For years, digital advertising has relied on A/B tests, broad segments and historical performance. Now a new layer is being added to the mix. Brands are experimenting with predictive creative systems that read signals from audiences in real time and adjust what they see almost instantly.
Instead of one film and a fixed media plan, marketers are starting to work with dozens of creative variations. Machine learning models monitor how people watch, react and scroll. On the back end, emotion AI tools analyse facial expressions, voice tones or interaction patterns to identify whether a piece of content is landing as intended. The output could be as simple as switching a thumbnail or as complex as rewriting a line of copy on the fly.
The ambition is clear. If emotions drive a large share of purchase decisions, brands want to know how people feel in the moment, not just what they clicked last month. The debate now is about how far this can go, and how much of that emotional decisioning should be handed to algorithms.
From automation to anticipation
Globally, the emotion AI market is projected to grow from around 5.7 billion dollars in 2025 to nearly 38.5 billion dollars by 2035, according to recent forecasts, with marketing as a key use case. That growth is driven by a simple promise: if brands can sense attention, confusion or delight earlier, they can predict which creative will work before spending heavily on distribution.
In India, the broader AI context is already mature. Adobe’s 2025 AI and Digital Trends India snapshot reports that 23 per cent of Indian businesses are seeing measurable ROI from generative AI. Seventy three per cent say AI speeds content ideation and production, and 67 per cent report higher productivity and efficiency. This base of AI-led content workflows is what makes predictive creative possible.
Prativa Mohapatra, Vice President and Managing Director, Adobe India, summed up the shift by saying that “Indian businesses are setting the global pace for realizing ROI on AI initiatives as most are improving scale, speed and efficiencies.” Her point is reflected in marketing teams where AI already handles large parts of copy generation, asset resizing and versioning. Predictive creative is the next layer, where the machine does not only generate assets but also decides which ones to show, to whom and when.
Early experiments have often come from global brands. Lexus worked with emotion AI vendor tools to tailor video creative based on viewers’ facial expressions and attention signals for a campaign in Europe and the US. In another example, analytics company Realeyes reported that pre testing emotion data helped Flowers Foods improve the efficiency of its video creative by around 25 per cent, as weaker edits were dropped early. These are not yet fully real time, but they show how emotional response is starting to influence creative decisions before media spends go live.
Indian brands move from segments to moments
Indian brands are now building their own versions of this playbook, often starting with personalisation and then layering predictive elements. Mondelez India’s Oreo “Say It With Oreo” campaign used AI text and voice engines to turn people’s prompts into personalised voice messages in Farhan Akhtar’s voice, turning awkward conversations into playful exchanges at scale. Cadbury Silk’s “The Story of Us” invited couples to generate personalised animated love stories using AI, creating hundreds of thousands of variations around a common emotional idea of everyday love.
Gifting brand FNP has gone a step further with fully AI built films. Its Raksha Bandhan campaign “Door ho ya paas, Rakhi banayein khaas” uses AI generated visuals to tell stories of siblings connected across borders, while the brand’s systems optimise which edits work best across platforms.
For Avi Kumar, Chief Marketing Officer at FNP, the goal is to use AI without losing emotional nuance. “AI lets us move from segments to moments. It personalizes three layers: the product, the story, and the journey,” he told Adgully, describing how FNP adjusts offerings, narratives and timing based on data. Predictive creative sits on top of this stack, deciding which version of that story to push when someone engages with a Rakhi reel, abandons a cart or revisits a gifting page late at night.
Banking and fintech marketers see similar possibilities. MVS Murthy, Chief Marketing Officer at Federal Bank, has described how AI changes the cadence of communication. “AI allows a marketer to be in an election kind of mode,” he said, contrasting always-on responsiveness with seasonal campaigns. In a predictive creative model, that “election mode” means running multiple message lines simultaneously, and letting models choose whether a user in a particular city and life stage should see a rational EMI message or a reassurance-driven film about security.
Payment brands also see predictive creative as a way to reduce friction rather than just increase noise. Anuradha Aggarwal, CMO of Amazon Pay, has spoken about this shift, saying, “AI-driven dynamic marketing, making personalized advertising, that’s something exciting.” In practical terms, that could mean an ad that changes the cashback focus or the use case shown on screen depending on whether the user has just paid a utility bill, ordered food or booked travel.
How predictive creative actually works
Behind the scenes, most predictive creative systems follow a similar pattern. First, they collect signals. On social platforms, those signals are often watch time, scroll speed, replays, taps on sound, comments and saves. On some closed environments, brands work with partners who can read facial micro expressions or voice modulations (with consent) through emotion AI tools. Those signals are then converted into scores for attention, sentiment or emotional intensity.
