Order from Lab F830DE

Retail search logs

Real retailer search logs: what the shopper typed, the ranked list the site showed, what was clicked, carted and bought, joined to the product catalog as it was on that date. Event-level, with the unclicked results kept, from several retailers in different verticals, licensed for commercial training with weights kept after the term.

How answering works

Any number of people fill one order. The lab receives one dataset, in its own columns.

Your agent reads the order
It looks at what you hold and says which columns you can fill, and what is missing.
Mora checks what you send
Every record: no copies, no personal data, not already public, not on Mora from someone else.
The lab pays per record it accepts
The lab named the price. You fill as much of the order as you can.

The columns the lab wants

You do not need every column. Mora says which ones each dataset fills.

Column Type What it must hold

search_id id One search event; joins searches, impressions and events

session_id id Pseudonymous session, so reformulations can be followed

searched_at timestamp When the search ran

query_text text The raw query as typed, typos included

filters_applied list Facets and sort order active for that search

ranker_version text Which production ranker or A/B arm served the list, and whether it was randomized

position number Rank of each product in the shown list, starting at 1

product_id id SKU shown; joins to the catalog

sponsored boolean Whether the slot was a paid placement

price_shown number Price at the time of the search

in_stock boolean Availability at the time of the search

event label click, add_to_cart or purchase, with event_at and quantity

product_title text Catalog title

product_description text Catalog description, brand and attributes

category_path text Catalog taxonomy path

valid_from / valid_to timestamp Catalog snapshot window, so price and existence can be checked against the search date

Column	Type	What it must hold
`search_id`	id	One search event; joins searches, impressions and events
`session_id`	id	Pseudonymous session, so reformulations can be followed
`searched_at`	timestamp	When the search ran
`query_text`	text	The raw query as typed, typos included
`filters_applied`	list	Facets and sort order active for that search
`ranker_version`	text	Which production ranker or A/B arm served the list, and whether it was randomized
`position`	number	Rank of each product in the shown list, starting at 1
`product_id`	id	SKU shown; joins to the catalog
`sponsored`	boolean	Whether the slot was a paid placement
`price_shown`	number	Price at the time of the search
`in_stock`	boolean	Availability at the time of the search
`event`	label	click, add_to_cart or purchase, with event_at and quantity
`product_title`	text	Catalog title
`product_description`	text	Catalog description, brand and attributes
`category_path`	text	Catalog taxonomy path
`valid_from / valid_to`	timestamp	Catalog snapshot window, so price and existence can be checked against the search date

How much, in the lab's words

At least 20 million search events (about 500 million impression rows) over 12 months without gaps, per retailer, from 3 or more retailers in different verticals, each with a catalog over 100,000 products. Delivered as Parquet in a bucket.

searches
impressions
events
catalog
experiments
judgments