BOMBAYDC Edits: Research lead Cheshta Gulati drives the case for ‘reassurance’ as a key asset in e-commerce

BOMBAYDC Edits delves into the people, principles, and processes behind our work, providing meaningful conversations and valuable insights.

BOMBAYDC’s latest report, ‘The Human Override’ argues that beneath the frictionless interfaces of e-commerce lies a quieter gap. It’s the absence of reassurance, guidance, and emotional validation that physical retail once provided naturally. The report argues that the next competitive edge in commerce won’t come from reducing delivery times, but from restoring the human touch. As AI evolves from generative assistance to agentic automation, the real opportunity lies in designing ‘human-aware AI’.  These are systems that don’t just execute tasks, but understand context, hesitation, and intent. It’s a shift from optimising transactions to nurturing trust.

Independent journalist Payal Khandelwal speaks with Cheshta Gulati, research lead at BOMBAYDC, about the thinking behind the report, the development of their Decision Comfort Index, and why reassurance may be the most undervalued currency in digital commerce today.

Can you give us a comprehensive background to the report?

The report primarily examines how the retail and lifestyle industry has successfully engineered frictionless journeys: faster deliveries, seamless interfaces, easy returns, access to international products, and 24/7 availability. Convenience, in many ways, is no longer the problem. However, in that race toward efficiency, something has been lost: the human touch.

While transactions have become easier, the emotional dimension of shopping — guidance, reassurance, empathy, contextual understanding — hasn’t evolved at the same pace. The data strongly suggests that the desire for human connection hasn’t disappeared. In fact, consumers still really value it. And this is where the report shifts focus: how can e-commerce solve for the human touch in a digital-first environment?

The broader context is, of course, AI, which is now embedded across work, commerce, and personal life. The report outlines two broad types of AI currently shaping commerce: Generative AI which supports conversation, content, and assistance, and Agentic AI - systems that can take action on a user’s behalf. Agentic AI further optimises convenience: you approve a task, and it completes it for you; you like a product, it purchases it; it asks for payment confirmation, and the transaction is done. Again, this ultimately solves for efficiency.

The report proposes a shift toward ‘human-aware AI.’ The suggestion here is to deliberately design AI systems that are sensitive to context, emotion, nuance, and intent. To build AI that doesn’t just execute tasks, but understands the human behind them. Human-aware AI is positioned as the bridge between frictionless commerce and emotionally resonant commerce. It suggests that the next competitive advantage in retail would be sensitivity. 

During the research, was there a specific moment, insight, or data point that really stood out that clearly indicated that convenience alone is no longer enough to win in the Indian market, particularly in the context of e-commerce and quick commerce?

The journey today is largely frictionless, and in many ways, the operational side of shopping has been optimised. And yet, people still hesitate. We see it in behaviour all the time: consumers add products to their carts but abandon them before completing the purchase. The friction isn’t technical anymore, it’s psychological. 

Usually, when we walk into an offline store, we ask questions, and a sales assistant answers them. That plays a huge role in decision-making and builds confidence in real time. That layer of human support, which exists naturally in physical retail, is largely missing online. Convenience has been solved for, Confidence has not.
Shoppers still ask themselves: Am I buying the right product? Will this work for me? Is this worth the money? That confidence gap remains. Around 68% of Indian consumers want guidance, particularly for high-value purchases. 78% prefer human support over fully automated interactions. That’s a significant signal.

How was the Decision Comfort Index framework developed? What specific behavioural markers did you look at to determine whether a user is experiencing psychological friction versus simply browsing casually?

The Decision Comfort Index (DCI) emerged from our ongoing work with clients. Typically, clients share detailed data analytics with us: where users drop off; where they spend a disproportionate amount of time without converting; which screens stall the journey; and where the funnel breaks. That data gives us a clear view of what is happening across the product experience, but not why it’s happening.

