For two decades, the PLP (Product Listing Page) has been the engine of online retail. Paired with a standard companion, the filter panel, it became the industry-standard way to help shoppers find what they need. But the filter was designed where search was keyword-based, data was scarce, and AI was science fiction. That world no longer exists.
With natural language commerce, and machine customers, shoppers talk to online stores the way they talk to a knowledgeable friend. And the brands that make this shift first will not just win customers. They will fundamentally redefine what it means to shop online.
Filters Friction
According to the Baymard Institute, and many others, poor search and filtering experiences contribute to an average e-commerce cart abandonment rate of 70.19%, one of the highest across any digital channel. A further study by Forrester Research found that 68% of shoppers will abandon a site entirely if they cannot find what they are looking for quickly.
The filter paradigm forces customers to translate their intent which is human, fluid, and contextual into rigid taxonomic choices “Occasion: Formal” doesn’t capture “garden wedding.” “Category: Tops” doesn’t capture “something I can wear to a very formal event and still feel like myself.”
McKinsey’s 2024 State of Consumer Survey found that 71% of consumers now expect companies to deliver personalized interactions and express frustration when they don’t. Yet most e-commerce search experiences are the antithesis of personalisation.
The gap between consumer expectation and e-commerce reality has never been wider. And it is costing brands real money. Research from Salesforce estimates that poor product discovery costs retailers up to $300 billion in lost revenue annually in the US market alone.
$300B
Estimated annual revenue lost by US retailers due to poor product discovery (Salesforce, 2024)
Why Now? The Convergence That Changes Everything
The technology to replace the filter has existed in fragments for years. What’s changed is the convergence and the speed at which it’s becoming accessible.
Large Language Models (LLMs) like GPT 4o and Claude can now understand intent, nuance, context, and ambiguity in natural language at a level that would have been impossible two years ago. When a shopper types “something my mother would love for her 60th birthday, she’s very classic, budget around xx” – an LLM can parse that fully and return relevant results.
Vector search and semantic embeddings have matured dramatically. Tools like Algolia NeuralSearch, Typesense, Pinecone etc. allow retailers to index product catalogues in a way that captures meaning, not just keywords. A product described as “minimalist white trainers” will now surface for the query “clean, simple sneakers” without requiring exact keyword matches.
Real-time inference at scale is now commercially viable. The compute costs that once made AI-powered search prohibitively expensive for mid-market retailers have plummeted.
The result is a new category of commerce experience that we call natural language search
71%
of consumers expect personalised interactions — and express frustration when brands fail to deliver (McKinsey, 2024)
What Natural Language Search Actually Looks Like
Natural language search eliminates the PLP as the intermediary between intent and result. Instead of routing shoppers through a browse architecture built around internal product taxonomy, it takes their query however it’s phrased and returns a curated, ranked set of results that match the intent behind the words.
Here’s what that looks like across different retail verticals:
A shopper on a fashion platform types or says : “I need something for a client dinner that isn’t boring but still feels professional.” Natural language search understands occasion (business, evening), tone (creative, confident), and eliminates anything too casual or too formal. It surfaces a curated edit not 340 filtered results.
A shopper on a home goods site says: “I want my living room to feel more Scandi but I have a tight budget and a toddler.” The system interprets aesthetic preference (Scandinavian minimalism), practical constraints (child-safe, durable), and price sensitivity returning products that meet all three without a single dropdown being clicked.
A shopper on a beauty platform asks: “What’s a good moisturiser for someone who’s oily in summer but dry in winter, and is also vegan?” Filter panels typically cannot handle compound, conditional queries like this. Natural language search handles it natively.
“Natural language search doesn’t just improve conversion. It changes the entire emotional experience of shopping making it feel less like a database query and more like a conversation.” Imagine combining this with a framework that evaluates Price, Quality, Risk, Scope and Timeline against every product keeping the customer intent in mind.The User experience is profound and precious.
