Posted In: Digital by
Emma Tarleton,
August 7, 2026
By Emma Tarleton, Account Director
The impact of AI on everyday behaviour is becoming impossible to ignore, and nowhere is that more obvious than in the way people search.
Gone are the days when we all relied on two or three carefully chosen words to get the result we wanted. We are increasingly used to interacting with tools like ChatGPT and Gemini conversationally: giving them more context, adding nuance and expecting them to work out what we actually mean.
I know my own behaviour has changed. I regularly use the microphone to unload everything in my head into an AI tool and let it make sense of the chaos for me.
Search behaviour appears to be moving in the same direction. My Google Account Manager recently shared with me that long-form searches of five or more words had grown 12% year on year, significantly faster than shorter searches.
Whether AI is solely responsible for that shift is harder to prove. But one thing is clear: people are becoming more comfortable expressing what they want in much more detail.
And that matters for Google Ads.
For years, good paid search management was largely about predicting what somebody might search, building tightly controlled account structures around those queries and matching the right advert to them.
That model is becoming less useful.
Today, the challenge is less about predicting every possible search term and more about giving Google enough high-quality information to understand intent, while retaining control over where that automation can and cannot go.
I can almost hear my former self objecting as I write this.
I am naturally sceptical whenever Google launches a shiny new automation feature and tells advertisers that it is going to transform performance. That scepticism has served me pretty well over the years.
But AI Max has surprised me.
I have now tested it across accounts with different budgets, industries and levels of existing performance. I would not pretend every test produces identical results, but I have seen enough positive outcomes to believe it is something advertisers should be actively testing rather than dismissing.
The reason becomes clearer when you consider how search behaviour is changing.
Google has also told us that usage of AI-powered search experiences is growing rapidly. As people ask longer, more specific and more conversational questions, advertisers need a way of matching against that wider range of intent without attempting to predict every possible query manually.
That is where AI Max becomes useful.
Its value is not simply that it opens up additional reach. It gives Google more flexibility to interpret the intent behind a search and decide whether your business is relevant.
That can be powerful.
But only if you give it the right foundations.
Single Keyword Ad Groups used to be the bread and butter of how I managed Google Ads accounts.
Beautifully curated ad groups. One keyword in each — maybe two if I was feeling particularly adventurous and needed a plural variation.
I lived and died by Exact Match because it meant the advert I had lovingly written could be specifically tailored to the query somebody had searched.
And for a long time, it worked extremely well.
But the environment has changed.
Match types have become broader, Google has become better at interpreting intent and users themselves are searching in far more varied ways.
If somebody can express the same underlying need using hundreds of different combinations of words, building an individual ad group around every variation stops being practical.
More importantly, it can spread your data too thinly.
Instead, I increasingly favour grouping keywords around genuinely distinct themes and intentions.
That gives Google’s systems more data to learn from while still giving the advertiser control over the proposition, landing page and type of customer they are trying to reach.
The skill is no longer in building the biggest possible account structure.
It is in deciding which distinctions genuinely matter.
Responsive Search Ads already moved us away from writing one fixed advert for every search.
We provide Google with a range of headlines and descriptions, and the system decides which combination is most appropriate for an individual user.
Text customisation within AI Max takes that idea further.
If search queries are becoming longer and more varied, we simply cannot manually write bespoke advert copy for every possible expression of intent.
Google can.
Using information from your website, existing advertising assets and generative AI, text customisation can adapt messaging to make it more relevant to the individual search.
In theory, that gives us many of the benefits we originally wanted from tightly controlled structures like SKAGs: relevance between the query and the advert, without having to create thousands of individually managed ads.
But there is an important caveat.
The quality of the output is heavily dependent on the quality of the inputs.
If your website is vague, your messaging is inconsistent or you give Google complete freedom without any restrictions, you should not be surprised when the output is poor.
Which brings me to the most important part.
AI can make us faster and, used well, better at our jobs.
But poor inputs still create poor outputs.
The same principle applies whether you are prompting an LLM or giving more control to Google Ads.
I often describe AI as magic, and sometimes it genuinely feels like it. But it is not magic in the sense that we can stop thinking.
Before handing more of your account over to automation, you need to decide where Google can have freedom and where a human still needs to retain control.
My approach is relatively simple:
The aim should not be to prevent Google from making decisions.
It should be to create an environment in which it can make better ones.
One of the things I have always enjoyed about paid search is that it never stays still for very long.
There is always something new to test, something to complain about and something we previously considered best practice that suddenly needs reconsidering.
That can make us resistant to change.
If an account structure has worked for years, it is understandable to want to keep using it.
But “we have always done it this way” is not a particularly strong strategy when the behaviour of the people we are advertising to is changing underneath us.
My favourite answer to almost any paid search question remains the same:
Test it.
I am fortunate to work across a wide variety of accounts, which means I can test new features in different industries, at different spend levels and against different commercial objectives.
The answer is rarely that one tactic works everywhere.
But the broader direction is becoming clearer.
Good paid search management is becoming less about controlling every individual query and more about deciding where machines should have freedom, what information they need to make good decisions and where humans still need to intervene.
AI is not removing the need for paid search specialists.
It is changing what good paid search specialists need to be good at.
Privacy Policy
