Last week, the Rimm-Kaufman Group disclosed what is apparently an inadvertent bug in Google’s Broad Match algorithm. Last summer’s switch from minimum bid to first-page minimum bid wasn’t supposed to impact broad match logic at all, but it did in a big way. The result? Agencies and advertisers who pay attention to the numbers saw a dramatic and essentially unmanageable decline in the performance of broad matched keywords, forcing down bids to remain efficient, dropping traffic and sales, with damage to Google’s revenue as well.
We hope that Google will fix this bug for the benefit of all parties. While they’re at it, we’d like them to consider a few more controls that we think will also allow smart advertisers to play the game more aggressively.
First, we’d like to separate broad match from advanced match. These should be two different match type choices. Old Broad Match (“OBM”) fired ads only when all the words in the keyword phrase appeared in the user search, but they could be in any order and combined with any other words. This gave us the ability to capture the crazy word orderings that appear in human search behavior: “I need to spend $5000 on a ring, a big diamond, or my girlfriend will kill me” without having to worry that Google will serve our “Tennis Bracelet” ad.
Furthermore, within OBM, we’d like the ability to accept or reject geo-specific search traffic. As we study the user searches on something simple like “furniture” we find a huge chunk of the traffic contains geographic modifiers, eg “Orlando furniture,” “furniture stores near 22901,” etc. Crate and Barrel with its wide brick and mortar footprint, might benefit greatly from these matches (though special geo-targeted campaigns would be preferable). Overstock, and other online retailers might find that this class of traffic converts poorly, as the user likely wants to find a physical showroom nearby. We’d like to get the benefits of broad match without having to pay for traffic that we find converts poorly. Sure, you could conceivably control this now through copious lists of negatives, but a check box might make time consuming/expensive negative keyword build-outs unnecessary.
Advanced Match can be absolutely amazing. Your keyword is “Sony XBR-123,” the user searches for “XBR123blk” and Google makes the match. We’ve seen them pick up stale model numbers in a user search and match them to an ad for the new model correctly! Incredible and for many clients truly valuable. Picking up typos and misspellings that we didn’t anticipate can also benefit account performance.
Smart marketers would be more comfortable using Advanced Match if we had controls over other behaviors. For example: A yes or no on competitor’s trademarks would be useful. While negatives are easier to manage here, a check box would be helpful to catch misspellings and typos of those trademarked terms. Granted, trademarks are dicey, and Google may not want to make promises here given how rapidly trademarks can arise. They could however offer a good faith effort without making a guarantee and that would be fine all around.
With respect to the buggy behavior we described last week, advertisers would like to prevent self-competition as much as possible. I can’t conceive of a situation where an advertiser would benefit by having Google pick a different ad than the one the advertiser designated to be served for user search “Aquarium filters”. Moreover, the person searching for “Filters for a salt water aquarium” might be advanced matched to any number of keywords, but the user, the advertiser and Google would all benefit by making that choice as tight as possible a match.
Having all keywords in the account—paused or active—act as “exact match negatives” for every other keyword in the account would accomplish a great deal towards cleaning up Advanced Match mismatches. Giving preference to the ads that match the most words in the user search when no exact match exists will also provide more relevant and better results all around.
Would these changes benefit Google? Absolutely. More relevant ads with better landing pages encourages users to click on sponsored links. Serving the most relevant ads will allow advertisers to control the game, and compete more aggressively in targeted areas. This will raise CPCs because conversion rates improve, the advertisers make more money and can afford to spend more. The users are happy, the advertisers are happy, and Google is too.
One more idea: Imagine if advertisers were allowed to target based on the buying behavior of the user. Online only businesses would bid hard for the traffic of proven online buyers. Brick and Mortar businesses would bid hard for the traffic that doesn’t buy online. Win-win-win.
Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.