The shopper feels heard rather than processed. And in an era where emotional connection drives brand loyalty, this matters enormously. The buying and selling experience is now personalised according to the customers priorities. Overtime the algorithm learns to respond better and the PQRST framework adapts to reimagine buying decisions.
Beyond the UX story, the commercial impact of natural language search in combination with the PQRST framework is significant and measurable.
Conversion rate uplift: Early adopters of intent-based search are reporting search-to-purchase conversion improvements of 2x to 3.5x compared to traditional keyword search with filters. This is consistent with data from the e-commerce personalisation space, where Barilliance reports that personalised product recommendations drive up to 31% of e-commerce revenues.
Average order value: When shoppers are served contextually relevant results particularly when the system understands occasion, aesthetic, and intent they are more likely to purchase multiple items that cohere. Natural language search creates implicit cross-sell and upsell opportunities without the clunky “you might also like” widget. When we balance PQRST the buying and selling process is efficiently seamless and personalised
Reduced returns: A significant portion of returns globally stem from products that didn’t match what the shopper actually wanted
Customer lifetime value: Shoppers who find what they want quickly and easily come back.
The Organisations Leading This Shift
The early movers in natural language commerce are setting a new benchmark for what shoppers expect.
Amazon’s Rufus, launched in 2024, is a conversational AI shopping assistant that allows US customers to ask questions, compare products, and get personalised recommendations through natural dialogue. Within its first months, Amazon reported that it was already influencing millions of shopping decisions. The message to competitors was unambiguous.
Zalando, Europe’s largest online fashion platform, has been investing heavily in AI-powered search and styling tools that move beyond the filter paradigm. Their AI-driven outfit builder and semantic search capabilities are positioning them ahead of rivals who still rely on legacy filter architectures.
Google’s Search Generative Experience (SGE) is reshaping how shoppers discover products even before they reach a retailer’s site. This means that retailers who haven’t invested in structured, intent-friendly product data will lose visibility because Google’s AI can no longer surface their products effectively.
What This Means for Your Brand Right Now
If you’re a retailer, brand, or digital commerce leader reading this, the question isn’t whether natural language search will reshape your category. It will. The question is whether you will lead that change or follow it.
Here are three things to do immediately:
- Audit your current search performance. Audit your current search performance.
This data will tell you exactly how much revenue your current filter model is costing you. - Map your product data for semantic readiness.Map your product data for semantic readiness.
Natural language search requires rich, descriptive product data not just SKUs and categories. Occasion, aesthetic, material feel, lifestyle context, and emotional resonance all need to be embedded in your product attributes. This is a data exercise as much as a technology one. - Pilot a natural language search experience.Pilot a natural language search experience.
You don’t need to rearchitect your entire e-commerce platform to test this. A focused pilot on a specific category, with a defined user group can deliver evidence within weeks. Tools and platforms now exist that make this faster and more accessible than ever before.
The Filter Had a Good Run. Its Time Is Over.
The Product Listing Page and the filter panel served e-commerce well for twenty years. They were the right solution for a world of limited compute, keyword search, and early digital shoppers willing to do the work themselves. That world is gone.
Natural language search is not a feature upgrade. It is a paradigm shift from search-as-database-query to guided search-as-conversation. And the brands that make this transition now will not just see better conversion rates. They will build the kind of effortless, intuitive, personalised shopping experiences that create genuine brand love.
The filter is dead. The conversation has begun.
About Pathfinder by BOLDEST
Pathfinder is BOLDEST’s AI tool designed to eliminate the PLP and filter paradigm and replace it with an intelligent, conversational commerce experience through contextual answers responding to natural language queries with a framework called PQRST. Pathfinder embeds calls to action, reduces abandonment and vastly increases customer, supplier and workforce experiences. Built for mid-market to enterprise retailers, Pathfinder integrates with your existing product catalogue and delivers intent-driven search results that convert. To see a live demo or discuss how Pathfinder could work for your brand, contact us.
