When I was at BN around 7 years ago or so, we were experimenting with this capability on the site, and we had some processes like 1) rules around allowing no co

When I was at BN around 7 years ago or so, we were experimenting with this capability on the site, and we had some processes like 1) rules around allowing no competitive sales ads and 2) if the probability of purchase calced by a set of models was higher than the estimated ad revenue, we supressed ads. This worked really well, in that "browsers" who tended to visit and not buy on the site (and who, based on models combining digital and loyalty card data, were not expected to go to a store) might see ads (non-competitive, so no other booksellers, for example) and every visit, even non-purchasers, had some chance of revenue on the visit. Obviously, we knew that many visits are shopping visits on the journey to buy in a later visit or at store, so the models needed maturing, but the concept had legs. But as BN looked to add 3rd party sellers of the same products they offered in store and on site (ala amazon, ebay, etc.) and with the myriad leadership changes, clearly they went away from our original approach... And you wind up with this comment's experience.