Artificial intelligence is rapidly moving up the marketing agenda, but the speed of adoption should not necessarily determine the speed of investment. For businesses, the challenge is to determine when, where, and how much to invest in AI without underfunding the activities that are still delivering results.
The findings from Digital 2026 Mid-Year Global Update Report, by We Are Social and Manochi, put this debate into perspective: today, 14.8% of people discover new brands through AI tools, compared with 32.4% who discover them through search engines and 31.2% through television.
The gap is also significant when it comes to pre-purchase research: 45.8% of connected adults use search engines to research brands, compared with 22.1% who turn to AI tools.
This does not mean companies should wait to incorporate AI. It means something more important from a budget management perspective: innovation does not require abandoning what already works.
The first decision: protect what is already delivering results
The conversation around AI is often framed as a replacement: less search, more AI. But the data does not yet support such a simple conclusion. Although the use of search engines to research brands has declined by more than 10% over the past 18 months—from 51.3% to 45.8%—search remains the dominant channel at this stage of the customer journey.
Google also receives approximately seven times more unique monthly visitors than ChatGPT, while its traffic to leading websites has even increased slightly over the past six months.
That is why, before reallocating investment, the question should be: which parts of our marketing mix are working, and which ones need to be tested?
The 80/20 rule: invest in results and create room to experiment
Simon Kemp, the study’s lead analyst, offers a clear benchmark: allocate at least 80% of the budget to what we already know works and use the remaining 20% to experiment.
The principle is simple. The 80% protects channels and activities with proven results. The 20% creates room to test AI tools, new formats, emerging channels, or new ways to optimize the current marketing mix.
But experimentation does not mean investing without clear criteria. Every test should have:
- A specific objective.
- A success metric.
- A clear criterion for deciding whether to scale or stop.
If an initiative consistently delivers results, it stops being experimental and can earn a larger share of the budget.
AI should not receive budget simply because it is AI
This is probably the most important point for marketing teams. A new tool does not automatically become a good investment simply because it uses artificial intelligence.
Before allocating more budget to it, the company should be able to answer three key questions: What business problem does it solve? What metric should it impact? And what result would justify increasing the investment?
This approach helps prevent the pressure to “not fall behind” from turning into budget decisions that are difficult to justify.
First-party data should carry more weight than the hype
The report also warns that headlines about digital behavior can exaggerate or distort trends. With AI, the gap between perception and actual behavior is particularly relevant.
Before changing the investment mix, companies should compare market trends against their own data and, ultimately, their business impact.
Because a global trend does not necessarily mean that your audience behaves the same way. And if your own data contradicts the hype, it should carry more weight in the decision.
The goal is not to invest less in AI. It is to invest better.
The strategic question for marketing teams is where AI can generate an impact we are not achieving today—and how we will prove it.
The 80/20 rule provides a starting point for answering that question: protect what is already delivering results and create room to experiment with what could deliver them tomorrow.
Because in a market where technology is changing rapidly, the advantage lies in knowing what deserves investment, what deserves experimentation, and what deserves to be scaled.