This is total speculation, and yet another long post, so not sure if people will want to participate or not, but found the topic intriguing (could be that Psyche major kicking in). Feel free to disagree, would like to see others opinions.
I'm sure I missed a factor or 3, so please add more factors that I've missed. Of course, every factor can't be accounted for, but the major ones can be.
In order, would you agree that the rate is determined by:
1. Ad Position - no matter how good your copy, this is going to be the major factor.
2. Relevant Keyword (I'm sure there will be disagreement that this should be #1, but I've just seen too many unrelated KWs in the top positions get decent CTRs).
3. Title.
4. Competiting ads near yours - makes your ad look better or worse (or every ad looks the same).
5. Branding - have they heard of/done business with you before?
6. Total number of ads
7. Description
8. URL
Is there a major factor I missed?
Now for numbers, the reason someone clicked on your ad, which is total specualtion.
1. Premuim Position: 5%
2. Position: 30%
3. Relevant keyword: 25%
4. Title: 15%
5. Branding: 10%
6. Description: 10%
7. URL: 5%
Other factors:
Competition: +/- 25%
Total ads: +/- 15%
So, if your ad was:
1. Premium Position, 2nd place: 2.5%
2. Number 2 position: 25%
3. Keyword very closely related: 20%
4. Very Good title: 10%
5. They've seen your banners: 3%
6. Average descrition: 4%
7. URL: Included /Folder relating to description: 2%.
Excellent competition: -20%
Full 8 ads on page: -15%.
Would 31.5% of people who would click on an ad, click on your ad?
This can be taken even further, as the recent study by iProspect [clickz.com] (I consider this data very speculative, but useful for this formula) determined that 27.7% of G searchers clicked on an ad.
So, 27.7% of 31.5% is 8.72%.
Would the above ad gather an 8.72 CTR rate?
Some day soon, eWhisper will be writing the definitive book on AdWords!
AWA
I've had ads in the top premium spot get 60% CTRs and others get 6%, both had the keyword in the title and a decent description.
The more the top-10 spots are influenced by anchor text and don't have very exact matches of the keyword in the title or description snippet, the more the ads get clicked, especially those with keywords in the title, regardless of position.
RvN
If you have more time available on this topic you may want to run a few actual ads against your prediction model and see how well it does....
Because all the ads are syndicated, I can't seperate out AOL from G in the ad CTRs. So I looked at the logs, and adjusted the .277 multiplier from iProspect to .4 as AOL users have a 50/50 natural vs PPC CTR.
I then ran the best preforming ads at .9 of the max points allowed for title/description, and all of them came within +/-1.5% of the actual CTR rate (These were all for ads that have over a 10% CTR). While this is quite a variation still when you're talking about 1.5% itself being a good CTR rate for some KWs.
I think I'll set up a few campaigns later to expressely test this - should be interesting.
FromRocky,
I belive that dynamic titles can be as much of a hinderance as a boon, so that aspect could be directly related to how good of a description you have.
Budgeting should not have any impact on CTR, that would be frequency. However, 'ad blindness' or seeing the same ad over and over could have an impact.
Skibum,
The quality of the top 10 serps is a good point. I've also seen ads that do very well when the top serps are excellent for those keywords. I think that using the .277 multiplier by iProspect (or some multiplier of how often ads are clicked as opposed to natural serps) does take this into account to some degree - however, you have a good point to dwell upon there. Instead of making it its own category, it could be added to the 'competition' factor as that would have a direct relationship between both ad and serp competition.
AWA,
I'm still waiting for that G job posting to write AdWord FAQs and eBooks - I just keep seeing ad manager & client service positions :)
Is there a major factor I missed?
Yes, and it's a huge one. The offer.
Another factor is the quality of the target term. A term closely associated with a shopping search will have a much higher CTR than one associated with an information or entertainment search.
Yes, and it's a huge one. The offer.
Wouldn't this be directly related to description as that is the offer in an ad - or do you mean some call to action?
Obviousally 'pre screening' as in putting in price to discourage some clicks would lead to a lower description % as while it might be very well written, its leading to less clicks.
Offer is something different from a call to action. Let's consider our favorite term here: widgets. There are many different offers associated widgets.
* widget books
* widget staffing
* widgets themselves
* widget repair
* widget consulting
* widget training
Differences between these offers are quite noticable if you have more than one offer.
Analysis paralysis there eWhisper, but some stuff to stir the grey matter.
The issues I see are this :
Relevancy is the #1 factor in our book. Using dynamic titles and not letting Google serve in favour of ad popularity (i.e. ads served in rotation 50/50), we found that dynamic titles with the same description are around 50% better in terms of clicks than static titles.
Quality of the creatives, yours and theirs and spamminess of the organic play a big factor, probably as close to the position.
Up top always wins, searchers are lazy, the first thing that remotely looks enticing wins the day. The sample size will give different results. If you are top 1 as opposed to 1st on the side, difference is massive.
A couple of things you omitted.
