Why 61% of Online Retailers Get Search Wrong, and How Shardul Aggarwal Is Changing That

In a market flooded with AI tools promising to transform online retail, Shardul Aggarwal, Founder and CEO of ConvoSearch, is focused on a problem most merchants overlook: the search bar.

His platform now processes over one million shopper sessions monthly across more than fifteen e-commerce brands, generating over two million dollars in incremental revenue for clients.

His thesis is direct: if your search bar does not understand what customers are asking for, nothing else matters.

“Decisions to buy don’t happen at checkout,” Shardul explains. “They happen at discovery. If a shopper cannot find what they came for, they leave. And most e-commerce sites are failing at this basic step.”

1. The Foundation: Why Search Is the Hidden Revenue Leak

The numbers reveal a significant gap. According to Algolia, site search users convert at rates up to 50 percent higher than average visitors. These searchers represent only 15 percent of total traffic but generate 45 percent of e-commerce revenue. Yet 61 percent of e-commerce sites perform below acceptable search standards, and 15 percent have search functionality that is fundamentally broken.

This is not a minor inefficiency. Research compiled by Baymard Institute estimates that $260 billion worth of lost orders in the US and EU alone are recoverable through better checkout and discovery design. The customers most likely to buy are often the ones who leave frustrated.

“An e-commerce brand can spend millions on ads to drive traffic,” Shardul notes. “But if those visitors search for something and get zero results, that investment is wasted. The bottleneck is not traffic. It is discovery.”

The problem: Traditional e-commerce search relies on keyword matching, a method that dates back to the earliest days of online retail. The system looks for exact or near-exact matches between the words a customer types and the words in product listings. When the vocabulary does not align, the search returns nothing.

A concrete example: A store sells brown leather shoes called “Sunset Sneakers,” a flagship product featured in advertising. A customer sees an Instagram ad, visits the site, and searches “brown shoes.” The result: zero matches. The system cannot connect “brown shoes” to “Sunset Sneakers” because the keywords do not match.

“This is the vocabulary gap,” Shardul explains. “The disconnect between how shoppers think about products and how merchants describe them. It exists on almost every e-commerce site, and most brands do not even realize how much revenue it costs them.”

2. The Intelligence Layer: Moving from Keywords to Intent

The solution requires a fundamental shift in how search systems operate. Instead of matching keywords, modern AI can interpret what shoppers actually mean.

Shardul’s approach at ConvoSearch focuses on three layers:

Semantic Understanding: The AI analyzes shopper intent, product attributes, and contextual signals to connect queries with relevant products even when exact terminology differs.

Behavioral Learning: The system learns from patterns across its user base, continuously improving its understanding of how shoppers express purchase intent.

Personalized Discovery: Search results adapt based on individual shopper behavior, purchase history, and browsing context.

“We are not just fixing search,” Shardul says. “We are building the revenue automation layer for e-commerce. The goal is to connect shopper intent to purchase, turning outdated search and static merchandising into intelligent, high-converting experiences.”

This approach aligns with broader industry movement. According to McKinsey, personalization efforts can drive 5 to 15 percent revenue lift for retailers, with some implementations achieving even higher results. The more effectively a company applies data to understand customer intent, the greater the returns across both revenue and efficiency metrics.

Amazon’s launch of Rufus, its AI shopping assistant, in early 2024 signaled that major players recognize the category’s importance. The company reported 250 million Rufus users within the first year, with customers who engage with the AI assistant 60 percent more likely to complete purchases. Google has made similar moves with Vertex AI Search for commerce.

(Image: ConvoSearch)

3. The Payoff: Measurable Revenue Impact

When search works correctly, the business results follow. Client outcomes from ConvoSearch demonstrate the scale of impact across different markets and product categories. Uncle Reco, an Australian pop culture merchandise brand processing approximately seven million dollars in quarterly sales, achieved a 24 percent revenue lift after implementation, translating to 1.3 million dollars in quarterly incremental revenue. 

