The future of website traffic: AI overviews and beyond

Bringing people to your website – and tracking how they got there – used to be much easier. Clever SEO practices could game Google’s algorithm and get you to the top, and there was far less content to compete with across marketing channels.

You could see which keywords users searched to find you, and which ones led to a conversion or sale. Keywords were also much easier to decipher and target – search terms used to be simple and direct, while people were first figuring out how to use search engines. Now they’re highly varied and complex.

Add AI into the mix, and things get even more complicated. It’s becoming increasingly difficult to predict how people will discover you, and to track their behaviour once they do.

In this post, we’ll look at:

• Key challenges around traffic and tracking
• The emergence and evolution of AI tools in search
• How AI search and tools are impacting the marketing funnel
• What Google’s AI mode could mean for marketers
• Tips on what to do next

Traffic & tracking challenges

The modern marketer faces a mounting list of challenges they’ve not had to deal with before. Changing algorithms on various platforms are reducing organic reach. Factor in the sheer volume of content you’re competing with to get users’ attention – it’s estimated that we create 463 exabytes of data daily – then actually getting people to engage with your content and visit your website is becoming even more challenging.

Once users are on your site, you have decreasing visibility of how they discovered you, how they got there, and sometimes whether they’re there at all. Deprecation of third-party cookies will eventually make tracking users across multiple websites and attributing conversions and engagement impossible. The future of website tracking in general is moving more towards an emphasis on user privacy and control. Google has been limiting and removing access to keyword data on their platforms for years – outside of paid traffic, you’re having to do a lot of guesswork about the queries people use to find your content.

As if that weren’t enough to deal with – enter AI.

The emergence of AI in the marketing funnel

AI can be a powerful tool to help marketers. Whether it’s used for productivity, predictive analytics and forecasting, content generation and personalisation, or to improve the customer experience, it can be incredibly helpful in marketing strategy and delivery.

But there’s another facet of AI that marketers are having to get their heads around. AI-powered search puts the power in the hands of the users and leaves the marketers having to figure out where they fit in. Increasingly, users can get direct answers to their complex queries at-a-glance, either within their chosen AI platform or in the AI overview of a search results page (SERP). Websites are shifting towards becoming data sources, rather than destinations.

How AI is breaking the funnel

Although your content or brand may appear – or be synthesised – in AI overviews and responses, the user may have got their answer, so there’s no guarantee they’ll click through (known as “zero click behaviour”). It’s estimated that the organic click-through rate is reduced by 30-70%, depending on the query. Outside of this, there is currently no sustainable way to track brand or content mentions in AI overviews at scale, and the way that AI click-throughs are attributed on analytics platforms like GA4 is currently not helpful without some serious customisation of the data and reports.

Then there’s the issue of synthesisation – the large language models (LLMs) we’re referring to here aren’t regurgitating your content verbatim, they’re reviewing and evaluating content from multiple sources and forming them into what they deem will be the most useful answer for the user.

So to put it simply, you’re getting less visibility and less control over how your brand and content are represented online, and who’s engaging with it and when.

Google’s AI mode

Just when marketers thought they had enough to contend with when it came to AI search, at their I/O conference in March, Google announced AI Mode – an expansion to AI overviews driven by their Gemini 2.0 engine, which will add more advanced reasoning, thinking and capabilities. There’s a very detailed and fairly technical unpicking of their patent and how it works here – but the key point is that it will use more contextual information, including past user behaviour, to provide detailed and highly personalised responses. It will go beyond retrieving and repeating information to evaluating complex relationships between concepts, considering context, weighing up alternatives and generating a response that’s been deliberated with the thought and consideration a human would apply, but in less than a few seconds. That means it’ll be able to answer more complex queries, but AI responses will become increasingly varied from user to user, even more so than they are now. All of which adds up to even less visibility and control.

Where do we go next?

It all may sound a bit doom and gloom in terms of the outlook for marketing, and particularly marketing on organic search. In the immediate term, it’s about controlling what you can control.

Continue focusing on:

• Creating user-centric experiences and well-structured websites that are easy to navigate and extract information from (for both users and search engines/LLMs)
• Producing high-quality content that’s useful to the user
• Tracking the user behaviour that you can and using this to influence your development roadmaps and content plans to better cater to user needs.

And a few things you can add to your strategy:

• Pay closer attention to tracking branded traffic, direct traffic and brand mentions
• Gain an understanding of how AI search operates and some techniques you can use to try and make your content appeal to LLMs