A Very Strange AdWords Test

[abbottsys]

We spend lots of time carefully crafting and optimizing our AdWords ad copy - the headline, and 2 lines of body copy. We tweek, we tune, we measure. Finally we get a truly great ad.

I have several such ads that have performed superbly for years, pulling in major business. But am I absolutely sure it's because of my great ad copy? How do I know?

It occured to me that this could be measured just like anything else. So I'm running a variation against my best ad, but with bogus line copy. The variation reads:

Headline
Headline
Headline

Where "Headline" is the real headline from by best ad, but lines 2 and 3 just repeat this headline - i.e. there is no ad copy written for lines 2 and 3.

If my ad writing skills for lines 2 and 3 are so good then this ad should perform terribly compared with my *real* ad that I spent ages optimizing. Is ad copy for lines 2 and 3 actually needed? I'll let you know

[RedFly]

Originally Posted by abbottsys

We spend lots of time carefully crafting and optimizing our AdWords ad copy - the headline, and 2 lines of body copy. We tweek, we tune, we measure. Finally we get a truly great ad.

I have several such ads that have performed superbly for years, pulling in major business. But am I absolutely sure it's because of my great ad copy? How do I know?

It occured to me that this could be measured just like anything else. So I'm running a variation against my best ad, but with bogus line copy. The variation reads:

Headline
Headline
Headline

Where "Headline" is the real headline from by best ad, but lines 2 and 3 just repeat this headline - i.e. there is no ad copy written for lines 2 and 3.

If my ad writing skills for lines 2 and 3 are so good then this ad should perform terribly compared with my *real* ad that I spent ages optimizing. Is ad copy for lines 2 and 3 actually needed? I'll let you know

Looking forward to the results. My faith in humanity depends on your new test ads performing terribly.

[glengara]

At least if they don't tank you'll be able to console yourself with your headline writing skills :-)

[abbottsys]

Originally Posted by RedFly

.... My faith in humanity depends on your new test ads performing terribly.

LOL! I know exactly what you mean. I'm keeping my fingers crossed for a total flop

[abbottsys]

Originally Posted by glengara

At least if they don't tank you'll be able to console yourself with your headline writing skills :-)

Good point! I was careful to use the same headline in both ads, and it's a headline that took me ages to perfect.

BTW: The status so far.. the "bogus" ad just started getting clicks. No conversions yet.

[AdWordsRep]

A Very Strange AdWords Test

abbottsys, when I saw the subject line I instantly knew it would be you that started the thread. And I mean that as a compliment.

I am just waiting for the book to be published:

"The Complete AdWords Tests of Abbottsys -- 101 Useful Tests by the King of AdWords Testing"

When you publish, I hope I can be one of those folks who writes a glowing blurb on the back cover.

AWR

[abbottsys]

Originally Posted by AdWordsRep

...
"The Complete AdWords Tests of Abbottsys -- 101 Useful Tests by the King of AdWords Testing"...
AWR

LOL! What a great idea! I guess my need to test comes from my background, many years ago, as an experimental particle physicist in Europe. The guys I worked with were brutal. I mean, you could drop an apple and tell them it fell due to gravity and they would say, "oh really, are you sure, we'll test that hypothesis". LOL!

Meanwhile, prelim results from my latest ad test are either great or very disturbing, depending on your point of view

More to come..

[glengara]

Sounds ominous for Daves' faith in humanity......

Can you give any info on operational area, typical viewer profile?

[abbottsys]

PRELIM RESULTS

1) The bogus ad is performing exactly the same as the real ad in terms of #clicks and CTR.

2) Still to early to quote conversions because volume is not yet statistically significant. But, *if* all I was measuring was clicks and CTR then I would conclude both ads are performing identically. But I'm not, conversions rule, so I'll wait for those numbers to come in.

3) I have a couple of theories for the above, but I'll hold off drawing conclusions until all numbers are in.

BTW: we're in the medical field, generating leads for a high ticket elective medical procedure (no, not plastic surgery.)

[abbottsys]

RESULTS

Good grief!

1) The bogus ad has matched the real ad in terms of clicks, CTR and conversions. In short, its performance is superb.

This is a surprise, and *may* have implications for all AdWords advertisers and Google. I need to ponder this a bit more. I'll post my analysis and thoughts tomorrow.

[searchengineman]

You need to replicate your results with another AD, before you jump to conclusions.

That must be one hell of a headline. To not be mauled by the "Quality Score Google Bear".

Sad thing is if any Google Bear Keepers, are watching this thread. Don't be surprise if your trick becomes another SPAM sandwich, when they close the hole.


