From keywords to intent: how AI is reshaping modern Search

With the rise of AI and changes in how people search, we’re now seeing much longer queries – often full sentences. A couple of years ago someone might have searched for something like “red dress”. Now it’s “red dress cap sleeve sequin for work Christmas party try-on near me.”

As queries get more specific, brands can match intent much more precisely. That doesn’t just improve efficiency; it improves relevance and user experience. If you have that exact product, available near the user, you’re much more likely to win the sale. Conversion rates are increasing as a result, which is driving efficiency for brands.

Absolutely. The operational challenge is the sheer number of possible keywords. You end up with millions of different queries, and no one person can think of – let alone manage – them manually in an ad account.

As Google has previously shared, around 15% of searches every day are completely new, reflecting how dynamically consumer intent is constantly evolving. At that scale, automation becomes essential, which is why leaning into tools like Performance Max, Demand Gen and broad match allows marketers to respond to real demand while applying human oversight and clear strategic guardrails.

Demand-led growth, by nature, needs to be led by demand.

If you sell garden furniture, you need to be ready to push budget when the weather improves. Sometimes that’s February, sometimes it’s May. What you can’t do is force spend into a fixed calendar when demand simply isn’t there.

This approach allows brands to align spend with consumer readiness, improving performance and reducing wasted exposure, rather than chasing activity for activity’s sake.

Beyond budget agility, there are two core elements: creative and data signals.

Signals are a relatively new concept in marketing. I think of them like prompts. The richer and more detailed the signal, the better the output.

For example, we work with a tile client whose Performance Max campaign was already performing well. But when we overlaid sample-request audience data, performance accelerated because the system could better understand who was most likely to convert.

If brands don’t yet have rich first-party data, there are still starting points. Enhanced conversions can help mitigate signal loss from declining consent rates by backfilling data server-to-server, and uploaded customer lists can provide early signals. What matters is progress, not perfection.

Up to 49% of campaign performance comes from creative.

Historically, everything centred on one hero TV ad. Now it’s about building modular creative systems with multiple variations.

For example, with a mattress client, we tested a designed asset against an organic-style person-to-camera video. The designed one won head-to-head. But when we combined both into a third variant, that third variant dramatically outperformed each individually. It’s about having the right ad for the right person — some people respond better to designed assets, others to organic-style creative.

We also use a social-bias framework, including social proof, awards, authority cues, next-day delivery and “Which? Best Buy” messaging. Every client we’ve applied this framework to has seen strong performance improvements.

Ultimately, it’s about giving algorithms enough creative variety to assemble the right message for the right person in real time.

AI is incredibly powerful, but it works best with structure. At Connective3, we use a “rocks in the jar” analogy.

• Exact match terms are the rocks. They are your hero terms and fill the jar first.
• Broad match terms are the pebbles. They add incremental reach and performance.
• Performance Max is the sand – filling the remaining gaps and identifying additional opportunities.

Once the jar is full, Demand Gen campaigns are what grow the jar itself by creating new demand. This approach lets brands capture every relevant opportunity while still maintaining strategic control.

Measurement needs to stand up at board level. This requires marketing to speak the language of the CFO, not just marketing metrics. That means shifting the conversation beyond return on advertising spend (ROAS) and cost per acquisition (CPA), towards profitability, margins and business impact.

A good example is a client that had significantly overbought garden furniture. As warehouses filled up, delivery times increased and conversion rates fell. Rather than optimising in isolation, we ran an in-stock labelling campaign designed to clear inventory and relieve operational pressure.

It may not have been their highest-margin month, but it solved a critical business problem and improved customer experience. When marketing performance is measured against what matters most to the wider business (not just marketing efficiency), it builds trust. And that trust is what unlocks budget, confidence and long-term growth.

When you are running 15 or more tests a year, not all of them will deliver a clear performance uplift and that has to be accepted upfront. A test that does not perform as expected is not a setback. It is still a learning opportunity. You analyse what happened, adjust, and move forward.

Psychological safety comes from treating everything as a test and reinforcing that all outcomes generate valuable data. Knowing what not to do can be just as important as knowing what to double down on.

The key is strong alignment between brand and agency. Decisions are made together, expectations are clear, and outcomes are shared openly, whether they are positive or more neutral. That removes fear and enables faster learning. The only real risk is running activity without proper test design. If a test is not structured correctly, you do not learn anything. With the right framework in place, every test moves the organisation forward.

First, deeply understand your business realities, including operational constraints such as stock levels or delivery capacity that can directly impact performance.

Second, partner closely with your agency to translate those realities into Search strategy.

Third, be prepared to invest in areas like data and creative that historically haven’t received the same focus.

If you start seeing more agile budgets, richer signals and creative variation scaling effectively, you’re building a Search strategy designed for how people actually express intent today.