As a result, a larger number of sites can get clicks from AIOs. However, more diversity is offset by fewer URLs cited in AIOs and fewer outgoing clickers due to more in-depth answers. A tad over 12 million URLs appear in search results compared to 2.7 million in AIOs (23.1%).
How Many AIOs Contain The Search Query?
It’s unclear whether AIO answers contain the search query. Since queries really represent user intent, which is implied rather than explicit, it’s possible that they don’t.
As a result, tailoring content too much to the explicit query and missing intent could lead Google to not pick it as a citation or source for AI answers.
The data shows that only 6% of AIOs contain the search query.
That number is slightly higher in SGE, at 7%, and lower in live AIOs, at 5.1%.
As a result, meeting user intent in the content is much more important than we might have assumed.
This should not come as a surprise since user intent has been a key ranking requirement in SEO for many years, but seeing the data is shocking.
How Different Are AIOs In Vs. Outside Of SGE?
SGE is Google’s beta testing environment for new Search features. It is not, as commonly mistaken, equal to AI Overviews.
Since Google has experimented with new AI features in SGE, the question arises of how different AIOs are in vs. out of SGE. Can we learn anything from AIOs in SGE about what’s to come?
I looked at over 8,000 AIOs in and outside of SGE and found that 30% of AIOs have very different content in SGE compared to live. SGE results are likely not an indicator of what’s to come, at least at this point.
The length of SGE vs. live AIOs varies but is the same on average: 1,019 in SGE vs. 996 live.
For example, the AIO for the search query “Marketing manager” has 347 characters in SGE vs. 1,473 live.
But most AIO answers look like “P&L,” which has 1,188 in SGE and 1,124 in the live results.
We cannot conclude that SGE results (and the potential future of AIOs) are longer (more succinct) or shorter (more detailed). I will analyze the results further.
On the domain level, the following 10 domains would see the biggest relative visibility increases if SGE was a predictor of future performance:
byjus.com.
geeksforgeeks.org.
timesofindia.indiatimes.com.
amazon.com.
ahrefs.com.
github.com.
medium.com.
pcmag.com.
techtarget.com.
coursera.org.
The top 10 domains that would be set to lose the most relative AIO visibility are:
support.squarespace.com.
knowledge.hubspot.com.
quickbooks.intuit.com.
allrecipes.com.
bhg.com.
bankrate.com.
cnbc.com.
nerdwallet.com.
thespruce.com.
tiktok.com.
Meaning
All of this means three things:
1. Optimizing for AI Overviews is similar to Featured Snippets with the difference of being more user-intent focused.
Featured Snippet-optimization is very exact match-driven – you need to match the question and clearly indicate that the answer relates to the question. Not for AIOs.
For AIOs, we can tweak our content to match the AIO answer or give a better one, but reflecting “useful” information in the search query context is much more important than the exact wording.
Three challenges stand in the way:
Understand and target what sections appear in AIOs, like lists, comparisons, “what is…” or “how to…” explanations, etc.
Keep track of AIOs since they tend to change fairly often, which means we must adjust our content and impact expectations accordingly. Just recently, Google started testing a sidebar with links instead of a carousel.
Rank in the top 10 positions, ideally top 3, for a query is not a pre-requisite but increases your chances.
2. SGE is useful for monitoring potential AIO design changes but not to predict how AIO answers might change. One threat to keep an eye on is citation-less AIOs.
3. Social could make a comeback! Many years ago, social signals were hyped as SEO ranking factors. Today, the strong prominence of social networks like YouTube and LinkedIn in citations offers an opportunity to impact AIOs with social and video content.
Thinking Ahead
AIOs do the opposite of leveling the playing field. They create an imbalance where a few sites that get cited get more visibility than everyone else.
However, they also shrink the playing field by answering user questions better and more often than Featured Snippets.
The risk of getting fewer clicks grows with better AIO answers – but there is also the risk of fewer ad clicks. Organic and paid results always existed in balance. The quality of one impacts the other. Unless Google embeds new ad modules – which is likely – better organic answers will come at the cost of ad revenue.
At the same time, Google is pulled forward from competitors like OpenAI and Perplexity, which constantly ship better models and increase the chance of searchers not using Google for answers. It will be hard for Google not to iterate and innovate on AI in the search results.
Differences in AIO design might arise between the EU and non-EU countries. New regulations and fines will lower the appetite for tech companies like Alphabet, Meta, or Apple to launch AI features in the EU.
The result could be two internets that allow us to compare the impact and changing AI landscape in countries like the US.
Stay tuned for round two.
Boost your skills with Growth Memo’s weekly expert insights. Subscribe for Free!
Google AI Overview Study: Link Selection Based on Related Queries
New ways to connect to the web with AI Overviews
Featured Image: Paulo Bobita/Search Engine Journal