In a conversation with Brij Pahwa, Editorial Lead, Ashley D'Souza, Chief Digital Officer at Hindustan Coca-Cola Beverages, offers a clear view of how AI is reshaping marketing. From the role of data and MarTech to the limits of automation, D'Souza emphasizes one core idea: tools alone do not drive outcomes. Judgment, context, and “taste” do.
Over the last five years, how has digital transformed?
There’s been a sea change. It’s accelerating. It’s important that all of us absorb these changes. Leaders also need to lead the change, work with teams, make them comfortable, and help them adopt it. Technology is evolving, but how people adapt and evolve is equally important.
Has the pressure to prove ROI made marketing more difficult?
Marketing has never been separated from results. It has always been about numbers. What has changed is how we measure them.
Today’s technology stacks, whether in media planning or programmatic advertising, are completely numbers-driven. The tools available now provide much better control.
The next phase will be even more outcome-led. Today, there is still a gap between what tools suggest and where human judgment comes in. Over time, tools will get closer to outcomes, recommending strategy and execution.
So yes, it has always been about numbers. The difference now is better measurement and better control.
As generative AI advances, does human creativity diminish?
It’s a trade-off.
Marketing has always been about putting a story in front of consumers. Now, with personalization at scale, the challenge is engaging each consumer authentically.
If campaigns become generic, where teams simply input prompts into AI, they will lack engagement. You may have a campaign, but not the connect, conversion, or impact.
There will be more campaigns. Effectiveness will depend on the teams using the tools. Tools alone are not enough.
Is there fear among teams that AI could replace roles?
It is changing. We use GenAI extensively. It will increase volume, but not replace judgment.
The key question is: what are the right marketing objectives? That comes from high-judgment individuals.
With AI-generated content, taste becomes the differentiator. The ability to judge what good looks like. That requires understanding the consumer and their needs.
AI is a force multiplier. But judgment remains human.
What does an ideal MarTech architecture look like today?
Four layers are essential. First, data. It must be robust, accurate, and real-time. Second, the semantic layer. It defines what the data means. Third, the application layer. It enables teams to access data, generate insights, and apply them to campaigns. And across all layers, privacy is critical. Guardrails must be built in from the start.
The stack must be end-to-end. Bolt-on tools do not work effectively.
FMCG brands struggle with first-party data. How do you solve that?
It requires investment.
Companies that prioritize first-party data through research, sampling, and direct engagement can build it. Otherwise, they rely on proxies like secondary data.
It is not readily available. It must be built deliberately.
How do you handle LLMs trained on global data?
Generic LLMs do not perform well without context. They need grounding in local realities such as country, region, and brand context. Applying global insights blindly will not work. Organizations must invest in contextualizing models. Different approaches can work, including SLMs, RAG, or controlled LLM layers. The right choice depends on the use case.
Should FMCG brands move aggressively to D2C?
Digital is an add-on, not a replacement. We need to go where customers are.
The approach is not either-or. It is offline plus online. The human connection remains important.
India is nuanced. Different consumers operate at different levels of digital adoption. The strategy is simple: meet customers where they are.
What is your view on AI agents?
Agents can handle mechanistic tasks. But the key question is not whether they can do something. It is whether they should. For example, AI voice calls are widespread. But their effectiveness in driving engagement is debatable.
Agents can execute tasks but effectiveness depends on context and judgment.
AI can also appear confident when it is wrong. That is a risk. So the focus should be on where automation adds value and where human intervention is needed.
Related Articles
