Another adCenter bug

We’ve found several bugs in Microsoft adCenter, usually related to their web interface. This is a keyword addition bug. When entering keywords, either through the online interface or CSV import, certain keyword combinations are ignored if other variations with conjunctions exist, such as those containing “for” or “on”. Example: If “butter for toast” has been imported already into an ad group, try adding “butter on toast” or “butter toast”. acCenter ignores these new keywords, without even so much as a warning message. This is problematic if an advertiser wants to use exact and phrase match precisely as they are defined. This may be quite important if the presence, absence or difference in conjunctions is meaningful.

Bookmark and Share

2 Responses to “Another adCenter bug”

  1. C0rk Says:

    This occurs due to an adCenter process known as ‘normalization’ where extraneous/stop words like ‘or’, ‘and’, and ‘to’ and special characters like “?”, “!”, and “#” are removed and/or ignored.

    In adCenter’s matching, the KW “bike-repair” will match the query “bike repair” and the query “bike-repair”, so there is no need to add all variants.

    adCenter is essentially preventing your from adding KWs that would otherwise be treated as duplicates in terms of their matching. However, this does not account for the lack of ‘reactive messaging’ which should be provided or the concern you raise about the value of being able to bid/link conjunctions differently. (Are you optimizing conjunctions differently?)

    It’s important to note that normalization does NOT account for plural forms, so be sure to add those.

    You can find more details about normalization by going to the adCenter help section and searching for “normalization”.

  2. Josh Says:

    These are good points, C0rk, although this contradicts the meaning of exact and phrase match. While in most cases the conjunctions / stop words will not have an appreciable impact, it would be good to be able to test this. There are some cases where stop words could make a significant difference in meaning. As a somewhat silly example, if I am interested in searching for or selling the book “The Cat in the Hat”, I would not want my query to be normalized to “cat hat,” a search for feline clothing. I suspect that most instances where this is a problem would be branded searches.