Shared Listings and Seller-Level Attribution
Where several sellers compete on one listing, the rank belongs to the listing. This breaks competitive analysis in easily missed ways.
On marketplaces where multiple sellers offer the same item under one listing, rank measures the listing's total sales. Attributing it to a seller requires knowing which seller was making the sales, and that changes hourly.
The structure
One listing, one identifier, one rank.
Several sellers offering the same item, competing on price, fulfilment and rating.
One of them holds the featured position — the default add-to-cart seller — and captures the large majority of sales.
The featured position rotates based on price, availability, fulfilment method and seller performance, sometimes several times a day.
What this breaks
"Competitor X is selling well" is unsupported if X is one of six sellers on a listing and does not hold the featured position.
Rank tracks the listing, so a competitor losing the featured position sees their sales collapse while the rank stays healthy.
Your own sales can fall with no rank change, which is confusing until you know the mechanism.
Market share analysis by listing rank attributes all of a listing's volume to whoever you assume is selling, which is frequently wrong.
Capturing seller information
Record the featured seller with every observation. It is displayed on the listing.
Record the offer count where visible — how many sellers are competing.
Record the featured price, which is the price associated with that seller.
These three fields turn an ambiguous listing rank into an attributable one, at least approximately.
Sampling matters here. The featured position rotates, so a daily observation captures one draw. Hourly sampling gives a proportion of time held, which is the useful measure.
The share-of-featured-position measure
Proportion of observations where a given seller held the featured position, over a period.
Combined with listing rank, this approximates each seller's volume: listing volume multiplied by their share of the featured position.
It is rough. Sales are not uniform through the day and position rotation correlates with demand periods.
It is far better than assuming, and it is the only seller-level attribution available from outside.
Repricing dynamics
Automated repricing produces characteristic patterns: frequent small price changes, rapid response to competitors, and oscillation.
A listing with several automated repricers can spiral downward as each undercuts the others.
Observable as: price series with high-frequency small movements, and featured position rotating rapidly.
Analytically this matters because rank on such a listing reflects a price war rather than demand, and the volume is being sold at declining margin by whoever is currently losing the race.
When the manufacturer sells too
A brand selling directly on its own listing alongside resellers usually holds the featured position most of the time.
A brand that loses the featured position to a reseller has a channel control problem visible in the data.
Tracking featured-position share on your own listings is a genuinely useful operational measure that many brands do not collect.
For analysis, the practical rules
Never attribute listing rank to a seller without featured-position data.
Capture the featured seller and offer count if competitive analysis on shared listings matters to you.
Prefer marketplaces or categories with single-seller listings for clean competitive comparison, and note when your analysis crosses into shared-listing territory.
When reporting, say what you attributed and how. "Estimated from listing rank and observed featured-position share of 62 percent over 30 days" is defensible. A seller-level revenue figure with no mention of the listing structure is not.
Sampling the featured position
Featured-position share is only meaningful if the sampling captures the rotation.
Daily sampling gives one draw and is close to useless for this purpose.
Hourly sampling over a fortnight gives a usable proportion.
Weight by demand period if you can. Position held during a peak buying hour is worth more than the same position at 4am, and an unweighted proportion overstates a seller who holds the position overnight.
Record the featured price alongside, because the position and the price move together and the pair explains the rotation.
Note the offer count. A listing with two sellers behaves differently from one with fifteen.
Where hourly sampling is not feasible, state that the attribution is coarse and present it as a band rather than a figure.