Often, the “why” is inferred. We rely on patterns from previous projects, industry benchmarks, or educated assumptions. Deeper analysis can sometimes reveal correlations — for instance, if we study additional variables, we might conclude that a drop-off is occurring because of complexity or unclear information. But analytics alone rarely give us a precise, structured understanding of the psychological drivers behind user hesitation.

That’s where the DCI comes in. It evaluates the entire journey leading up to conversion, and tries to identify potential friction points, particularly those rooted in psychology. When a digital product is evaluated using DCI, we can assess whether drop-offs are being caused by logistical barriers such as confusing navigation, unclear calls to action, or slow loading times, or by psychological barriers like cognitive load, loss aversion, and decision anxiety. Once we understand whether the friction is emotional, cognitive, or structural, we can design targeted solutions. 

DCI doesn’t replace analytics; it builds on them. It helps translate behavioural data into human insight, allowing us to detect and address the deeper frictions that traditional metrics often overlook.

Beyond reducing cart abandonment and acknowledging that brands may have multiple objectives, how can DCI help predict long-term value, such as repeat purchases and customer loyalty?

If you don’t truly understand the why behind the hesitation patterns, you’re missing something critical: how the customer is actually feeling during the journey. Your product may be excellent, but if the digital experience creates friction or doubt, the customer might choose to buy the same product on another platform that feels more trustworthy or easier to navigate. That happens all the time.

The gaps, whether informational, emotional, or trust-related, can be identified and closed with the help of DCI. And the impact goes beyond just improving a single transaction. If you create a journey where users feel confident, understood, and supported (and of course if the product ultimately delivers), you don’t just win a conversion, you build the foundation for repeat purchase and long-term loyalty.

Is there a particular brand or platform where you’ve applied this framework?

Yes, we applied this framework with Lotto when the brand was re-entering the market. It wasn’t exactly a rebrand, but a strategic comeback. As part of that process, we conducted a behavioural diagnosis of the digital buying journey. 

When people buy shoes, there’s an inherent need to touch, feel, and try them on. Fit is critical. In a physical store, you can wear the shoe, walk around, and immediately assess comfort. Online, that tactile support is missing, and we identified that ‘size guide’ is often where drop-offs happen. There are two psychological reasons for this. It creates cognitive overload, as size charts can be complex and overwhelming, especially when users are trying to map their measurements across different standards (UK, US, EU, etc.). Our working memory is limited, and when information becomes too dense or difficult to process, the brain simply disengages.

Second, there’s loss aversion. Humans naturally fear making the wrong choice. The pain of getting the wrong size, the hassle of returns, wasted time, and uncertainty feels greater than the pleasure of a successful purchase. So at that moment, doubt peaks. 

We recommended two layers of support at that exact decision point. First, if the user had previously made a purchase, we suggested showing a personalised prompt to reduce cognitive effort and increase confidence. Second, we added a clear message on the size guide screen, where the loss aversion peaks, that if the size doesn’t fit, the customer can exchange it for free. Lotto introduced a framing along the lines of: ‘Try it on for 15 days.’ That subtle shift gave users a sense of ownership, and they could feel like they’re just “trying” rather than “committing”. Interestingly, when people feel ownership, even temporarily, they’re more likely to keep the product. And this is a clear example of how the Decision Comfort approach works.  

You’ve already addressed part of my next question which is about how the fear of making the wrong choice often outweighs the excitement of making the right one. But specifically, how can agentic AI effectively de-risk high-stakes purchases without overwhelming users with even more information? 

Agentic AI by itself isn’t inherently human-aware. It can take action, automate tasks, and complete transactions on a user’s behalf, but unless we intentionally design it to account for human psychology, it won’t provide empathy or contextual sensitivity. So when we talk about ‘human-aware agentic AI,’ what we really mean is designing those systems with behavioural insight embedded into them.

