Published: 07 June 2026 by Dan Williams
Series: Notes on eBay & Discovery
Observations on why modern marketplaces are appearing to optimise buyer journeys rather than individual search results.
Search Is Only The Beginning
When most sellers think about visibility, they naturally think about search.
The logic is simple.
A buyer searches.
A listing appears.
A sale happens.
For many years that was also how I thought about online marketplaces.
However, after spending time reading eBay’s engineering articles and observing buyer behaviour, I increasingly suspect that search is only one part of a much larger process.
The more I read, the more it appears that modern marketplaces are not simply trying to help buyers find items.
They are trying to help buyers complete successful shopping journeys.
That might sound like a subtle distinction.
I suspect it is not.
Shopping Is Different From Searching
One of the interesting eBay engineering articles I came across was written back in 2011.
In Click Modeling for eCommerce, eBay discussed how shopping behaviour differs from traditional web search behaviour.
The article made a simple observation.
A web search user may find one relevant result and stop.
A shopper often behaves differently.
They compare.
They browse.
They evaluate alternatives.
They look at good deals and bad deals.
Only then do they make a decision.
That feels entirely familiar.
Collectors do it.
General shoppers do it.
Most of us do it.
Finding an item is often only part of the process.
Choosing an item is the real objective.
The Successful Buyer Session
If there is one phrase that repeatedly comes to mind when reading eBay’s engineering articles, it is: successful buyer session.
That phrase rarely appears explicitly, although it seems to sit behind many of the decisions being described.
A successful buyer session may include:
- finding the right item
- comparing alternatives
- discovering related items
- building a basket
- purchasing multiple items
- returning later and completing a purchase
The important point is that success appears broader than simply serving a search result.
Search begins the journey.
It does not necessarily end it.
Why Recommendations Matter
This helps explain why recommendation systems appear throughout modern marketplaces.
Over the years eBay has discussed:
- co-purchases
- co-views
- related items
- complementary recommendations
Complementary Item Recommendations at eBay Scale article
These systems are not merely decorative additions to a page.
They appear designed to help buyers continue their journey.
A buyer looking at a listing may also be interested in:
- similar items
- complementary items
- alternative items
- additional items
The recommendation itself may not generate the sale.
Sometimes the value of a listing comes less from the sale itself and more from the role it plays in the wider buyer journey.
However, it may help move the buyer towards a successful outcome.
That distinction feels important.
Different Buyers Want Different Things
Another recurring theme in eBay’s public articles is personalization.
As early as 2011, eBay was discussing personalized search.
More recently, they introduced concepts such as price propensity.
eBay Makes Search More Efficient Through Personalization
The basic idea is straightforward.
Not all buyers behave the same way.
Some prefer auctions.
Some prefer fixed-price listings.
Some regularly purchase lower-value items.
Others repeatedly purchase specialist material at higher price points.
If buyers behave differently, showing every buyer exactly the same results may not always be optimal.
Again, the goal appears to be helping the buyer make a successful decision.
Not simply displaying a list of matching items.
Similarity Creates Discovery
In the previous article I discussed vector similarity and recommendation systems.
The technical details are less important than the underlying principle.
Modern marketplaces increasingly appear to think in terms of relationships.
Listings may become connected through:
- titles
- attributes
- images
- buyer behaviour
Those relationships then create opportunities for discovery.
A buyer enters through one listing.
They discover another.
Then another.
Sometimes the journey becomes more important than the starting point.
The recommendation system is not replacing search.
It is extending it.
What eBay Appears To Have Been Building
One thing I found fascinating while reading eBay’s engineering articles was how consistent the direction of travel appears to be.
The technologies changed.
The terminology changed.
The scale changed.
Yet the overall objective remained remarkably similar.
2006
Best Match becomes the default search experience.
Focus: relevance.
2011
Behavioural signals begin playing a larger role.
Clicks, bids and purchases influence ranking.
Personalization appears.
Using Behavioral Data to Improve Search
Personalized Search at eBay
2015
Peer groups and popularity modelling.
Items increasingly evaluated within context.
Peer Groups in Empirical Bayes
2018
Intent becomes more important.
Skipped listings become signals.
Titles are evaluated in relation to buyer intent.
Finding Desirable Items in eBay Search by a Deep Dive into Skipped Items
2019
Complementary recommendations.
Co-purchases.
Co-views.
Related inventory.
Complementary Item Recommendations at eBay Scale
2020
Personalized Best Match.
Price propensity.
Different buyers see different experiences.
eBay Makes Search More Efficient Through Personalization
2024
Embeddings.
Vector similarity.
Semantic recommendations.
eBay’s Blazingly Fast Billion-Scale Vector Similarity Engine
2025
Domain-specific language models trained on e-commerce data.
Scaling Large Language Models for e-Commerce
Looking back through these articles, what strikes me most is not what has changed, but how what has stayed the same.
The technologies and terminology have evolved.
The scale has become extraordinary.
And the details differ, but the objective appears to be remarkably consistent – helping buyers make better decisions.
Why This Matters To Sellers
One reason I find all of this interesting is that it changes how I think about my own listings visibility.
Traditional thinking often treats visibility as ranking position.
Modern marketplaces appear to think more broadly.
Visibility may include:
- search results
- recommendation modules
- related items
- discovery pathways
- personalization
A listing does not necessarily exist in isolation.
It exists within a larger ecosystem.
The better a marketplace becomes at connecting relevant buyers with relevant inventory, the less important a single ranking position may become.
Closing Thoughts
Search remains very important. Without search, buyers would struggle to find many of the things they are looking for.
However, the more I read and observe, the more I think search is only the beginning.
Modern marketplaces increasingly appear to focus on what happens after the search.
Discovery.
Comparison.
Recommendations.
Personalization.
Successful buyer sessions.
Viewed through that lens, many of the technologies discussed throughout this series begin to make more sense.
They are not separate systems.
They are different ways of helping buyers move from intent to decision.
And perhaps that has been the goal all along.
— Dan Williams
Continue Reading
Previous: Cassini, Best Match and the Rest of eBay Discovery
Next: Beyond SEO: Visibility, Discovery and Selling on eBay
Read More
This note forms part of Notes on eBay & Discovery: a collection of observations about visibility, recommendations and buyer behaviour on modern marketplaces.