|All else equal, impressions tank on one kw overnight|
QS, CTR, Position all the same, no seasonality at play, impressions floored
| 2:36 pm on May 24, 2011 (gmt 0)|
I have a campaign containing an exact match keyword ABC. Prior to Feb 17th, impressions ran about 1000 with a CTR of 30%, after the event impressions drop to 25. CTR remains the same, QS remains 10/10, average position *increased* from 1.4 to 1.3.
I've got data for the last two years on the same keyword/ad/campaign for the same period and there's no seasonality changes, particularly not ones that happen overnight on the 17th Feb.
The main ad that's being triggered in the campaign has a keyword insertion, though at the moment I'm not sure that's relevant as AdWords data is telling us the drop off came from the exact match k/w.
The ad kw remains in the doldrums to this day.
I'd really appreciate any ideas where to look next for some clues as to where those impressions went..
Thanks in advance.
| 5:29 pm on May 24, 2011 (gmt 0)|
-keyword bad (sometimes keywords get deemed adult, gambling, medical) and do not show up on Google.com
-check your negative matches
-another advertiser is bidding on the keyword using the same display url
-first page bid - if you fall to the next page and that page is not visited, your ranking doesn't change
| 5:29 pm on May 24, 2011 (gmt 0)|
I meant, keyword banned, not bad. :p
| 5:53 pm on May 24, 2011 (gmt 0)|
Thx for reply and pm, all sound advice but ... Keyword not banned, showing at position 1 page 1 etc. I'm going to investigate the negs though, handn't thought about that, perhaps one has gone funny.. The odd thing though is all the traffic was coming from exact match on the phrase - not sure how a neg could affect an exact match in part only and not cut out all the traffic.
| 6:54 pm on May 24, 2011 (gmt 0)|
No problem. I hope you findout soon.
have you checked your search querey report to see if another keyword is catching the impressions?
| 6:27 pm on May 29, 2011 (gmt 0)|
have you looked at a pre and post impression share report?
How can I see impression share metrics?
| 4:53 am on May 30, 2011 (gmt 0)|
Good point, confirms what we know.. that campaign dropped 20% impression share overnight while all else remained equal. Lost IS (rank) reports 68% post event, 64% prior to the event.
None of the figures appear to match up particularly well unless I'm misunderstanding - if Lost IS wasn't significantly different, then surely impression share shouldn't be 20% different?
| 10:44 pm on May 30, 2011 (gmt 0)|
you IS changes are nowhere near your 1000 -> 25 change...
either the traffic abated for some market reason...
amircat is right, another keyword is picking it up.
when keywords overlap (and they almost always do), a change made elsewhere can affect the other. perhaps you have a phrase or broad or mod broad whose performance has increased or whose bid was raised...
when you do as amircat suggests and try to "see" if another keyword has swept it up, the reporting can be confusing because of the nature of the search query report.
you might try adding your exact match keyword as an exact match negative to all other ad groups to see if that sorts out the match overlap.
but do keep in mind, within the ad group where it sits now, adding negs elsewhere won't clarify the overlapping within it's own ad group.
you might try raising its bid a bunch, to see if you can see the traffic is still there, and if so, restructure things going forward to make sure that exact match has no competing (aka overlapping) keywords that make encroach.
| 9:13 am on May 31, 2011 (gmt 0)|
Thanks. We're digging through that now. We're pretty confident it's not market - Google Insights and our own data for four years don't reflect it.
So we're back to keyword overlapping - history log shows no changes around those times, and the traffic didn't go anywhere else; so perhaps G changed the way it matches sometime around then that we weren't aware of.
| 2:46 pm on May 31, 2011 (gmt 0)|
keep in mind that there's a lag between edits and performance being recognized by G's algo, so you can't always tie the time of an edit to an overlapping keyword switch...
and, since G is constantly evaluating performance, there can be multiple switchbacks even for a fairly crisp usurpation.
if its critical, and you believe the traffic is unabated (insights is fuzzy), exact neg it everywhere else and have it stand alone in its own, new ad group.
| 3:02 pm on May 31, 2011 (gmt 0)|
I think that's going to be our next move, we're running out if options. Other anomalies are starting to show too, huge drops in CTR with positions, bids, impressions remaining equal in campaigns that have run for years that aren't prone to sudden change. One of our campaigns got switched to auto-max-clicks without anyone pushing the button (Gs investigating), so much sees to be tipsy turvy right now!
Thanks for all the input, if I find an answer that's postable I'll add it.
| 2:21 pm on Jun 1, 2011 (gmt 0)|
"One of our campaigns got switched to auto-max-clicks without anyone pushing the button (Gs investigating)"
nothing in change history?
no matter what you may find, prudent to have all users change their passwords. not saying this is the probable cause, pwords should get updated once in a while, and the time smells right.