Now, what is the one non-negotiable human trait that digital products must replicate if they want to survive the shift from generative AI to agentic AI? If I had to choose one, it would be ‘reassurance’. Digital products must anticipate hesitation and respond with trust-building cues, whether that’s validation, guarantees, social proof, contextual guidance, or personalised suggestions. Ultimately, what physical retail does naturally, which is building confidence through human interaction, digital platforms must now deliberately recreate. 

Why has reassurance, despite being such a fundamental human need in decision-making, not been prioritised in digital experiences?

I also think part of the issue is the limited integration of behavioural science into product and platform design. Research often gets approached in a surface-level way. Teams may rely on existing analytics, industry benchmarks, or what’s already available publicly. But that doesn’t always capture the deeper psychological pain points of users. 

It’s a structural gap, as many e-commerce and digital product teams are heavily focused on performance metrics, feature rollouts, and operational efficiency. Deep behavioural research, on the other hand, takes time, investment, and cross-functional alignment, and not every organisation is able to prioritise that. If more brands invested in behavioural research, not just usability testing, but true behavioural diagnosis, they would likely uncover multiple hidden gaps in their digital journeys. And those gaps often aren’t obvious from analytics alone.

This feels like a good segue into the next part, where I’d love to talk more about behavioural science. Can you tell me a bit more about your role at Bombay DC, and what your journey has been like so far?

At Bombay DC, I work specifically in behavioural science, although not all of my work is purely behavioural science-led. 

Behavioural science is a rigorous, time-intensive process. It requires research depth, stakeholder alignment, and buy-in. And not every client who approaches us is necessarily looking for that level of involvement. That said, a significant part of our work does include a lot of structured analysis. We audit digital platforms, identify usability gaps, study friction points, conduct stakeholder interviews, analyse competitors, and speak directly with customers.

For example, we worked with an edtech platform, which caters to CBSE students primarily between grades four and nine. It’s a school-aligned app which knows about upcoming exams’ portions and schedules, and tailors content accordingly. The client’s goal was to increase word of mouth downloads, and therefore we knew we had to make the app valuable for students. We began by conducting interviews to understand students’ needs, expectations, and pain points. And then developed hypotheses and recommendations grounded in behavioural science principles such as loss aversion, goal gradient effects, and cognitive overload. Every recommendation was backed by behavioural reasoning, and that gave the recommendations more strategic weight.

Research doesn’t guarantee success, but it significantly increases the probability of effectiveness. Ultimately, recommendations that are research-backed are not just better argued; they are more likely to deliver meaningful outcomes.

If brands treated behavioural science as core infrastructure rather than an add-on or optional layer, how would that fundamentally change digital products? And how would the role of AI within those digital products evolve if behavioural science were embedded at the foundation?

If behavioural science were treated as a core part of product infrastructure rather than an add-on, digital products would fundamentally be designed around how people actually think and behave, not just around what seems logically efficient. 

One of the central ideas in behavioural science is that humans are not purely rational decision-makers. If brands truly internalised that human behaviour is often irrational, emotional, and context-driven, they would design products very differently. Features and functionalities would be built around real cognitive patterns, biases, and emotional triggers, rather than assumptions about “ideal” user behaviour.

Now, when you layer AI into that system, especially human-aware agentic AI, the shift becomes even more powerful. AI becomes proactive rather than reactive. The result isn’t just higher conversions; it’s stronger loyalty, deeper engagement, and sustained trust. 

From a behavioural science perspective, as AI begins to make more decisions on behalf of users, how do we design interfaces that preserve user agency and trust? 

From what I’ve seen in research, both younger users and even older demographics are becoming increasingly critical about how their personal data is being used. Across industries, there’s a growing demand for transparency from digital platforms.

Users need to clearly understand what the AI is doing, why it’s making a recommendation, and how their data is being used. If that layer of clarity is missing, trust erodes quickly. When people feel like decisions are being made for them without visibility or explanation, it can feel manipulative.

Designing for an agency means ensuring that AI supports decisions rather than overrides them. The system can suggest, guide, or automate, but the user should still feel in control.