When broad matching it appears as if the overall minimum CTR will no longer be relevant. Instead, if you bid on widgets and only sell red widgets, you will probably find blue widgets getting automatically disabled but your ad still showing for red widgets.
What I would like to know is how will they determine the relevancy - will it be based on the text in your advert or will it be based on CTR or something else?
future improvements will include other keyword matching options
Any thoughts about what they will be?
I found the mail to and the verbage could not be more vague (and that from a company that preaches relevancy). I also wonder how they automatically going to determine relevancy...
We'll see, hopefully this will not be as buggy as the other recent "improvements".
ps. the demo works for me but it is a basic tutorial on matching options.
Now all they need to do is optimize on (exact phrase, location, conversion, time) ..
That is, only show an ad on a particular site at a particular time if historically it is shown to convert well under those circumstances.
Cause let's face it. Searchers care about getting results that they will actually use, not just blurbs that are attractive to click on for a particular query.
For some ads, I have vaulted to #1 on some searches, back to the regular place on others. Google had to make this change in my opinion. Broad match is totally worthless for a searcher.
jd
Now, G will keep more 'grannular' track of those particular EMB's and ONLY switch off those which are not performing well.
This is a good improvement. I would still like to *know* when EBM is 'off' for a term but this is an improvement.
Recent improvements to promote ad relevance within the AdWords system will apply for broad matched keywords for advertisers in all countries and all languages, for ads shown only on Google.co.uk.
Is this trial only impacting Google.co.uk at the moment?
In general I think this move is good news and nice to see them actaully telling us about it. The mind boggles at the beast of an algorithm that must be required.
AWA any chance of a bit of a run down on how relevancy is determined?
Got to supprt Nyet with request for knowing what your ads do not show for.
When they rolled out EBM it was easy to miss.
I would like more clarity on how the new logic works, but I suspect they don't want to give too much about the ALGO away.
besides, NOT knowing will give us plenty to talk about here
Can we confirm that exact match keywords will be shown before broad match keywords (all other things being equal)?
I can't see anywhere in the FAQ or email that says this, and it would be a big change in how Google operates. All I can see from what they have said is that they will only partially disable broad matched terms instead of disabling them across the board.
I'm just getting to WebmasterWorld, and wanted to respond to a few posts in this thread. So, here they are - more or less in the order in which they occurred. ;)
How am I taking it is exact match is now more important than broad match. This is EXACTLY how it should be. If this is the change, kudos for google for fixing this issue.
Johnnydequino, this is not really the case - although it may occur as a result of the quality improvements. Rather than favoring exact match as part of the algo, the system determines position in the usual way (Max CPC x CTR) for each and every broad keyword variation - and ranks accordingly.
Since the exact match is likely to have a better CTR (assuming a well targeted ad), then it may well appear to be 'favored'.
Now all they need to do is optimize on (exact phrase, location, conversion, time)...
blaze, the ads quality improvements will be ongoing, and will take into account more and more factors over time.
Cause let's face it. Searchers care about getting results that they will actually use, not just blurbs that are attractive to click on for a particular query.
Well said, blaze. This is exactly what the ads quality improvements are all about.
Now, G will keep more 'grannular' track of those particular EMB's and ONLY switch off those which are not performing well
That is an excellent summary, nyet.
This is a good improvement. I would still like to *know* when EBM is 'off' for a term but this is an improvement.
I'll pass this on, as usual, at the end of the week.
Is this trial only impacting Google.co.uk at the moment
Excellent question, running scared, and sorry for the confusion. The intent of the message you quote was to convey that this is operating - at present - on Google and Google 'properties' only - and not partner sites. I'll alert the writing team to this potential point of confusion.
AWA any chance of a bit of a run down on how relevancy is determined?
I would like more clarity on how the new logic works, but I suspect they don't want to give too much about the ALGO away.
The algo takes into account many, many factors, and the factors considered will increase over time. However, nyet is correct in that I can't really be specific. (In point of fact, the exact algo will not be known by anyone in AdWords support. We are aware of it only in an overall sense, in order, of course, to protect competitive information.)
Wow, long post! I'll check back during the day and see if there is anything else needing a response.
AWA
Is the system going to learn over time to take into consideration other (as yet) unknown factors?
I have always thought that a search engine could *learn* if it could 'keep its eye' on multiple searches by one user. "watching" and considering each search string and 'remembering' those word associations and relating it to the eventual page the user visits.
The only problem was how to watch? You could not rely on IP #'s because those can change. Could the cookies for conversion tracking be the key?
[edited by: nyet at 6:13 pm (utc) on June 30, 2004]
Rather than favoring exact match as part of the algo, the system determines position in the usual way (Max CPC x CTR) for each and every broad keyword variation - and ranks accordingly.
Does this mean if I broadmatch: Green widget.
That Google is actually setting a different CTR (and thus CPC), for the broadmatch variation of the same keyword such as:
Green widget could have a 1.5%ctr while
green widgiting has a 2.3%ctr
while they are actually just part of the Green Widget broadmatch?
I hope that example makes sense.
p.s. I think the CTR for the 'hidden' EBM terms is (was?) not done on a term by term basis but by groups like +1 word, +2 words etc. just a guess.
Today's announcement was a step in the right direction, but the true formula for PPC should be exact, phrase, broad. It makes sense. Searchers are happy. Google still makes money. Advertisers actually get on the first page for relevent keywords.
I don't understand googles logic sometimes. Oh well.
Thanks for your post AWA
jd
I'm almost certain that the info below is correct. However, I really hate to be only 'almost certain' - so I'll confirm and post again when I've done so. This may be tomorrow, or it may be later today depending on if I'm (feeing) lucky.
* The CTR shown in your stats is the aggregate total for the broad matched term
* Each variation has it's own CTR to determine yes-no in terms of when EBM will be switched off for that term (to quote nyet), and...
* Each variation has it's own CTR which determines position, when calculated with the Max CPC of the broad matched term.
But, hey, don't quote me till I'm 100% certain, okay? ;)
AWA
Any chance that the new improvements will include capability of adding unique title and description data to the powerposting upload sheet? This is currently not an option. This would certainly help make things more relevant.
Latimer, I'll certainly pass your suggestion along - although it sort falls under the umbrella of 'feature', as opposed to a 'system improvement'. I'll add it to the Advertiser Feedback report, though, this Friday. ;)
My total Google costs jumped about 60% yesterday over the monthly average. Can anyone else share their cost experiences?
Although it is tempting to evaluate recent changes to a program right away, dnathan, I'd really suggest letting the system sort of 'reach equilibrium' for a week or more before drawing conclusions. Especially 'actionable' conclusions.
The 60% increase you mention might be related to ads quality improvements, but equally, it might not be. It could be something as simple as your product or service receiving some media attention today, and enjoying a spike in interest as a result. (I've seen this happen countless times. Often involving Britney.)
AWA