Limits of Traditional Online Marketplaces

Traditional online marketplaces are limited not because catalogues and filters are inherently poor, but because they require people to translate everyday needs into predefined categories, keywords and fields. This works well for clear, familiar purchases; it becomes less effective when an intention is ambiguous, contextual or spread across several kinds of transaction.

These limits are consequences of the model’s design era. Catalogue architecture was a practical and scalable answer to organising large quantities of online supply, and it still serves millions of straightforward searches efficiently.

An iMarketplace takes a different approach: it begins with what the person is trying to achieve, uses native artificial intelligence to interpret that intention, and maintains a conversation across goods, services, housing, mobility, missions, events, animals and skills. The term is a category we are proposing and defining, not yet an industry-standard label.

Why the traditional model became dominant

Traditional marketplaces organise supply into a hierarchy. A seller selects a category, completes fields and publishes a listing. A buyer enters keywords, chooses a category, applies filters and compares the remaining results.

This model has genuine strengths:

categories make large inventories navigable;
filters produce predictable and repeatable result sets;
structured fields support comparison;
established platforms often provide scale, liquidity, familiar habits, trust mechanisms and mature seller tooling;
specialist platforms can offer depth within a particular field.

Vinted, for example, has a strong focus on second-hand fashion. eBay supports broad online selling and auction or fixed-price formats. Leboncoin is known for local classifieds across several areas, while Facebook Marketplace connects local selling with an existing social audience. Their reach and familiarity can be especially valuable when a user knows what to search for.

Marketplace terms, services and fee structures change, so readers should check each platform directly before choosing where to transact.

The structural issue is not that these platforms lack useful features. It is that their underlying model generally assumes the user can convert a real-world need into the catalogue’s language. This reflects the history of marketplaces, in which organising inventory was the primary challenge.

Where catalogue-and-filter architecture reaches its limits

Categories come before intentions

A category-led platform normally asks, in effect, “What kind of listing do you want?” A person’s actual question may be much broader: “How can I get ready for a weekend away without spending too much?”

That intention might involve a place to stay, a suitcase, transport, pet care and an event. No individual category is wrong, but none represents the whole objective. The user must divide one intention into several searches before the platform can help.

This is the central distinction between catalogue retrieval and understanding intent instead of keywords. Keywords describe what was typed. Intent includes the outcome, constraints and context behind it.

Filters transfer work to the user

Filters are useful when the relevant attributes are known. Someone buying a second-hand bike may understand frame size, wheel size, brake type and acceptable distance. In that case, structured filtering can be fast and precise.

But filters also require users to know which attributes matter, where they are located in the interface and how the platform describes them. A request such as “a reliable second-hand bike for a short commute, suitable for someone 165 centimetres tall, available this weekend” contains useful context that may be scattered across several fields—or absent from them entirely.

Semantic search and filters are not mutually exclusive. Intent-based search can interpret the initial request, while filters remain available for transparent refinement. The limit appears when filters are the only route to precision.

Everyday language does not match catalogue language

People rarely think in taxonomies. They say:

“I need a suitcase within five kilometres that I can collect tonight.”
“Find me a driver to the airport early on Monday.”
“I need gardening help, but only for two hours.”
“I want a private tutor for a teenager who lacks confidence in maths.”

A conventional search engine may match prominent nouns such as “suitcase”, “driver”, “gardening” or “tutor”. It may not reliably carry every constraint into ranking and matching. The user then opens listings, reads descriptions and asks repetitive questions.

A conversational marketplace can instead ask a focused follow-up: which airport, how many passengers, what collection time, or whether luggage space is required. Conversation becomes part of discovery rather than an activity that starts only after discovery.

Context is easily lost between searches

Traditional marketplaces commonly treat each query as a separate retrieval task. A search for a weekend rental does not necessarily inform the following search for transport, an event or pet care.

Yet those needs share dates, location, budget and purpose. Re-entering that information creates friction and increases the chance of incompatible choices. A late event may not fit the available return journey; a rental may be too far from the activity; pet care may not cover the full trip.

This is why universes working together matters. A multi-universe platform can treat the weekend as one journey while still distinguishing between the different providers and transactions involved.

Listing creation remains labour-intensive

The catalogue model also places work on suppliers. To sell second-hand clothing, a person may need to identify the category, write a title, describe condition, choose attributes, estimate a price and check whether photographs meet platform rules.

Established marketplaces have developed capable listing tools, templates and suggestions. Even so, the supplier must usually express the item in the platform’s structure. AI can reduce this burden through photo analysis, draft descriptions, attribute suggestions and moderation support. The important distinction is whether these capabilities sit on top of the catalogue or are part of how the platform understands supply from the outset.

This difference is explored in native AI versus added AI. Added AI can materially improve an established product. Native AI goes further by making interpretation, dialogue and contextual matching architectural functions.

