By Rachel Leaver
Every brand today is under the same pressure: move faster, stretch budgets further, and still make confident decisions. But speed without the right foundations doesn't produce better insight, it produces confident-sounding guesswork. The gap between fluency and truth is the real risk nobody is talking about.
A year ago, we launched a global GenAI-powered insights solution. Since then, it’s been used across 60+ countries and embedded into the way teams make everyday decisions, from strategy and planning through to innovation and creative development.
Much of what’s been written about GenAI-powered audience intelligence is still grounded in pilots or small-scale deployments so here’s what we’ve learned from putting it to work at enterprise level over the last 12 months. Use these learnings to evaluate what to look for in an AI-powered insights capability, and what to watch out for.
What has changed with AI powered insights
Within the first year of deployment, our teams were able to easily:
Explore audience’s attitudes, needs, and behaviours in real time without waiting for new primary research or having to look through multitudes of spreadsheets and insight presentations.
Accelerate insight generation by quickly testing questions, hypotheses, and scenarios against a data-grounded consumer view.
Bring the consumer perspective into everyday decisions across strategy, planning, innovation, and creative development.
Democratise access to insight so non-researchers (strategists, creatives, and client teams) can engage directly with consumer understanding.
Stress-test experiences, messaging and content early by pressure testing propositions and creative territories and killing weak ideas faster.
Make it Real: Moving beyond static insights with engaging, conversational outputs and vivid, human-like quotes that bring the audience voice into the room
The benefits that showed up
One advantage of operating at this scale over time is that you start to see patterns that don't show up in pilots. Some of the most significant value came from places we didn't expect.
It broke down data silos. Different teams often commission research or run standalone analyses. Connecting that intelligence means brands extract more from existing investment and stop commissioning work that already exists somewhere else.
It strengthened insight capability. Great insights typically come from diverse teams interpreting the same evidence in different ways to get to new spaces. While an AI isn’t “diverse,” it can interpret the data or research outputs in a different way that you might not have thought, enabling you explore areas you wouldn’t have got to on your own.
It sparked creativity. The more ‘human’ interpretation layer helped strategists elevate their thinking and move from data to a sharper, more imaginative direction.
It drove decisions with real impact. From ranking messages by audience relevance to shortlisting partnership opportunities and reshaping advertising strategy, the solution has turned new insight into clearer, more confident action.
Key learnings (and watchouts) for your solutions
AI will always give you an answer (even when it probably shouldn’t). Don’t confuse fluency with truth. Build the habit of asking: what led to that answer and checking the underlying data that supports it.
Know your data sources. Users need to understand the datasets (and the LLM context) to judge whether the evidence can legitimately answer the question
Unique data is the differentiator. Data is king. Without proprietary or distinctive data inputs, your outputs (and the decisions they inform) will converge with competitors’. Data quality and distinctiveness are what drive differentiated thinking.
Plan for model drift. Performance can change over time, so continuous user feedback and monitoring are essential. This isn’t something you “user test once” and move on, rigorous testing needs to be ongoing.
The measure of an AI insight capability
A year in, the biggest takeaway is simple: GenAI-enabled insight tools create real value when they’re grounded in strong data, embedded into everyday workflows, and paired with the right user habits (curiosity, scepticism, and feedback). Technology matters but adoption, data, and governance matter just as much. As our insight capabilities continue to evolve, that principle is what keeps outputs meaningful, decisions defensible, and the humans in the loop where they belong.
If you're exploring what an AI-powered insight capability could look like for your organisation, dentsu.Connect is built on exactly these principles. Combining data, AI, and human judgment, it turns audience intelligence into business decisions across media and creative.
