I would say this is of course situational, but I would expect lower conversion rates and possibly quality score also. However if you really want to look at Adwords traffic to your main shopping page, looking at content network may be an idea…

Mr Burke and I had a meeting of minds – Unless Rob/Laureen object, this level of granularity hasn’t been disclosed to us. Google focuses on ECTR, but they do not specify whether this comes from the exact match term in each keyword, or the variations.

Reason being – It is unclear that if the actual variations from these keywords are held accountable for the exact match version of CTR in determining quality score. So, if variations had exact match CTRs this would still increase overall QS without the exact match of that term being bid upon or included.

However, in the example of 1 campaigns, 1 ad groups and 1 keyword being widgets (broad), and there is a negative exact match [widgets] there still has to be quality score assigned. In this case we can assume that the variations are accountable for being a part of the overall QS.

It makes our brains hurt… J

Ben: “Uncow Glop (what he calls me) … what’s the square root of negative infinity?”
Glenn: “I don’t know Ben, what IS the square root of negative infinity?”
Ben: “Slaps his head with his hand and says ‘it makes my brain hurt!’”

Say you’re selling widgets and you have the keyword “widgets” in broad or phrase match in order to pick up the many variations of things that people can type in (including misspellings). But, you also know that the exact search term [widgets] isn’t cost effective, so you make that an exact negative (-[widgets]).

How does Google then calculate quality score for that broad or phrase match keyword, if that precise term is never searched on? Does Google just ignore CTR as a factor in QS for that keyword? Does Google use the average CTR for that keyword across other advertisers? Anyone have any ideas?

Thanks for your reply.

When I made the post, suggesting to only bid on broad match kws… I did this partly as a devil’s advocate; exercise, but I do admit to having seriously thought of implementing this strategy.

What I have implemented though is a the strategy of separating keywords into tier 1 versus tier2; adgroups, tier1 being those that have shown a positive ROI from the alltime keyword value report. I find this a very helpful aid in guiding max cpa bidding while using conversion otpimizer.

The conversation behind the keyword roofing broad match and [roofing] exact match are drastically different, for instance, and respond better to different ads. In those cases the differences in the CTR of the exact match version alone would improve your keyword level QS, and put you at a competitive advantage.

Often you want to bid on exact and broad match differently because they hold different values to you, because the clicks yield different conversion rates.

We’ve found that impression share goes up for exact match versus phrase or broad because you’re considered speaking more directly to the precise search query, i.e. you’re considered more relevant. When you’re making money on a term, you want to be in that ad auction as often as possible.

Also, Google reports QS on the keyword level, but keeps it on many levels, including a QS for ad/keyword pairings. Often that level of QS is different for broad versus exact versions of a keyword because broad traffic performs so differently. If you had different ads winning for exact and broad match, this again would be affected.

All of these issues involve greater control (of bids, of ads, of message, of impression share, of ad position) and greater potential relevance which is the key to making money in the AdWords auction.

Thanks for your reply. I’ve reread your response and Brad’s article again.

Increasing quality score on broad match is possible by increasing the probability (CTR) that the search query will exactly match the broad keyword.

Ok, now that we understand that confusing issue… why wouldn’t we just add the exact keyword to the adroup?

One reason NOT to include the exact match version… the ad auction on the exact match version would be much more competitive than the broad match version with all our negatives in place, leading to a higher cost/click, yes?

That being said, does this make a strong case, NEVER to bid on broad match, and if so, then adding hundreds of negative kws would be an invaluable exercise in winning the keyword auction?

I suggest you use the Adwords Editor interface instead. It is much easier.

Jayne