Big list vs Small List 6:21 pm on Mar 24, 2013 (gmt 0)
Can a VERY big list of well thought out negative keywords have a very positive impact on an account?
I read elsewhere of people using thousands of negative keywords to improve the performance of AdWords accounts and before I go off and do more research, I like to check in here first.
For instance, can a negative keyword have a positive impact even if it isn't showing up in the SQR (Search Query Report) or as it is now termed the MSQ (Matched Search Query through Analytics)?
We all know about the positive use of general negatives like job and resume and such (for non-job related sites) but will these type words have a positive impact even if they aren't actually driving those inadvertent clicks?
I think it would - because it would lower inappropriate impressions - which in theory will improve CTR and cost per conversion.
Agree or disagree? And why?
Agree, because QS is primarily based on CTR, getting rid of irrelevant imps, helps.
7:55 pm on Mar 24, 2013 (gmt 0)
others (on another channel) seem to disagree that negatives have minimal to no impact on QS...
10:17 pm on Mar 24, 2013 (gmt 0)
Well let's use our heads and figure this one out... lets take a precise example and there are plenty of them.
1. You have a search term "Portsmouth Real Estate". and you are in Virginia selling Portsmouth, Va real estate.
2. When someone in Portsmouth, NH searches for Portsmouth Real Estate and your ad comes up "Portsmouth, Va Homes for sale and another realtors ad comes up "Portsmouth, NH homes for sale"... whose ad will normally get clicked.... of course if the person is really looking for real estate in NH... the won't click the Va ad.... so we just got an impression that's very unlikely to get clicked.. every time that happens.. "impression and no click".. your QS suffers a little bit more... not only that, I don't want him to click my ad and make me pay a dollar since he won't buy from me anyway... so not only did I lose QS score points it cost me a freaking buck cause I did not use negative keywords effectively... I should have all states but my own in the negative box.. so that alone is 51 negative keywords right off the bat... Make sense?
10:30 pm on Mar 24, 2013 (gmt 0)
Thanks TommyTX, logically I totally agree that this type of approach will improve the performance, hands down.
...however in your particular case, I don't think negatives would prevent the search from showing as the searcher didn't enter a state name - more like geotargeting would help but that's a different subject.
Marketing wise, yes, using negative state names in that general case makes a lot of sense.
Where I'm going with this is the going consensus (that I'm trying to confirm or disconfirm) seems to be that QS applies only to exact match queries, so negatives that eliminate broad match impressions don't apply.
it simply comes down to CTR, as it appears you have already identified. I have run numerous experiments that shows data to backup the theory that improved CTR increases QS, and improved QS lowers CPC. Your original question was does a large negative list "...have a very positive impact on an account?"... According to my experiments I would say yes as I am defining "...a very positive impact..." as a lower cost with higher conversions. For instance... I have a DUI Attorney client located in LA, and when they came to me their average CPC for the LA County area was $46.00 a click... after I included a larger negative keyword list we saw an immediate reductions in CPC of $6.00 that first week. After continuing to make adjustments to their Ad Groups to make them more relevant and also restructering their ad variations we saw their QS increase 1-2 points for their main keywords and their CPC drop another $6.00 the following week as well.
hope this helps...
I've been seeing it too, for sure.
I've been picking off the bad ones for years and enjoying the benefits.
I always wondered if there was any data proving or disproving that pre-emptive negative keywords (those not found in detailed search query reports) also had a positive impact.
I'm trying to get some insight about the rifle-shot type of management where you are picking off negatives that you know vs the blindly eliminating tons of words that you don't see in query reports.
It is more complicated than we're all saying, of course.
For exact match, for example, negs shouldn't matter.
And the QS shown is for the exact match version of your other match types, so there's data you don't see - your QS for a non-exact match.
Eliminating searches you know don't convert well enough, makes sense whether if affects QS or not.
Rodger that RF!
Other reading suggests there are at least 2 QS type metrics - that which is shown (like toolbar PR) and that which is what is really going on (much like what we learn in advanced backlink analysis).
Fortunately, we have a reasonably decent view of what we're told are the actual queries, and that presently isn't being obscured as much as it once was.
Thinking this through (with all your help of course) has brought me to think up an advanced negative test case which I'll be running to see if I can boost ROI.
Heh I've worked on accounts where I've run up against the limit on negative keywords (at one point it was 5000, not sure if it's changed - I don't work on that account anymore)
Part of the problem was really really broad terms that could mean one of a dozen things, and part was because the client had let it their display network campaign run pretty much on autopilot for years and years and there was a lot of cleaning up to do.
Hi Netmeg - thanks for chiming in!
Where's eWhisper when you need him?
It seems there's two sides to this discussion - one is control and the other is QS.
As for control; negatives are amazing. We have many accounts with more than 100k negatives in them (and some are often 1 million - don't try this without the API). Google changed the negative keyword max from 5000 to virtually unlimited a while ago. These large negative lists are mostly for control - not for QS reasons.
From a control standpoint, you might use negatives to block ads from showing in one ad group so they show in another one. This is common when you have ad groups that are general (say website hosting); others that are discount based (cheap website hosting); and others that are specific (vps website hosting). By using your negatives to 'shape' which ad shows; your overall CTR and conversion rates should go up. In fact, if you don't use negatives in that situation, you will probably have some of the 'wrong' ads shown; and if someone is looking for a VPS, you never want to show them a general hosting ad. So using negatives for ad shaping based around what type of ad the end user should be seeing is essential in many accounts.