<snip>
Courtesy of <snip>

SearchEngineMan.

[abbottsys]

CONCLUSIONS

Before I give my conclusions let me just summarize this test so folks don't need to scroll through the whole thread.

I ran 2 ads against each other. The first ad (the "real" ad) was one of my long term top performers with all ad copy carefully tuned:

headline
line 2
line 3
display url

The second ad (the "bogus" ad) ignored lines 2 and 3 by simply replicating the headline:

headline
headline
headline
display url

And the test result was that the bogus ad performed just as well as the real ad in terms of clicks, CTR and conversions. The ads are lead generators for a high ticket elective medical procedure (not plastic surgery.)

So, my conclusions:

1) Clearly all I can say at the moment is that this worked for me, and for this particular ad. It would be very interesting to see if others can replicate these results.

2) Notice that the bogus ad is not really bogus, since it uses a powerful and proven headline.

3) Still, if these results can be replicated more broadly they mean that the real power of AdWords text ads lies in the headline, and lines 2 and 3 are far less important, or possibly not needed at all.

4) If it turns out that 3) is true, then both advertisers and Google could benefit if Google displayed reduced text ads on Google.com and the Search Network. A "reduced ad" would simply be 2 lines, the headline and the display URL. How would everyone benefit from this? Easy, twice the number of ads would fit on the SRP, so many more advertisers would get first page exposure. And twice the number of ads would also mean more revenue for Google.

5) It's interesting to note that "reduced ads" are a common format for AdSense publishers who use them on the Content Network. So, Google must have performance metrics for these ads, at least on the Content Network.

6) I would be very interested to hear from anyone who does a similar test.

I do *not* regard this as a "trick", I regard it as a test that could justify running reduced (2-line) ads on google.com and the Search Network. At the very least Google could try this on a controlled basis to gather stats. If successful it could open up vast new distribution inventory for Google.

AWR - is this worth a mention in your Thursday dispatch to the team?

[Mel66]

Fascinating stuff, abbottsys. I'm wondering which of our clients would be willing to have a go at this type of test.....

[searchengineman]

I've been vetoed for this strategy, by our office.

It looks Spammy. Unless you're marketing Parrots or Politicians!

SearchEngineMan

[danielanaidu]

If ad formats are changed to just two lines (a headline and the URL) aren't we going to end up with several ads that look identical? Already right now, many ads have the same headline and are only differentiated by the 2nd and 3rd lines. What will happen if the 2nd and 3rd lines are eliminated?

Aren't we always hearing complaints that the search engines don't give us enough room in the ads to truly specify what visitors will see on our landing pages?

[abbottsys]

Originally Posted by danielanaidu

If ad formats are changed to just two lines (a headline and the URL) aren't we going to end up with several ads that look identical? Already right now, many ads have the same headline and are only differentiated by the 2nd and 3rd lines. What will happen if the 2nd and 3rd lines are eliminated?

Aren't we always hearing complaints that the search engines don't give us enough room in the ads to truly specify what visitors will see on our landing pages?

Very valid points. But, if visitors are not reading lines 2 and 3 then what's the difference. As regards ads all looking the same, yes, for those folks who just rely on dynamic keyword insertion and do not bother to craft a real benefits oriented headline.

My test, if anything, says you should *really* work on a custom headline. And don't forget, 2-line ads are by no means new, they are used all over the Content Network (actually, these are 1-line ads, the display URL is not used). And these 1-line ads work. I know because I have an AdSense account and test them.

[abbottsys]

UPDATE

Since I gave the results of this test (see the post below, "CONCLUSIONS") the bogus ad has now gone on to outperform the real ad. It is now significantly better. This turned out to be a very strange test indeed. What does it say about adwords ad writing

[abbottsys]

Originally Posted by abbottsys

UPDATE

Since I gave the results of this test (see the post below, "CONCLUSIONS") the bogus ad has now gone on to outperform the real ad. It is now significantly better. This turned out to be a very strange test indeed. What does it say about adwords ad writing

FINAL UPDATE
Today I noticed that the bogus ad is outperforming the real ad by 2 to 1. In terms of clicks, CTR and conversions. So the test is over and I'm placing the bogus ad into production. Except, I think I should stop calling it the "bogus" ad, and now call it the "best" ad

[Mel66]

Amazing. This is a great example of why everything should be tested - sometimes what we think is the very worst ad / landing page / offer / whatever will perform the exact opposite of what we expect.

[Pakinsto]

Hi abot, i´m gona check your idea, i´ll tell u how it goes