected guidelines for making search engine users happy, before experimenting with the results in a training set. The training set is reviewed by a panel of judges, after which changes are made to the ranking algorithm and tested on real users. The process is cyclical, and search engines will make changes to their algorithms almost every day.
According to Nagu, the 3 key factors for user satisfaction are:
- Relevance of the result – does the result provide the information the user was searching for?
- Quality – can the user trust this information?
- Context – is it timely and relevant to the location
At a more granular level, Nagu explained the key components of the quality factor:
Trust – can I trust this author, content and site? Would I give them my credit card number?
Utility – is this content useful, with sufficient details? Does it provide enough unique value to users?
Presentation – is the content well presented and easy to find? Can I distinguish ads from the main content?
The machine learning used in search engine algorithms is ultimately trying to learn about what real human users want and serve them the most useful and relevant results. It is therefore crucial that SEOs focus on understanding their users and anticipate what they want.
Stephanie Wallace carried on from Nagu’s talk, discussing how Google’s ranking factors have changed hugely in recent years. While ranking factors have traditionally focused on backlinks and keyword-density, Google is increasingly prioritising user signals, content quality and social signals. This makes sense, as the ultimate goal for search engines is to keep users happy and any engagement metrics that reflect this should be rewarded, which is why machine learning systems like RankBrain are increasingly important.
So, what does the impact of machine learning look like in search?
For Stephanie, this means optimising for consumer discovery rather than traditional search. In practical terms, this can be broken down into three main sections.
1.Optimising for discovery
SEOs must widen their SEO strategies from traditional ranking factors such as backlinks and keywords. This primarily means optimising for RankBrain and creating a better user experience.
- Using natural language – make your content conversational and easy to understand
- Prioritising reputation and trustworthiness – users want assurance that the content they are reading is reliable, which means that search engines are increasingly rewarding websites whose articles always cite their authors and include detailed biographies. If possible, editorial content should be shared on social media as an added indication that the content is trustworthy. This ties into E-A-T – Expertise, Authority, and Trust — three factors that Google uses to measure how much trust it should place in a brand or website.
- Ensuring that users can easily navigate through the website, especially on mobile devices
2. SERP optimisation
Stephanie argues that the modern homepage is on the Search Engine Results Page (SERP) itself, as local business and brand panels provide detailed company information without the user even clicking through to the website. With this in mind, SEOs should be leveraging schema markup and Google My Business to maximise their real estate in the SERPs and better engage users.
Stephanie also advocates for SEOs to make use of the Google Brand Panel to increase brand awareness. This SERP functionality allows websites to promote their blog posts, products, recent social media announcements and offers.
3.Optimising for voice search
Stephanie cited a recent study conducted by PWC which revealed that 65% of 25-49 year olds speak to their voice-enabled devices at least once a day, while another study predicts that half of online searches will be conducted by voice search in 2020. Given these findings, it’s clear that SEOs need to understand how to optimise their sites for these new technologies.
SEOs should analyse the type of voice search queries that they could optimise their site for and ensure that they add the appropriate mix of content to target these opportunities. For example, informational voice queries will often be pulled from the featured snippet, while local search queries will be pulled from the local pack.
Stephanie ended her talk by mentioning a new type of schema markup specifically targeted towards voice search – Speakable. The schema markup is still in the beta stage, and is currently only used in news results. However, the markup will inevitably lead into wider voice search opportunities and SEOs should look to incorporate it into their strategies in the coming years.