Because I do affiliate arbitrage, and send traffic directly to the parent company's website, I have no visibility into the content network, and no easy way to track conversions sepearate for different sources. (i.e. no referrer log available, and tags to distinguish search from content would do me no good, since I have no access to the server logs.)
As a side-effect, I came up with a way to have seperate bid prices for Search Network. There is one caveat - it assumes that you want a lower bid price on Search Network. This is a pretty good bet, as so far the test is showing that prices on Search Network are lower than on Search.
1. Create a new campaign. If you have a campaign "widgets", create a new one, "widgets search network". Duplicate the adgroups you have in your widgets campaign. Put in a dummy ad (you will delete this later) and a dummy keyword, say, [delete this dkljfkldjflk] for each adgroup. Set the bid and budget to .01 for now. Once you have created the adgroups, pause the campaign.
2. You can repeat this for "widgets content network", if you are also using Content Network. This is really only useful for people like me who don't have access to server logs, and want to use this for conversion tracking purposes.
3. Use the new Copy tool to copy your ads and keywords from your widgets campaign to your new widgets search network and (optionally) widgets content network campaigns.
4. The Copy tool does not copy negative keywords or campaign negative keywords! You have to copy these manually. Do this.
5. Make sure you have your Campaign settings as you want them on the new campaigns. Budget, no optimized serving, etc.
6. If you want to track conversions for each network, you can use "edit CPCs/URLs" to ad some unique tracking code for each network. For example, I used unique affiliate tracking IDs. Fortunately, my affiliate programs allows me to have multiple tracking IDs.
IMPORTANT! Make sure you use "edit CPCs/URLs". Do NOT edit the ads in any way! If you edit the ads, they will require review before running. The "copy" tool gets around requiring review, as long as you do not alter the ads in any way.
7. Now go back and set your budgets and bids for the new campaigns. This will only work if you set a LOWER bid for search network than for search.
8. On your original campaign, set it for Google Search only. On your new "search network" campaign, set it for Google Search and Search Network. It isn't possible to set it for Search Network only. But, fortunately, we can trick the system to do just that. (Optional) set your "content network" campaign for content network only.
9. Delete the dummy ads and keywords that you created earlier. This was necessary because there's no way to create an empty campaign or empty adgroups. (Unless you use the API.)
10. Un-pause your new campaigns.
The original campaign will run on Search Network only, per the checkboxes. The "search network" campaign will run only on Search Network, even though you have selected Google Search and Search Network. This is because you have identical keywords in both campaigns, and lower bids in the new campaign. The Search Network exposures will all go to the campaign with the higher bids - your original campaign.
p.s. It is at this point unclear to me whether the "content network" only campaign will actually work. So far, I have no exposures. Google does allow you to select only Content Network. But can you actually do this? Aren't quality scores based only on Google Search? But if you've never enabled Google Search for a campaign... Catch 22?
The Search Network part of this seems to be working. I just set this up this afternoon, so at this point I have less than a half-day's data, and no conversion information until I get it from the affiliate site tomorrow morning. Thursday morning, to be really meaningful (full day working this way).
As an example of the difference between Search and Search Network, I am paying an average of .23 on Search and .07 on Search Network. 6.8% CTR on Search, 3.8% on Search Network. This is very preliminary, though, as it's a small sample. A few days of running this will give more meaningful data, including conversions.
Now, Google: give us a way to do this without so much fiddling around. Having more control is always good.
I may need to reevaluate my decision to turn off Search Network. Obviously Search Network is a bargain (even at 1/2 the CTR) if it can also deliver on conversions.
Why dont you redirect your destination url through a page on your own domain, that way you will get server stats.
Thanks - good idea. In fact, I thought of that as well, and it's in the works.
I'm waiting for a virtual server slot to become available later this month. I've got several websites in the works in addition to this, so the server will get good use.
In the mean time, this will give me some clue.
And I thought I would share the technique, as in addition to giving you the ability to tag Search vs. Search Network for affiliate programs, it also allows you to place seperate bids for Search and Search Network.
Finally, I have had some doubt as to whether or not Search Network CTR's are able to drag down Search position if they are significantly lower. This should solve that mystery. It appears to me that Search Network contributes to the ad CTR, and can bring down position in Search unnecessarily.
There shouldn't be an issue of ad review, since using Copy is supposed to bypass the need for review. I note that the copied ads in my "content network" campaign started running immediately.