The Closet Lover, a Singapore-based fashion retailer, saw even faster results, with revenue increasing by 60 percent within fifteen days of deployment and adding sixty thousand dollars to monthly results. The platform also serves Oakland Roots SC, the professional soccer team representing the San Francisco Bay Area, powering merchandise discovery for fans across product categories. The diversity of these clients, spanning three continents and ranging from fashion to sports merchandise to pop culture collectibles, suggests the approach works regardless of industry vertical.

“Every vendor claims AI capabilities,” Shardul notes. “The only metric that matters is whether your revenue goes up. We consistently deliver 20 to 25 percent lift in revenue per visitor. If a solution cannot demonstrate that impact, the AI is not working, regardless of how it is marketed.”

4. The Gap: Why Independent Merchants Are Left Behind

Amazon and Google are building AI search for their own ecosystems. Amazon’s AI helps shoppers find products on Amazon. Google’s tools target enterprise retailers with significant technical resources.

The gap is in the middle market: the millions of independent merchants on Shopify, WooCommerce, and similar platforms who lack access to comparable technology.

“A merchant doing a few hundred thousand dollars a year should be able to offer the same intelligent search experience as a billion-dollar retailer,” Shardul argues. “That technology should not be gated by company size or technical sophistication. We bring enterprise-grade product discovery to independent e-commerce.”

This focus on accessibility stems from Shardul’s own experience. As an undergraduate at the Indian Institute of Technology (BHU) in Varanasi, he built his first company, 2xE, helping local artisans create e-commerce websites. The platform reached over 100 stores before shutting down.

“That failure taught me everything,” he recalls. “I kept asking why some stores converted visitors into buyers and others did not. The answer almost always came back to discovery. Could people find what they were looking for?”

After completing a Master’s degree in Human-Computer Interaction at the University of Maryland, Shardul took a product management role at Thomson Reuters, where he built the company’s first personalization and recommendation system. The project increased returning user sessions by 41 percent and improved activation-to-retention conversion by 17 percent.

“That experience at Thomson Reuters showed me what was possible,” he says. “When you understand user intent and build systems that respond to it, the business impact is measurable. The same principles apply to e-commerce, but most online stores are still using search technology from 20 years ago.”

5. The Playbook: Building Search That Actually Works

“Start by measuring what is broken,” Shardul advises. “Most merchants have no idea how many searches on their site return zero results. That number is usually shocking.”

Audit Current Performance: Identify zero-result searches, abandoned search sessions, and conversion rates for searchers versus non-searchers.

Map the Vocabulary Gap: Document the disconnect between how customers search and how products are described.

Prioritize High-Intent Queries: Focus first on searches that indicate strong purchase intent but currently return poor results.

Measure Revenue Impact: Track lift in revenue per visitor, not just search relevance scores or click-through rates.

Iterate Based on Behavior: Use actual shopper patterns to continuously improve results, not assumptions about what customers want.

ConvoSearch was selected for Inception Studio’s Cohort 17, a San Francisco nonprofit accelerator that takes no equity and focuses on experienced founders. According to the program, its community consists of 77 percent repeat founders, with 30 percent having previous successful exits. Portfolio companies have raised over 125 million dollars since the program launched in November 2022.

The company is also developing CloutFarm, a marketing and influencer arm that extends its growth toolkit for e-commerce brands. The vision encompasses the entire shopper journey from discovery through purchase.

Bottom Line

The e-commerce industry has spent years optimizing the journey to the store: ads, influencers, email campaigns, social media. The journey inside the store remains largely broken.

Site search represents one of the highest-leverage opportunities in online retail. The customers most likely to buy are the ones who search. When they cannot find what they want, they leave.

“We are at an inflection point,” Shardul concludes. “AI has made it possible to understand what shoppers actually mean, not just what they type. The brands that adopt this technology will capture revenue that others are losing every day. The ones that do not will keep wondering why their conversion rates are stuck.”

The search bar is no longer a feature. It is infrastructure. And for most e-commerce brands, that infrastructure is overdue for an upgrade.