For about two months people have been seeing strange queries in their AdWords search terms. These queries look like regular queries except that there’s the word “false” included – at the beginning, at the end or somewhere in between. Segmenting search queries by network quickly revealed search partners as the source.
I did some digging and compared search terms and referrer logs to find out where exactly these terms where coming from – it’s eBay. It’s not the first time they’ve been doing things like that: They’ve been including the word “enabled” in search queries for a long time.
Naturally, I wanted to find out what false meant. The word “true” also appears in search queries, although it’s much rarer. Looking at the referring pages from false and true queries I noticed that many of them included the URL parameter “_nkw”, which carries the term someone searched on eBay. It looked like true or false could refer to whether a search had returned any items on eBay. However, this was a dead end.
I also looked at some of our pure eBay campaigns. Those campaigns only cover eBay category pages relevant to our clients (as described here) and true or false appear in roughly 25 % of reported search queries. In these cases, the terms usually appear more than once, mostly three times. Those search queries look like this:
false false original category query false
From those multiple occurences I concluded that there is no single meaning behind true and false. Apparently there are several things here that can be true or false.
I also looked at the numbers for queries containing true and false. It appears that traffic from those queries actually converts OK and generates an average ROI (better than average in some, worse in other accounts). Compared to the traffic we get from eBay category pages, traffic with true or false produced much better results.
By the Way…
Aside from true and false I also found some other search query artifacts that I hadn’t seen before:
-
lh_sitewidecondition_new
-
lh_sitewidecondition_used
-
lh_sellertype_private (extremely rare)
Those, too, appear in some queries. Although these terms are very rare, the meaning of lh_sitewidecondition_new and lh_sitewidecondition_used on eBay seems obvious to me: Someone has limited a search to new or used products.
Conclusion
The inclusion of parameters like true and false is probably a bug and makes for some strange looking queries. As advertisers we can turn this to our advantage and handle at least some of eBay’s traffic separately.
One way to deal with this is to use these terms as negative keywords and be done with it. If you know for a fact that eBay traffic is worthless for your business, this is a good idea (you might also want to exclude “enabled”).
However, for many advertisers this is still valuable traffic, especially for retailers. So as usual, check your own data and see how queries with true and false have worked out so far. Aside from that, you probably want to exclude the term lh_sitewidecondition_used.
Martin Roettgerding is the head of SEM at SEO/SEM agency Bloofusion Germany. You can find him on LinkedIn.
Nice sleuthing, Martin! I’ve been seeing a lot of these queries in my reporting as well and was wondering what was going on. Kind of makes you wish that you could segment by search partners, if you didn’t want to before.
hi martin, because I have to share every single post in your blog you would do me a favor an put a google+ button too …;-)
Very interestin article!
We’ve seen lot’s of those queries and i usually exclude that kind of traffic from our campaings as it didn’t deliver very valuable trafic…
However the idea of building separate Ebay-Campaigns seems tempting to me, as this could mean to actually bid separatly to at least one search partner.
The main problem right now is that the search partner trafic influences overall AdGroup performances and bidding on AdGroups or Keywords without separating the origin is hard business and often leads to wrong decisions.
How do you handle that issue?
Hey Thomas,
For the most part, we don’t address this issue below the campaign level. Covering eBay categories in separate campaigns seems to make sense since the traffic value is considerably lower and it’s easy to set up. But besides eBay and maybe a few other search partners who use query substitutes there isn’t much you can do and so we have to live with sub-optimal bidding if we want to include search partners.
Some people deal with this by duplicating campaigns, once with and once without search partners, using lower bids for the latter campaign. I don’t think this would work well, but we’ve never tried it…
Do you have an approach to deal with this?
Hey Martin,
we tried the duplicating, too.
It works pretty well as long as you don’t have the necesity to bid your campaigns… ๐
There it get’s difficult to always have in mind, once you down-bid a bad performing search-only also to bid down the search-partner adgroup/campaign.
If you don’t bid them parallely you lose the effect of seperating them by bid and campaign adjustments…
However i’m positive, that the new google beta of bidding campaigns on AFS down through your google account manager is a first step towards seperating search and AFS.
How do you handle the bidding issue?
I guess using some bid-management-tools you can easily connect the two campaigns and implement rules to automatically downbid the AFS-Campaign, when downbidding the regular one…
Hi Thomas,
This beta sounds interesting, but so far it wasn’t offered to us yet. Is this just about separate bids for Adsense for Search or for search partners in general?
I haven’t seen any reliable data on the performance of AFS, but I always figured that, among search partners, that’s the part that delivers value. Search partner traffic from static pages on eBay and other sites on the other hand seems to be less valuable.
On duplicate campaigns, that’s what I figured: You have to somehow connect the bids on those two or the system breaks down. With manual bidding on a few selected head terms I’d say this could work, but it doesn’t seem worth the effort. You would need enough traffic from search partners and the traffic value would have to be considerably lower than the one from Google. In my experience, even though search partner traffic is usually less valuable on average, in some cases you’d want to bid higher.
Hi Martin,
from my experience with some accounts, AFS delivers up to 40% of all conversions and i would hate to exclude the traffic, but i don’t like having little control about my bids neither….
I coulnd’t tell you whether the beta differs between AdSense and search partners in general, but i guess it woudn’t.
I’d suppose they don’t but i’ll ask and as soon as i get more information i’ll let you know!
Hi Thomas,
Awesome, thank you! (and tell them to make the beta available for everyone already ;))
Hi Martin,
confirmed, the bids are for search partners in general.
Why would you want to make a difference?
Greets,
Thomas
Ah okay, thanks!
Different bids for true searches and placements like category pages would’ve been nice because the traffic value seems higher with clicks from actual searches. But at the moment I’m willing to settle for separate search partner bids ๐
@Olaf: Done ๐
Nice article!
I guess if you wanted to stress-test ebay, you’d use an ebay-specific / ebay-only type keyword in a clone campaign, using it as a broad-match modifier and excluding it across other campaigns …?
+ebayspecifickeyword
Yes, modified broad match would do, but since ebay is usually easy to distinguish, phrase works well, too. I’ve outlined the process in some detail in this article: /2012/04/02/targeting-search-partners/