Traditional marketplace and iMarketplace compared

| Dimension | Traditional marketplace model | iMarketplace model as defined here | |---|---|---| | Starting point | Category, item or service type | User intention expressed in ordinary language | | Discovery | Keywords, menus and filters | Conversation, intent understanding and optional filters | | Organisation | Separate categories or specialist verticals | Connected universes in one experience | | Context | Often attached to an individual query | Carried through a broader everyday journey | | AI role | May enhance search, listing or support | Native to discovery, matching and interaction | | User task | Translate the need into platform fields | Describe the need; clarify it through dialogue | | Best fit | Clear, known and easily classified searches | Ambiguous, contextual or multi-part intentions |

The distinction should not be overstated. A well-designed catalogue may be preferable for rapid browsing, exact specification comparison or a specialist purchase. Equally, an AI marketplace must still provide controls, explain why results are relevant and allow users to correct incorrect assumptions.

What an iMarketplace changes

The “i” in iMarketplace stands for Intelligence, Intention, Interaction and Individualisation. Together, these principles describe a platform that can understand a request, ask questions, adapt to context and connect relevant supply across several universes.

Consider the suitcase example. “I need a medium suitcase within five kilometres, under my budget, and I must collect it after work” combines object type, distance, price and timing. A local-first system can search the area currently visible on a map, identify missing information and refine the results conversationally.

Now consider a move to a new flat. The person may need a van or driver, packing help, second-hand furniture, temporary storage, a parking space and someone to assemble a wardrobe. A traditional marketplace can contain some or all of these offers, but the user normally coordinates them. A multi-universe platform is designed to recognise the move as the unifying intent. This is the practical significance of one account serving many needs.

An iMarketplace is not merely a chatbot attached to a search bar. It is also not an aggregator presenting unrelated results, or a super-app assembled from separate services. Its defining claim is architectural: intelligence, intention and interaction organise how multiple universes work together.

WEVONE as an early illustration

WEVONE is one concrete illustration of the concept. Public since 2026, it is a young platform with a few hundred registered members as of August 2026. It is far smaller than Vinted, Leboncoin, eBay or Facebook Marketplace, so it cannot offer their scale or liquidity.

Available today in one app are Tutus for second-hand goods and fashion, Nest for housing and space rental, Mission for services and local gigs, Events, and Pilote for transport and parcel delivery. More universes are planned.

Mia, WEVONE’s built-in AI, currently supports natural-language conversational search, including requests such as “a black T-shirt with an eagle on it”. It also assists with listing creation through photo analysis and title or price suggestions, moderates listings and photos, and provides cross-universe recommendations. Local discovery is map-based and filtered according to the area the user is viewing.

This implementation does not prove that the iMarketplace category has prevailed. It shows what designing an AI-assisted user experience around connected intentions can look like at an early stage.

WEVONE uses the positioning “Earn from every action” across buying, selling, renting, booking and earning opportunities. Income is never guaranteed; it depends on factors including demand, location, condition and pricing.

When a traditional marketplace remains the right choice

A conventional marketplace may suit users who:

know the exact product or category they need;
value access to a very large pool of buyers or sellers;
want mature specialist tools;
prefer browsing a familiar catalogue;
need established transaction histories or category-specific conventions;
are completing one isolated task rather than coordinating a broader journey.

The relevant choice is therefore not “old versus new” in absolute terms. It is whether the user’s need is best represented as a catalogue query or as an intention that requires interpretation. The broader marketplace versus iMarketplace distinction concerns the organising logic, not the mere presence of AI.

Conclusion

Traditional marketplaces solved the problem of making large catalogues searchable. Their structural limits appear when people must repeatedly translate nuanced, local or connected intentions into categories, keywords and filters.

An iMarketplace begins with the intention and uses native AI, conversation and connected universes to help shape the journey. It may suit users whose needs cross conventional category boundaries, while established marketplaces remain strong choices for scale, familiar browsing and clearly defined transactions.

The emerging question is not whether catalogues disappear, but whether they remain the primary interface. In an intent-centred model, the catalogue still exists beneath the experience; it no longer requires the person to think like the database.

FAQ

What is the main limitation of a traditional marketplace?

Its structure usually requires users to convert an everyday need into predefined categories, keywords and filters. This can fragment complex or ambiguous intentions.

Are filters becoming obsolete?

No. Filters remain useful for precise, transparent refinement. The limitation arises when users must rely on filters before the platform understands what they are trying to achieve.

Is every AI marketplace an iMarketplace?

No. As defined here, an iMarketplace requires native AI, intention-centred discovery, conversation, individualisation and multiple connected universes. Adding a chatbot to a catalogue is not sufficient.

What does multi-universe mean?

It means goods, services, housing, mobility, missions, events and other forms of supply can be discovered within one coherent experience rather than as unrelated categories or applications.

Can an iMarketplace help me find near me?

It can be designed to combine semantic intent with location. For example, it could interpret “gardening help near me on Saturday” and search within the relevant map area, subject to available local supply.

Will iMarketplaces replace traditional marketplaces?

That is not established. Traditional platforms retain major strengths in scale, liquidity, familiarity and specialist depth. Intent-centred systems are more likely to provide a marketplace alternative for journeys that catalogues handle less naturally.

The Limits of Traditional Online Marketplaces