What six quarters of the AI evolution have taught us

For the past six quarters, we have worked with forward-thinking brands including the AA, Habito by Monzo, Domestic & General and Biffa to understand how AI is changing marketing.

That experience has given us a much clearer picture of what these changes mean for brands and the agencies working with them.

As the founder of an agency, what has struck me most is the speed at which this has moved beyond technology.

Eighteen months ago, most conversations about AI were about content creation and productivity. Today, AI is changing how people discover brands, research their options and make buying decisions.

This has forced us to rethink parts of our agency, build new capabilities and work with clients in a much more connected way.

Here are the most important things we have learned.

1. The real change is user behaviour

The industry can become overly focused on platforms, products and terminology.

AI Search. GEO. ChatGPT Ads. Conversational agents. Agentic commerce.

But the most important change is much simpler: people are behaving differently.

They are asking longer and more detailed questions. They expect platforms to understand their circumstances. They are using AI to research categories, compare options and create shortlists.

Instead of searching through a series of links, they are increasingly asking for an answer or recommendation.

This means brands are no longer competing only for a click. They are competing to become part of the answer and, ultimately, part of the decision.

In effect, brands now need to sell to the machines as well as to people.

That does not mean creating robotic content or attempting to manipulate an algorithm. It means giving AI systems the evidence they need to understand the brand, trust its claims and confidently recommend it.

Products, expertise, evidence and points of difference need to be clear and consistent across the brand's own digital estate and the independent sources LLMs use to form their answers.

Understanding that behavioural change has to come before deciding which technology, channel or tactic to use.

2 AI Search is bigger than SEO

When we first began working seriously in AI Search, it became clear that this was not simply another version of SEO.

Technical foundations and on-site content matter, but LLMs also rely heavily on what trusted third-party sources say about a brand.

That brings together areas that have traditionally been managed separately: SEO, digital PR, affiliate partnerships, editorial content, paid media, brand reputation and audience research.

Our work became focused on the prompts customers use, how brands appear within the answers, which competitors are recommended, what sources influence those recommendations and where the gaps exist.

We then used targeted off-site content and publisher activity to influence those sources, an approach that now forms part of our broader work in AI search visibility.

The results demonstrated that this was a measurable commercial opportunity.

Habito by Monzo increased leads attributed to ChatGPT by 17x. The AA moved from 22% to 100% AI visibility in ten weeks and subsequently expanded the programme from one category to five.

Our work with the AA also went on to win Gold at the 2026 Performance Marketing Awards.

The lesson was not that every business needed a new GEO team.

It was that businesses needed people capable of joining together disciplines that had previously worked apart.

3 Paid media and AI Search are converging

The next stage is already taking shape.

ChatGPT has introduced advertising. OpenAI is testing Sponsored Agents that allow users to move from an advertisement into a branded AI conversation.

Google is developing conversational discovery ads and business agents. Amazon already allows customers to question, compare and purchase products through Alexa.

For most of the last 20 years, performance marketing has been built around a familiar journey:

Search. Ad. Landing page. Conversion.

That journey will continue, but it will no longer be the only one that matters.

A customer may research a category through an LLM, encounter a sponsored recommendation, question a branded agent, compare products and qualify their decision before visiting a website.

The landing page is not disappearing, but it is losing its monopoly on persuasion and conversion.

This creates new questions.

What information powers the agent? Who controls the answers? How should the conversation reflect the customer's needs? How does it connect with product data, content, CRM and customer service? What happens when the customer moves between an independent AI answer and a paid brand conversation?

These questions go well beyond paid media. Agencies will need to understand how the different parts of the customer journey fit together.

4 The right measurement starts with the right question

AI is also exposing the weaknesses in conventional attribution.

Clicks and last-click conversions cannot fully explain a journey in which research, comparison and qualification increasingly happen inside AI platforms.

Marketing mix modelling remains valuable, but models built around historic channels and assumptions need to evolve as customer behaviour changes. Otherwise, brands risk making investment decisions based on an incomplete picture of how people discover and engage with them.

The starting point should be:

What is the commercial question we need the data to answer?

Campaigns must then be designed to create the evidence required to answer that question.

Audiences, channels, content, geography, timing, budgets and control groups need to be structured in a way that allows us to isolate effects and understand what is genuinely influencing performance.

AI-driven correlation and media-effectiveness analysis will be key. It allows us to examine multiple datasets, identify relationships conventional reporting may miss and test whether changes in AI visibility, media exposure and user behaviour are translating into commercial outcomes.

That means connecting:

• AI visibility, recommendations, sentiment and citations

• Paid-media exposure and conversational engagement

• Website behaviour, leads and sales

• CRM outcomes and customer value

• Incremental changes across the wider business

Measurement needs to be built into campaigns from the beginning, with activity designed around the commercial questions we're trying to answer.

5. Why no single team can “own” AI marketing

The brands that have moved fastest have not treated AI as the responsibility of one person or department.

They have built virtual sprint teams around a specific problem or opportunity.

Marketing, paid media, SEO, PR, content, ecommerce, data, technology and customer-experience specialists come together for a defined purpose. They establish what they need to learn, design an appropriate test and then use the evidence to decide what happens next.

The teams at Domestic & General, Habito by Monzo, the AA and Biffa have all demonstrated elements of this approach.

They moved before every process, reporting line and job description had been perfected. They were curious, flexible and prepared to work across traditional departmental boundaries.

In many cases, our role has extended beyond delivering the work. We have helped internal marketers explain the opportunity, bring different stakeholders together and create the organisational momentum needed to act.

That has been one of the biggest lessons of the past six quarters.

Adopting AI is primarily an organisational challenge rather than a technological challenge.

We've explored this challenge in more detail in our article on why no single team can own AI visibility.

6 Agencies need to change shape too

The same lesson applies to agencies.

AI will automate more campaign production, creative variation, monitoring, analysis and reporting.

The bigger change is that the problems clients need us to solve no longer fit comfortably within traditional agency departments.

An AI-discovery problem may require audience research, technical SEO, publisher relationships, content, paid media and analytics. A conversational-advertising project may involve media, product data, CRM integration, customer experience and compliance.

Technology is helping connect these disciplines too. Model Context Protocol (MCP) provides a common standard for connecting AI systems with external tools and data. For agencies, this creates opportunities to bring together information from media platforms, analytics and client systems, helping teams work with a more complete picture of performance.

The agency of the future will need strong specialists, but it will assemble them around the client's problem rather than around a predefined channel.

Its value will come from gaining a deep understanding of the customer, the market and the commercial challenge, then connecting the right people, data, media and technology to solve it.

That is the direction we've taken at Planit, bringing our different digital marketing capabilities together around each client's commercial challenges.

Signal gives us the understanding: audience insight, market intelligence and AI visibility.

Omnivisible turns that understanding into connected activity across paid media and targeted editorial environments.

Pulse uses AI-driven correlation and media-effectiveness analysis to determine what is genuinely creating commercial impact.

Together, they reflect the process we've learned to follow:

Understand deeply.

Ask the right question.

Build activity capable of answering it.

Measure what really changed.

What we have ultimately learned

No agency can credibly claim to have every answer about AI. The technology and user behaviour are developing too quickly. But after six quarters of practical work with forward-thinking brands, I am certain about one thing.

The businesses that succeed will be those that look beyond the constant stream of AI announcements and new tools. They'll take the time to understand what's actually changing, how it affects their customers and where AI can make a meaningful difference.

This requires connected teams, better questions, carefully designed activity and measurement that reflects how people actually behave now.

That is how brands will thrive through the AI evolution.

And it is what being a digital problem-solving agency now means to us.