There are many ways you can control ad displays in this way; like some companies have absolute monthly budgets. So you might have an brand, exact, phrase match, and broad/modified broad campaigns. In this case, you'll spend all you can on your broad, exact, and phrase and use the broad to help you hit run rates. To make this work; you need all of your phrase match words as negatives in the broad campaign; and all your exact match words as negatives in the phrase campaigns.
For larger accounts, I'm a huge fan of the negative keyword lists as I often find that the negatives are not applied correctly to multiple campaigns. If you download your campaign negatives, and put them in a pivot table - look at how many negatives you have by campaign - seeing high level numbers can be enlightening.
From control to QS... So negative words won't directly affect individual word QSs as they are only calculated when the query matches the word. However, part of QS is overall QS for the account. What negatives will do is block these irrelevant queries so that you don't show for low CTR terms that don't matter for you, and then the entire account's CTR/QS will go up; which then indirectly helps a keyword's QS used in the auction. Its a bit convoluted, and since overall history isn't that huge of a deal for most accounts; the odds of you actually noticing negatives helping QS is pretty low.
So, I do find negatives are very important; but they are more essential from ensuring the correct ad shows for each query as opposed to trying to eek out slightly higher quality scores.
Looking at massive numbers of MSQ terms has been quite eye opening indeed. And this is on well managed (manual) accounts. We'll be seeing some nice performance improvements as a result of this deep dive I figured out how to do.
I also get to look at Marin data occasionally but I don't know the tool well enough - does Marin and other auto management tools automatically mine and apply negatives? Do you trust it?
Now if we could just get G to show us the kw's imps w/o any clicks... :-)
2:33 pm on Mar 28, 2013 (gmt 0)
@RhinoFish ... exactly! been wanting that for some time now as well!
2:38 pm on Mar 28, 2013 (gmt 0)
don't expect it anytime soon... but there are ways!
2:53 pm on Mar 28, 2013 (gmt 0)
@chewy ... I'm looking into my keyword details report now, and am seeing a few search terms with impressions but no clicks (never noticed this before), but is there another way that you are referring to? Thanks!
3:33 pm on Mar 28, 2013 (gmt 0)
Pretty much proprietary - some big houses claim they can see it - experience in looking at billions of words across various performance metrics is the thing that becomes valuable. Just keep looking.
8:00 pm on Mar 28, 2013 (gmt 0)
If your account is in an MCC; you can easily get 0 click data.
If you run the search term report from the MCC; there's still the old option to include '0 impression' data' which will give you all the queries with 0 clicks but have impressions.
If you have nothing to do for a couple of hours, read this for a mind boggling argument on what is and what is not true on quality score.
[rimmkaufman.com...]
Wow! That even made me weary about QS...But no matter what google says, I will never be convinced that Negative keyword used correctly will NOT indirectly raise the quality score of you entire campaign. IT WILL! Since we know for a fact that bad impressions will decrease, and CTR WILL increase... it has to since you are showing more ads that are more likely to get clicked... and if CTR goes up so will your QS.... its easily proven... Simply take any account with the QS averages pretty stable.. apply very careful and correct negative keywords.. a nd watch you QS overall go up... it will happen...
1:32 pm on Mar 29, 2013 (gmt 0)
Didn't have time to read EVERY word of the article, but I find it funny that that the article title says "NEGATIVE KEYWORDS DO NOT AFFECT GOOGLE ADWORDS QUALITY SCORES"...yet the final comment by "Google Adwords Help" says..."Matthew Mierzejewski has also written a fantastic post on this topic and detailed how negative keywords impact Quality Score."...hmmm? Two articles that contradict each other? There wasn't a link to the latter so I could read that article...just found it humorous.
Taking all this in, and reviewing the original question again from chewy, I'm still of the belief that even though the QS of broad match terms are based off of it's exact match we would still see a slight increase in QS, OVER TIME, as our CTR would get better as we continue to add negatives as this would, as chewy stated earlier, "lower inappropriate impressions".
eWhisper, thanks, I didn't know that was still there! Always learning!
When combing Placements, the external report makes it harder to take actions, but still, great tip!
I wish they'd have this available at the account level.
eWhisper, aren't I looking for "0 clicks", not "0 imps"?
(yep, question for myself, will run and see)
Hmmm... thanks eWhisper... just ran a report from here and also added a filter to remove all terms with clicks so I could simply see terms that have impressions with no clicks... I immediately saw 3 terms that have generated around 400 impressions each over the past 90 days that haven't generated a single click. 2 of them I can see need to have improved Ad Ranking, but one is a definite candidate for removal...Thanks!
@RhinoFish ...you were saying "I wish they'd have this available at the account level."... When I just ran this report it appears that they offer the ability to select "Account and Campaigns" as a radio button. After you select it it opens your entire MCC where you can navigate to the account and/or campaign that you need...
Seems to work even without MCC but with some limitations that seem to apply even when viewing via MCC. Super nice!
7:04 pm on Mar 31, 2013 (gmt 0)
"@RhinoFish ...you were saying "I wish they'd have this available at the account level."... When I just ran this report it appears that they offer the ability to select "Account and Campaigns" as a radio button. After you select it it opens your entire MCC where you can navigate to the account and/or campaign that you need..."
It lets you select for the report, correct, but it's an external report, meaning you can't click a result and convert it into a negative keyword, you're not in the account when viewing the report, you're in exported data.
Still a very great tip from eWhisper!
No comments:
Post a Comment