Second, models classify people into moment based cohorts rather than only static segments. Instead of simply saying “urban millennial parent in Bengaluru”, the system may tag someone as “stressed commuter scrolling in silence” or “lean back viewer watching with sound on”, based on that session’s behaviour. This is where Avi Kumar’s line about moving from segments to moments becomes operational.
Third, the creative pool is tagged in more detail. A brand film is no longer just “Version A” or “Version B”. Each cut is labelled for pace, emotional tone, soundtrack, product window and call to action. Over time, marketers build a library that knows which combination tends to work better for “nostalgic, family” moments versus “quick, transactional” ones.
Finally, generative AI comes into play. Once the system has a prediction that a certain combination is more likely to work, it can auto generate small variations within brand guidelines. That might be rewriting the opening line to be more empathetic, softening colours for late night viewing, or switching from a celebratory track to a calmer one when it detects low attention.
Many global creative technology platforms are racing to offer this stack to brands. Some focus on pre testing, where emotion scores guide edits before a campaign. Others plug into ad platforms to adjust creative dynamically. In both cases, the promise is that creative decisions become as data informed as media buying, without turning every ad into the same template.
Measuring uplift without losing the human touch
One of the key attractions of predictive creative is measurability. Realeyes reported that using emotion based analytics helped Flowers Foods drop underperforming edits early and improve overall campaign efficiency by a quarter. In India, Cadbury Silk’s AI driven “The Story of Us” reportedly generated millions of views and hundreds of thousands of personalised stories, contributing to double digit growth for the brand during the Valentine’s period, according to award submissions and case studies.
For Indian marketers, these numbers sit alongside broader AI adoption metrics. The Adobe India snapshot notes that 73 per cent of businesses see faster content production with generative AI, and 67 per cent report higher productivity. Predictive creative is often positioned as a way to convert that speed into better business outcomes, by feeding performance data back into the creative loop.
At the same time, there is a consistent caution against over automation. In a recent MartechAI in depth piece, Nitin Saini, Vice President of Marketing at Mondelez India, put it simply: “The human touch is still central. AI is a creative multiplier, not a replacement.” For predictive creative, that often means humans still define the emotional strategy, write the core narrative and set red lines, while AI handles experimentation at scale.
Practically, brands are building governance around three areas. First, consent and privacy, especially when emotion AI is involved. Most mainstream campaigns today rely on aggregated, anonymised signals and avoid any personally identifiable emotion tracking. Second, bias and exclusion, to ensure that optimisation does not silently favour only certain demographics or regions. Third, brand safety, so that auto generated variations do not dilute tone or make claims that fall outside regulatory limits.
What could come next
As Indian brands deepen their use of customer data platforms and agentic AI, the building blocks for predictive creative are becoming more accessible. Customer journeys that were once mapped in static flowcharts are now being updated in real time by AI agents, which can also instruct creative systems to adapt.
In a banking context, that could mean a video explainer that changes length and framing based on whether the viewer is a first time credit card applicant or a long time customer checking reward points. In travel, airline or hotel brands might alter creative to highlight flexibility, safety or value depending on sentiment trends in social listening. In retail, emotion signals from in store cameras, when used in a privacy safe way, could inform the kind of content pushed to nearby digital screens.
Globally, some marketers imagine a future where a single brand film can have thousands of micro edits that no human editor could ever manage manually, each tuned to a moment in a person’s day. Indian marketers, however, are increasingly clear that such power has to be balanced with clear disclosure and cultural sensitivity. The conversation is shifting from “Can we do this?” to “When is it appropriate to do this, and how do we explain it to consumers?”
For now, predictive creative sits at the experimental edge of social media marketing. It is most visible in high interest categories like fashion, beauty, entertainment, fintech and gifting, where emotional stakes are already high and time spent on platforms is significant. If current adoption trends continue, it is likely to become part of the standard toolkit in the next few years, especially for brands that already see strong ROI from AI in other parts of their stack.
What remains constant is the need for a clear human point of view about what the brand stands for. AI can help sense and respond to emotions at scale. It can suggest which cut to run and which line to test. But the decision about which emotions a brand should evoke, and which ones it should avoid, still rests with marketers who understand the people behind the data.
Disclaimer: All data points and statistics are attributed to published research studies and verified market research. All quotes are either sourced directly or attributed to public statements.
Related Articles