I noticed something strange today that seemingly can't have anything to do with this experiment, but the notion that it might is nagging at me: my CTR in the "Search only" campaign shot up today to nearly 25%. And this is across two different adgroups in the campaign which are using completely different keywords. My average CPC also dropped, from .22 yesterday to .07 today. Search Network CPC is actually a bit higher than Search now, at .09, though not enough data there to be really meaningful. (There is enough data on the Search group, but not enough on the others.)
Most likely, I'm thinking that the higher CTR and lower CPC is due to my having "played chicken" with a competitor and won. The previous day (before starting this experiment) I had raised bids to a most unprofitable level - though I never actually paid anything close to that bid. I'm thinking that perhaps the competitor got burned and dropped out. The result is I now have a CPC that is 1/3 what it was before.
Anyway, if the CTR stabalizes, I will use the new CTR as the baseline to compare Search Network and Content to Search.
It seems you have to do some goofy stuff to be successful with Adwords. These kind of games shouldn't be necessary.
I'm afraid that at this point, all my experiment is showing me is that Adwords is an impenetrable black box. I already knew that. :)
How do the search only ads develop a CTR? I thought only Google impressions were used for data.
Sorry if I was unclear. By "search only", I mean Google Search.
I have three campaigns:
1. Google Search Only
2. Search Network Only (really, Google Search + Search Network, but with a lower bid than campaign 1)
3. Content Network Only
I am seeing CTRs for ads and keywords in both 1 and 2. (No clicks yet in Content, so I do not know yet about content. But I think I will see only total CTR there, not per-keyword.)
It's not true that Adwords does not maintain CTR for Search Network. It's just that it doesn't use CTR from Search Network in determining your quality score. It only uses the CTR from Google Search.
This points out another use for this technique of isolating Google Search and Search Network. If you do not seperate them you are blind to the CTRs that are actually being used by Adwords to determine quality score, since you will see a blended CTR for Google Search + Search Network, but Adwords is, in fact, using the CTRs only from Google Search!
My understanding is that keywords are in some sense "sticky" across your account. So, campaigns 2 and 3, at least in theory, are getting quality scores based on the CTRs in campaign 1, even though they report their own CTRs seperately.
Edit: - oops
I target the same keyword at different countries at different prices. Usually one CPC doesn't work for everything.
Even if one CPC is higher than the rest or one quality score outranks anothers, it is only designated to show in one spot. Therefor, your ads don't compete because they aren't trying to show in the same place.
Do all these campaigns have the same keywords?
Yes.
If I understand you correctly, you are saying that the keywords in campaign 1 will always show on Google's site instead of keywords from campaign 2 because the max CPC in campaign 2 are lower. I thought ads were shown based on quality score of which max CPC is only a part of. Isn't it possible keywords in campaign 2 get a better quality score and thus show up on Google even though their max CPC is lower?
See below on keywords. Position is based on quality score x bid price. In any case, the CTR on campaign 2 is lower than in campaign 1. Since everything else (keywords, ads, negative keywords) is the same, presumably #1 is getting Google Search only and #2 Search Network only.
It is looking like the CTR in campaign #2 is consistently about 1/2 of the CTR in campaign #1. I will have to run about a week to get good conversion stats.
My thought was that the keywords in account 2 and 3 might have a hard time getting a decent quality score. I am not sure they migrate across the enire account to help you out.
I assume you mean "campaign 2 and 3", not "account 2 and 3".
A Google rep told me that the keyword stats do migrate across an entire account. In fact, even if you delete a keyword and then add it back you still have your stats. They are both sticky and permanent, even if you delete them.
Google stats that the stats from Google Search are the only ones used for quality score. If the keyword stats didn't migrate across the account, what stats would they use for the same keywords in a content-only campaign? I do have a content-only campaign (#3) running with the same keywords. the one-day delay for exposures is curious, and I have no idea what it means.
Remember YMMV. Don't blame me if you rely on this and have poor results. Use this as a basis for your own experiments.
Just my luck - the online store who's affiliate program I am a member of ran out of the product I was using for this test. The ETA of the next shipment is 1-3 months.
It's a best-seller book, so 1-3 months might not be accurate, so I will wait a few days and see if I can re-start the test then. They ran out just before Christmas, also, but got stock back in a few days.
I don't have any meaningful conversion data, as the conversions plummeted yesterday when they ran out.
Here's the CTR data, anyway, for the two full days that I ran the test:
Google Search 40 clicks, 371 impressions, 10.7% CTR .10 CPC
Search Network 12 clicks, 262 impressions, 4.5% CTR, .10 CPC
Content Network 0 clicks, 1094 impressions
I note that I received one content click at the merchant's site, even though Google reports 0. Google must have filtered it out as invalid/fraud.
