6 jun 2025
The Drop-Off Before Cart Abandonment
eCommerce
Cart abandonment is just a symptom. The real problem starts with broken product discovery.
Cart abandonment has been in the spotlight for years, and for good reason: In 2025, rates hit a staggering 70.2%, the highest since 2013.¹ But a growing body of research shows that many potential customers never even reach the cart. Instead, they drop off earlier in the journey, frustrated by dead-end searches, irrelevant results, and clunky navigation.
To understand what’s really costing revenue, we need to look upstream, at moments of failed discovery and missed engagement.
The hidden costs of poor search
Online shoppers often behave like in-store browsers. They may not arrive with a specific purchase in mind, but they’re open to finding something they love. That’s where digital teams have a major opportunity to convert low-intent sessions into high-value outcomes—if product discovery works.
Consider these stats:
58.6% of shoppers are simply browsing without immediate purchase intent²
69% report irrelevant search results
80% bounce after a poor search experience³
If your team is focused solely on checkout optimization, that’s a late-stage fix. Most of the real drop-off is happening earlier.
And while $18 billion is lost to abandoned carts each year,⁴ more than $260 billion is potentially recoverable through better digital experiences.⁵ That’s a growth opportunity hiding in plain sight.
Why traditional eCommerce search is failing
Most eCommerce platforms still rely on keyword-based search. These systems assume users know what they want and how to phrase it. But that’s rarely true, especially for shoppers in discovery mode.
Keyword search was built for speed: help users find and buy quickly. But that misses the point. Shopping is often exploratory, even entertaining. Many users browse to compare or get inspired. If your search experience doesn’t support that, you’re missing revenue.
Add to that a few technical pitfalls:
Boost and bury rules may surface irrelevant products
Missing tags mean key items stay buried
No synonym handling (“joggers” won’t return “sweatpants”)
No support for intent-driven queries
No adaptation to stock, seasonality, or trends
And critically, personalization software and search are still separate systems. Most search engines still return identical results for everyone.
This is where AI-powered search stands apart. It doesn’t rely on keywords. It understands the shopper and helps them find products they didn’t even know they wanted. That’s where AOV and loyalty grow.
Intent over input: AI-powered discovery
AI Findr is a customizable AI-powered search engine designed to turn exploration into revenue. Unlike keyword systems, it understands user intent based on language, context, even tone, and actively guides the discovery process.
It starts with ingestion: AI Findr pulls in your product data, store policies, FAQs, and support content, enriches it with external sources, and structures it into a machine-readable format. This becomes the brain behind both search and business logic.
And because AI Findr is an LLM itself, it’s uniquely suited to prepare your content not just for internal use, but for visibility in public tools like ChatGPT. (More on that shortly.)
Here’s what modern discovery looks like:
A user types “best camera for travel”
AI Findr clarifies: “Do you prefer DSLR or compact?”
Business rules apply—promotions get priority
It adjusts for current inventory, seasonality, and even bundles
The experience adapts not just to what was typed, but to what was meant.
In a year when teams are under budget pressure, AI Findr consolidates functions. It replaces separate personalization layers with a single system that starts working the moment someone types or speaks a query.
What happens when product discovery works
Improved search experience pays off quickly.
In 30-day pilots with AI Findr:
Checkout completions increased by 15%+
Fully self-serve sales rose by 52.6%
Customer satisfaction (CSAT) doubled
Net Promoter Score (NPS) increased by 39 points
Beyond KPIs, it gives digital teams more control. It makes your catalog more discoverable, your content more useful, and your search strategy directly tied to revenue.
LLMO and the new discovery frontier
More often, product discovery doesn’t begin on your site. It begins with ChatGPT, Perplexity, or Claude. The emerging field of Large Language Model Optimization (LLMO) focuses on getting your brand cited by structuring content for AI-generated answers.
Your structured data now needs to do double duty: power great UX and feed LLMs.
Because AI Findr already structures your content into a centralized knowledge base, its sister product, LLM Findr, can repurpose that data and optimize it for ingestion by external LLMs.
Your content becomes findable across platforms without doing misguided LLMO busywork like creating site FAQs.
AI Findr enhances internal discovery. LLM Findr helps ensure your brand is cited externally, by the machines shaping the new buyer journey.
Final thoughts
Cart abandonment isn’t where the problem starts. It’s just where you finally notice it.
The real drop-off happens when discovery fails, when intent is misunderstood, when browsing becomes a dead end. AI Findr rebuilds the experience around how shoppers actually explore.
And with LLM Findr extending that architecture to public AI chat platforms, the full search stack—on-site and off—is finally ready for how discovery works now.
Watch this demo from BIMBA Y LOLA to see what AI Findr looks like in action >
_____________________________________________________________________________________________________
Footnotes:
1. Statista
2. Baymard Institute
3. 2023 Censuswide research for Nosto
4. Forrester
5. Baymard Institute