Persistency. Evidence we have suggests that there is some reward for being online 24/7 as opposed to dayparting or, reduction out of hours. We have campaigns that did well off the bet when they had the inherited CTR, they established credibility and CTR/conversion was great. Client then decided to do the hokey cokey with the campaign, in/out/in/out/shake it all about, and we're left with a fishy smell that won't go away, no matter how much we increase CPC to compensate.
Size of keyword list. Your overall CTR rate is governed by the collective effectiveness of your keywords. The more you have in a group the less relevant some of those keywords will be. Much better to siphon them into their own group/campaign.
This is the sort of thing to sit down with a few beers to discuss.......
Persistency. Evidence we have suggests that there is some reward for being online 24/7 as opposed to dayparting or, reduction out of hours
FromRocky, seems your suggestion was a good one. All my accounts run 24/7, so never had this type of data WD mentioned.
Size of keyword list. Your overall CTR rate is governed by the collective effectiveness of your keywords. The more you have in a group the less relevant some of those keywords will be. Much better to siphon them into their own group/campaign.
Agree completely. And that's why KW has the 2nd highest factor in the equation. The same KWs in a group will not have the same CTR, and your reason is a major part of that.
The formula would have to be for an ad per keyword. So it wouldn't reflect the ads CTR, but the ads CTR for that KW, or an aggregate collection of all the CTRs in that group.
This is the sort of thing to sit down with a few beers to discuss
Amen to that!
In my experience this is a key factor in overall succes. The problem is that propositions which I though were great did "nothing" and vise versa. A typical case where the marketer cannot be the target group...
I added under 'other factors', a +/- 15% based on the offer (thank you Cline for that suggestion). Anything with the word 'free' in it, gained a few points, anything that was 'pre-qualifing' visitors, lost a few points in most instances.
I didn't mess with the 'persistancy' issue, as I wanted everything running 24/7.
Of course, most of this was pure guessing as I was attempting to judge the quality of my own ads - and the average exposure people had seen to various ads before.
So, based off the above qualification, the numbers came in within a reasonable margin of error for most of the campaigns I ran (not saying the CTRs were what I predicted, the CTRs were in the order of highest to lowest of what I predicted to what actually occured). However, there were a issues that really stood out after analyzing things so closely.
The first is searcher intent towards targeted keywords.
This really stood out in a few places when plurals were used in both phrase and exact matching.
keywords substitued, so if you sell this - not necessarily data that pertains to your keywords
Its reasonable to assume if you sold 'Classic Mustangs', that you'd want to advertise on 'classic mustang' and 'classic mustangs'.
In reality, the majority of people searching for 'classic mustangs' are looking for pictures, information, and more general stuff. Some were buyers, and the keywords were still profitable, but the CTR was lower than for the keyword 'classic mustang'. Classic mustang had a much higher CTR & conversion rate.
When it comes to wholesale supplies, such as 'floral arrangements', then the plural keywords were more qualified than the non plurals. Both were profitable in phrase/exact match scenarios, but the plural made quite a large difference. This is where powerposting can be quite useful as you'd want a landing page that had the availablity of both - but the presentation can be quite different.
I'm putting some of the 'plural/singular' data together, as that entire issue could be its own article/thread
To make the origional formula work, one could consider the plural keywords much less targeted, and thus make it work, but I was running a test - and didn't want to change my data after the fact - rather figure out where my estimates and the CTRs were very different.
The second was SERPS.
Skibum suggested SERP quality would make a difference. I didn't know how to qualify this before I ran the test (looking at all the possible SERP results to rate them just didn't sound very exciting), so I didn't adjust the formula at all for it.
For several keywords that seemed to do a bit better than I'd thought - the top 3ish results were almost all affiliate type pages with links to other search results, or similiar type pages. I'd love to know how many people visited those pages, clicked the back button, and then choose an ad.
However, when the SERPS were excellent, I didn't notice that much of a decline in the CTRs. This goes towards the theory that its very worthwhile to run ads even if you're number one for a keyword.
What todo with dynamic titles?
Find Your {KeyWord:Classic Mustang}
Buy {KeyWord:Red Widgets} Today.
Dynamic titles were very hard to judge. As expected, when several people were using them, the static titles did better than the dynamic ones. When no one was using dynamic titles, they did great.
In both instances, adding one or two words to the dynamic title made a difference in both scenarios as if you were getting the signifigence of dynamic titles with the 'stand apart from the rest' quality of static titles.
The Keyword
The % it affects CTR is higher than anticipated.
I added some dynamic titles with specific offers to very vague terms just to see how the CTRs would do. My working theory was that if you added an untargeted keyword with a nice dynamic title and a well written ad, you'd get a good CTR rate for a high position. My theory was definately off on this one.
This isn't always the case. In some instances, the keywords generated some nice CTRs - in other cases, they were slowed.
A good keyword is much more relevant than a bad ad.
A bad ad with good keyword selection can get decent CTRs.
A bad keyword with good descriptions can totally fail.
No matter how the CTR is boiled down - the keywords (and negatives) are by far the most imporntant factor.
Thanks for this test eWhisper. When you say at the end "bad keyword" I assume you mean untargeted.
are by far the most imporntant factor.
What have you been doing lately? ;-]
All the Best,
C.K.
