Position (rank) bias
Items higher in a ranked list get disproportionately more attention and clicks, whatever their true relevance.
Three of the eight page areas.
Three ways to use it, and one to avoid.
- Treat the default sort as the biggest merchandising decision on the results page.
- Put best-match items in the top rows.
- Don’t read raw click data as proof of relevance.
- Avoid: presenting paid or margin-driven placement as organic relevance, which risks being a misleading practice.
Rank drove clicks, but not purchases once an item was clicked.
In randomised Expedia rankings, rank drove clicks but not purchase once an item was clicked.
Ursu's randomised Expedia rankings showed that rank caused more clicks but did not change purchase once a hotel was clicked. Joachims and colleagues found searchers favoured the top result even when the order was secretly swapped. Ghose and colleagues found ranking by expected value to the shopper raised revenue, but on a test site with 900 paid participants.
Five sources, each checked.
Last reviewed
Paid or margin-driven placement presented as organic relevance risks being a misleading practice; get it checked before using sponsored ranking.
This is simply a pointer to check, not legal advice. Seek legal advice before you put this principle into action.