I've set up an ad group with keywords like:
computer repair
computer repair service
computer repair consultant
The content match was on, and I wanted to get clicks from the relevant sites in the content network.
After settings things up, I left the account alone for a while. Once I returned to check on things, I saw that I got thousands of clicks from the content network. The actual niche is way too small to generate that kind of traffic, so I checked my server logs to see where the clicks were coming from.
I visited the pages on which my ad has been shown (judging by the referrer logs of my web server), and found out that those pages -- many different ones, not just one -- were related to things like:
auto repair
credit repair
home repair
For some reason, G picked the word repair as the main word in the adgroup, and started showing my ad on pages that deal with "repair" -- any kind of repair, nothing to do with computer repair.
Needles to say, the traffic was completely untargeted and irrelevant to my product.
Does G consider those to be invalid clicks?
Thanks.
This is because your keywords are broad-match.
Does matching apply to the content network? I didn't know that it did.
Maybe you can fix this by providing information in the Ad text that makes the nature of your product / service clear
The ad was really specific. I don't think you guys are getting the situation. Here are the stats for just a few hours of running that adgroup. This data is from the adwords report. I'm not including the hours with hundreds of thousands of impressions, just the "peaks":
Date / Hour / Impressions / Clicks / CTR
2/23/07 / 12:00 AM / 1,312,577 / 170 / 0.01%
2/23/07 / 1:00 AM / 3,793,600 / 497 / 0.01%
2/23/07 / 2:00 AM / 2,951,086 / 359 / 0.01%
2/23/07 / 3:00 AM / 1,495,059 / 215 / 0.01%
and a couple of final blows long after I'd paused the ad group:
2/23/07 / 12:00 PM / 2,147,176 / 266 / 0.01%
2/24/07 / 12:00 AM / 5,490,200 / 873 / 0.02%
Considering that the keywords were not something like sex or mp3, I don't see how one can get 2-5mil impressions per hour and the context still be in any way targeted or related. Especially, when the other ad groups in that same niche were getting maybe 500-1000 impressions per hour at that same time. When I say a niche, I really mean a niche.
And such low CTR only proves that the ad was completely irrelevant to pages it was on.
Or, you could spend an hour and identify 95% of the negative keywords you need to enter and benefit from the content network (typically) lower click costs and haul in the profits.
I don't see how I can cover "everything" related to repair but computer repair with negative keywords.
We had similar stats (impressions and clicks) for certain sites, mostly from big countries or on home pages of very popular sites (eg file sharing, games online).
Now, an "impression" can be very easily "created".
We did report an occurance of this, involving a gaming site, but Google would not recognize it as invalid.
Which countries are you running your campaigns for?
There must be a couple of sites that are generating most of these clicks, do you have stats by site/page?
There must be a couple of sites that are generating most of these clicks, do you have stats by site/page?
Yep. A lot of clicks came from one large site, but from different pages on that site. Neither the site itself not specific pages were in any way relevant to my niche.
I've been using adwords for a long time, and always used content network with great success, but I've never seen something like this happen.
I've been using adwords for a long time, and always used content network with great success, but I've never seen something like this happen.
Same here. We now have switched off the content network for most campaigns because of this. We won't switch it back on until we can actively select content network sites.
Recently I blew up my whole budget for a campaign within one hour and most (90%) of the cliks came from China!
It was a total waste.
I have now reduced the budget on this campaign and monitoring carefully.
However IMO, something is broken at G or they are again experimenting...