Predictive AI tools that scan social conversations, ad libraries, and consumer sentiment at a scale no planner could match are being pitched to Indian creative agencies as insurance against the one thing this business has never been able to guarantee: audience response. The promise is straightforward - test the idea before you shoot it, spend less on campaigns that were never going to work, and put more behind the ones that might. But the real question facing creative leaders is whether a probability score should drive how agencies approve or kill ideas, or whether the industry that has always run on instinct should keep the final call with humans.
The appetite for prediction is already here. Globally, 41% of marketing and advertising decision-makers said they planned to use AI for campaign activation in 2025, up from 31% the year before, according to DoubleVerify's 2025 Global Insights report. Adobe's 2025 AI and Digital Trends India study found that 23% of Indian businesses were already reporting measurable returns from generative AI, the highest share in the Asia Pacific region. Forrester research cited by WPP's global chief technology officer found that three in four ad industry executives said their companies were using generative AI tools in 2025, up from 61% a year earlier.
The detective and the forensics lab
What predictive AI tools are genuinely good at is compression. A human planner scanning hundreds of hours of content across platforms, languages, and formats to spot an emerging pattern would take weeks. AI can do a version of that work in hours, surfacing a spike in conversation before it becomes common knowledge to anyone watching a single feed.
Rachita Goel, Strategy Director at 22feet Tribal, draws a sharp line between the tool and the judgement. "AI can get you early signals, but it can't predict culture on its own," she said. "We like to think of it this way: humans are the detective, AI is the forensics lab. The lab can process evidence at scale, but it can't decide what the evidence means without a detective understanding."
Sundeep Sehgal, Senior Vice President and Executive Creative Director at VML India, argues that India's diversity makes cultural prediction especially unreliable. "India is not one culture. It is many cultures moving at the same time: different languages, regions, age groups, internet habits," he said. "Something can trend in Coimbatore and not matter in Chandigarh. AI can detect rising conversations and email spikes, but it cannot fully understand the context of why people care, the emotion driving it, or whether it will fade in two days."
Built to reward the familiar
The more uncomfortable problem sits underneath the accuracy debate. Every prediction model is trained on precedent, on what has already worked, which means it has no frame of reference for the idea that has never been tried. A campaign that looks nothing like anything the model has seen will, by definition, score lower than a safe and familiar idea.
Gopa Menon, Director and Principal at TheBlurr, a brand strategy and innovation consultancy, makes the case plainly. "Prediction tools are trained on what's worked before. So, by design, they'll always score a safe, familiar idea higher than a genuinely new one, because the new has no past pattern to match against," he said. "An agency that leans too hard on these scores ends up evaluating what won't fail instead of the one that would be different."
Rohit Dhamija, Founder and Creative Director at Good Design and Digital, sees the same mechanism in blunter terms. "AI is excited about the past, it is excited about content, it is excited about what has already been done," he said. "But advertising is not built on what has happened. Advertising is built on new ideas."
Dashboards are hygiene. Interpretation is advantage.
None of this has stopped the demand for these tools from growing inside client conversations. Agencies now build them into the pitch itself. "Yes, and rightly so," said Sehgal. "Marketing today runs at internet speed. Clients want early signals, volatility alerts, meme tracking, regional pulse checks." But he separates the tool from the value of a real advantage. "The smartest clients aren't asking what's trending. They're asking, 'What should we do about it?' Dashboards are hygiene. Interpretation is advantage."
Carmen Rodriguez, who runs the brand at Menon's firm, frames the same concern in terms of tone and voice: "The audience can tell when a brand is using a tool rather than thinking for itself. They don't want to be served content that feels like it came from a machine."
Goel, however, frames the client appetite as a genuine operational benefit. "We've been meeting clients who are increasingly curious about this, but what they're most excited about is the speed that dashboards have," she said. "They're eager about accelerating research time and getting to insights quicker."
Why this matters for creatives
The disagreement is not over whether the tools should be turned off entirely. It is over how much weight a probability score should carry once it reaches the room where the actual decision gets made - and how many agencies still have the conviction to overrule it when it says no.
"AI should be used as a tool, not as a surrogate truth," said Dhamija. "It can analyze, challenge, test, and identify patterns. But the decision to break a pattern still has to come from creative people, from planners, and ultimately from people who understand that culture doesn't always allow you to be formulaic."
For creatives, the practical takeaway is to treat prediction scores as a source of information rather than a verdict. The agencies that will stay ahead are the ones that use the data to sharpen their instinct, not replace it with a dashboard. If your work scores low because it's genuinely new, that's a sign the model is doing what it was designed to do - and the decision to back the work still belongs to you.
Your membership also unlocks:
