Search is evolving. A SERP (search engine results page) used to simply be a list of 10 search results, then came featured snippets, a small overview of content that provided helpful information to answer a search query, and now? The AI Overview.
Google’s AI Overviews are dominating the SERP. When an AI overview appears with a featured snippet, they can take up 76% of the screen space on a device. So, what do we do about them?
What is an AI Overview?
An AI Overview is a feature designed by Google to bring users the answers to their queries without the need for in depth searching. Pulling information from multiple websites to draw a concise summary to provide immediate answers, they take up the top portion of the SERP.
However, Google has now confirmed that the AI Overview is more than just a SERP feature, it is the new position 1.
The SEO landscape is in fluctuation, many SEO’s think SEO’s dead, many think it’s an opportunity. Can we optimise for them? Should we be optimising for them? Is there any point in anything anymore or is it all just a pointless endeavour into the soul sucking ether (feel free to insert more of your own dramatics here)?
The Problems with AI Overviews
As much as an AI Overview is designed to make the Google search results more friendly to a user, they aren’t flawless. Generative AI poses several issues, not just for users, but for SEO’s and businesses alike.
Zero Click Search
The main problem with AI Overviews is that they create a SERP that doesn’t require anyone to click onto a site. A user can get all the information they need at a glance without actually needing to read your full blog post or information guide.
The most frustrating part of it all? The overview may well be generated using your content.
So, if you’re a site that sells kettles and your site was thriving and gaining sales via informational posts that you’ve written on how to boil the best water for pasta, but Google has summed up your post without people needing to click onto your post – how are you going to secure the same amount of traffic and sales as before?
The same also goes for news publishers. Carly Steven, SEO and Editorial E-Commerce Director for the Daily Mail recently stated ‘On desktop, when we are ranking number one in organic search, it [clickthrough] effectively is about 13% on desktop and about 20% on mobile. When we are still ranking number one organically but there is an AI Overview present, that drops to less than 5% on desktop and 7% on mobile, so a pretty profound change in clickthrough.’
A quick look through GA4/Search Console when your site is featured in an overview can often show the impact – impressions are high, clicks are low.
Incorrect Advice
The point of controversy around AI generated answers is the accuracy of the information provided in the overview.
Studies have shown that 74% of Gen Z and 66% of Millennials find AI overviews useful, and only 33% of all search users say that they don’t completely trust AI overviews. So for the average search user, most people trust the AI generated answers to their queries.
But should we?
The fact of the matter is that crawlers can misunderstand key information that features in AI generated summaries. There have been several instances where Reddit threads have been used to form the basis of the overview and sarcasm has been misconstrued as truth – here are a couple of notable answers:
- One rock a day – The AI overview was caught saying that geologists recommend people should eat one rock a day as the supposed mineral content is good for us.
- Cheesy pizza – The Overview stated that to stop cheese from moving around on a pizza, the cheese should be glued to the base.
Although Google have said these were isolated incidents, there have also been instances of misinformation for other queries where the same question has been asked in a different way across multiple searches, and the Overview has given a different answer to each search.
Ai responses can provide valuable insights to a user, but they can also get things wrong (often called an AI hallucination). You can often find these hallucinations yourself if you go on a deep dive search on a topic that you really understand – and although you will be able to spot them, not all users will. If users are trusting the information that your content has been used to create, it is not Google misinterpreting the content that is going to get the blame – it’s going to be you, as the referenced source.
Changing Search Landscape
One of the main problems with AI overviews, is that they are consistently changing – along with the entire search landscape.
A search generative experience appears to be the only way forward, but as technology evolves, ranking factors evolve and the way that users search evolves, so it can feel like hiking a mountain undergoing a mudslide.
Gone are the days where a 12 month SEO strategy is going to work for ranking in search. AI demands fresh, up to date content and the best methods of ranking seem to change every few months and keeping on top of it all requires ongoing consistency.
One thing is for sure though, we can’t hide from what AI Overviews are doing to the search bar – we need to adapt strategies in order to appear in AI Overviews. People might not always click on the site, but there is a much higher chance of click through when you appear in an Overview vs when you don’t (in fact, if you’re not in the top 3 positions and an AI Overview is present, you can lose up to 27.04% of traffic on average).
What to do about it?
While there is no surefire way to ensure that every user is going to click on your site (but wouldn’t that be lovely), there are ways to improve your visibility in search. These actionable tips need to be backed up by quality content that is genuinely useful to a user – but they can improve your chances of appearing in an Overview, and in turn make you more visible to a user.
Long Tail Keywords
Data collected by Neil Patel shows that 77.2% of queries that feature 6 or more words will generate an AI Overview.
One of the main changes in search behaviour that has been noticed across the search industry is how people are structuring their searches. Questions are less general and more specific to a unique need. LLMs have changed how we ask questions – people now want a specific and fast answer, they don’t want to have to do a secondary search – they want everything in one go.
