Read
← Back to IdeasDialogue With Data: The Rising Bar For Customer-Centricity
Imagine a world where you could access the voice of your customer at any moment – from formulating the brief and developing the strategy to brainstorming the solution and iterating the execution. This isn’t a business utopia. This is the new reality synthetic audiences are making possible.
Oh, The AI Irony
Engaging with customers is easier and more difficult than ever.
In theory, businesses are now able to identify, engage, and deliver on customer needs faster and more effectively than ever before. However, this capability has a double edge.
On the one hand, expectations have skyrocketed. Customers now demand instant, relevant, and seamless end-to-end experiences.
On the other hand, there are more businesses generating more content than ever, turning the battle for engagement into a war of attrition.
The irony is that the very technology that has lowered the barrier to entry for brands has simultaneously raised the bar for customer expectations, making it more challenging for businesses to truly satisfy these needs.
In principle, the solution is nothing new: understanding what customers think, feel, want, and need remains the only path to serving them. But between budget squeezes and the need for speed, relying solely on traditional research forces an impossible trade-off.
In practice, the solution is something new. And what’s more ironic still is that the technology fuelling this paradox also holds the keys to solving it.
Dialogue with Data
Imagine if you could access your customers’ voices at any moment – working with them to consistently uncover their nuanced needs and individual insight, to maximise the chances of serving and engaging them.
“What’s your biggest frustration when you’re at checkout?”
“How do you feel about this campaign headline?”
“How do you feel about this campaign headline, now?”
“Red or blue?”
Well, with synthetic audiences, now you can.
Synthetic audiences are AI representations of human groups and individuals, built on real customer data and audience insights, allowing for the simulation of conversations, reactions, and focus groups.
They enable dialogue with data.
The uploaded customer data informs each persona’s personality, character, preferences and, thought process. The AI model underneath it acts as the central nervous system, coordinating responses. This interplay creates impressive predictive power, allowing these audiences to respond with an uncanny accuracy of tone, personality, and sentiment – even when the conversation ventures beyond the domains of their training data. And the richer, broader, and more accurate the data on the human audiences, the richer, broader, more accurate the synthetic audiences’ responses.
You need the human to access the synthetic, but once you have this data foundation, you gain access to highly accurate representations of these voices, whenever, wherever you need them.
The Synthetic Advantage
Synthetic audiences offer four main business benefits:
-
Availability: They’re always-on, ready whenever insight is needed.
-
Efficiency: They enable instant response without high ongoing recruitment costs.
-
Dynamism: They can assume contexts effortlessly, enabling broad situational simulations.
-
Resilience: They won’t tire or take offence, enabling you to explore sensitive questions that you might hesitate to ask real consumers.
This addresses all-too-familiar research frustrations:
“If only we could get their views on the product now.”
“We don’t have the budget or time to run another round of research.”
“We should have asked this question...”
“They might not feel comfortable answering that.”
But how accurate are they? Like any good answer, it depends. It depends upon the quantity and quality of data provided, how they’re configured, and whether they’re used for the right use-case.
A recent study led by Stanford University and Google DeepMind found that synthetic audiences trained on interview data matched human survey responses with 85% accuracy and mimicked social behaviour with 98% accuracy.
At Ogilvy, we conducted our own form of ‘Turing Test’ using two campaign routes for an automotive brand to ascertain the similarity in responses between human audiences and their synthetic counterparts. We began with the synthetic audiences, which preferred route B (surprising the team who were confident in route A). When tested with the real humans, they reached identical conclusions, highlighted the same specific reasons, and communicated this with remarkably similar language.
These results aren’t isolated. As the models which underpin them increase in sophistication, the predictive power and fidelity of synthetic audiences only grows.
The Actual + The Artificial
These results are impressive. But the utmost respect must be held for the 15% gap of unpredictable human behaviour that cannot be captured by these AI representations.
It’s often these non-statistically significant moments – the nuggets of randomness and unique tacit knowledge – that become the golden insights unlocking a strategy’s success or catching a crucial flaw.
And thus, most importantly of all, synthetic audiences should not replace real human research. Instead, they should work alongside it, providing access to customer perspectives when human customer interaction is otherwise not possible.
They are an augmentation, not an alternative.
The New Standard
These AI representations of target audiences – built and used correctly – unlock a new level of customer-centricity.
With the combination of the actual and the artificial, businesses can continuously consider the wants, needs, barriers, and drivers of their customers without sacrificing the speed required to be relevant or the depth required to be accurate.
Soon, every truly customer-centric company will have access to synthetic versions of their target audiences. The bar has been raised; reach it and start your dialogue with data.