How does this affect SEO? Long tail key words and queries are more likely to generate AI overviews.
Crafting your content to a long tail main keyword usually means more tailored content. Instead of asking ‘what mop do i buy?’, instead think of ‘what mop to I buy for cleaning grease residue’.
The more specific your content, the more likely to be featured in an Overview it is.
Plus, long tail key words tend to have lower competition scores and can often be easier to target.
Write for the Query
Intertwining with long tail key words, it is also essential to write your content to answer a query. For this you’ll need to have an understanding of your audience.
Examine your H1 – do you have a ‘Top 5 impulsive purchasing trends of 2023’ blog? Is that what your audience are going to be searching for? Would you type that into a search bar? If not, reexamine your content.
You could be losing valuable traffic just because your content isn’t answering a user query.
There are tools out there such as Also Asked which can tell you the questions your audience is asking. Another useful tool is Knowatoa, offering suggestions for questions aimed not only at users, but also at LLMs (which could be incredibly useful for the next stage of search.)
If you haven’t already, it’s time to think about the existing content on your site. If it’s old and can be updated, update it. If it’s old and can be restructured to more specifically target the needs of your user, reoptimise it. If it’s old and doesn’t serve a valuable purpose, kill your darlings.
Help Google Understand
The structure of your content is incredibly important, not only for helping readers understand, but helping the Google crawlers to understand.
Studies have found that Google’s AI Overviews love a list. So much so that 69% of all AI overviews feature an unordered list. Is publishing listicles the only way forward? No, probably not. But your heading structure should be clear, so that when the crawlers are establishing their lists they can do so easily.
Your headings need to be clearly worded. No poetic spiel. Try to summarise whatever is in the section of content in a few short words. Flag the information points for what they are, again, this will make it just as easy for a user to navigate as it will a crawler.
It is also incredibly helpful to summarise your content. If you’re writing a blog, your user shouldn’t have to wait until the end to find the point – give them the answers to their questions early on – be that in the introduction or in a Key Points summary. If your content is easy for a user to navigate, it will also be easy for bots.
Provide Quick Answers
Writing content shouldn’t feel like writing a dissertation, there’s no word count to hit. When you’re writing keep a focus on what exactly it is that you’re writing, highlight it, check back in after a few sections, make sure that it is all relevant.
It’s easy for content to become distorted and distracted, especially if you’re trying to take it in a direction specifically so you can target a keyword or two – but that no longer matters quite as much. Your content needs to be detailed and show expertise, but also be to the point and relevant.
It comes back to the basic SEO principles, EEAT and brand authority are both signals that Google is going to be looking for, but the focus is even more so on helpful content (and Google keeps doing those sneaky updates that prioritise helpful content, so you should be focusing on being helpful anyway).
The Importance of Brand
Recent updates have been proving that brand has never been more important. Branded keywords are less likely to trigger AI Overviews (only 4.79% do) – but when they do, they get a +18.68% CTR boost.
So what does this mean?
Taking steps to boost brand awareness have never been more important. Are you interacting with posts on Reddit that mention your brand? Are reviews being responded to? Are you creating discussion that sets you up as a reliable source of information?
All of these things give Google signals that you know what you’re on about – but they also mean that your audience is going to come to you. That’s where you’re going to guarantee a click. Your brand should be one of your main focuses. Digital PR services can have a direct positive impact here, but in a world where clicks are becoming harder to earn, investing time into expanding awareness of your brand is essential.
What’s Next?
Search is evolving and it’s doing it far faster than any of us anticipated. By the time we’ve fully figured out AI Overviews, the next thing will be upon us – so what is the next thing?
AI Mode
Currently being rolled out in the US, AI Mode was announced just a couple of weeks ago as Google’s next big thing in search.
As for what that means for SEO, it’s not entirely certain just yet – but it’s something to be aware of. AI Mode is designed to function just like LLMs, similar to ChatGPT, where longer queries can be answered in depth. The good thing is, Google have mentioned that the results will be fully cited, but it is yet to be seen how this will impact search and how users will respond to it.
Many people thought that LLMs would be the death of the Google SERP, but Google has only lost 1% of its traffic in the past year, so perhaps it will be the same here? Let’s not panic just yet.
The Expanding Search Landscape
The thing is, search is unstable (just like we’ll all be after trying to keep up with all of this) and nobody knows what’s coming next.
However, it appears that AI Overviews are here to stay. To get users to click on our sites, we need our users to have visibility of those sites, and that means at least trying to wiggle into the overviews.
The future might hold AI Mode, shifts in how LLMs are used for search (and tears, many many tears), but all we can do is continue to adapt strategies to meet to changing user behaviours.
