August 27, 2007 at 01:54CTR of search results
As you know, search engine optimization (the main method of attracting new visitors to a site without advertising) has long become a multimillion-dollar industry. According to various estimates, the volume of the search engine optimization market varies from $ 50 million to $ 100 million, which is comparable with the volume of the media and contextual advertising markets on the Internet separately.
Surprisingly, with such market indicators, the SEO services themselves are absolutely not formalized. They are not so formalized that customers who, ultimately, are only interested in the volumes of attracted traffic, do not know in advance what they are buying. More precisely, they cannot a priori compare SEO services with Internet advertising (whether it be contextual advertising or a media card) according to familiar indicators, such as CPC, CTR, etc.
I tried to find at least some research on this subject. I can’t say that I tried hard :), but in the end, all the information on this subject comes down either to the dull Ashmanov shit (attempts to find the CTR of the search results regardless of the subject and type of request), or to empty-minded conversations of optimizers who are afraid to discover their “secret”.
Many at this moment will probably say that, but bullshit, SEO is a “semi-legal” service, which in some cases violates the rules of search engines, which increases the risk of the latter or some kind of anathema from the latter. And the less you shine, the calmer your life.
This is clear!
Another thing is surprising. Surprising is the position of customers who are ready to allocate solid budgets for SEO only because somewhere on the Internet they read that, they say, this is cool.
In short, it's all the lyrics. My real thought is that if it’s time to start full-fledged independent research, the result of which would be real data on the effectiveness of SEO? TNS Gallup will soon begin to monitor display ads on the Internet - perhaps search engines should confuse in response to measuring the positions of companies in search results ...
Unfortunately, there are no such studies so far, and we have to be content with a trifle.
Introductory
There is a site of the company Pervostrakh (the name has been changed for obvious reasons), which is promoted by a number of insurance topics. These requests are:
-insurance
- car insurance
- Casco
- CTP
- property insurance
- insurance companies
- pervostrah (company name)
- first fear (slightly modified name of the company)
- vtorostrah (name of the company, competing with Pervostrahom)
Promotion of and behind the monitoring was conducted during June 2007.
Statistics of transitions to the company’s website Pervostrach was calculated using one of the free attendance counters.
The purpose of this study- empirically, obtain data on the CTR of individual items of Yandex search results for these requests.
Assumptions in the research model
1. The ratio of impressions of the first search page strictly for the query in question to the total number of impressions (all pages of search results) for queries containing the query in question, for Yandex is equal to the same ratio for Rambler.
2. The distribution of requests on weekdays is uniform throughout the month, that is, it does not depend on the specific date and day of the week. A similar assumption has been made regarding the distribution of requests on weekends.
3. The ratio of the average number of requests on a weekday to the number of requests on a weekend is assumed to be 2. This number is calculated based on statistics on the total number of hits in Yandex search for June - the number of hits on weekdays is twice as many as on weekends.
4. The CTR data of a certain search results item is calculated based on data for at least three days. That is, if the site for some request stood at the position in question for less than three days, then the CTR for this case is not calculated due to a large error.
Settlement Procedure
1. Number of impressions per day.
Let:
• Yah - the number of impressions (all pages of search results) for the query in question in Yandex for June,
• Rx- the number of impressions (all pages of search results) for the query in question in Rambler for June,
• R1 - the number of impressions of the first page for the query in question in Rambler for June.
Then, the required number of impressions of the first page for the considered request in Yandex (that is, the page on which the Pervostrakh website was placed) is calculated, remembering assumption 1, as a proportion:
Я1 = Ях * R1 / Rx
For the “insurance” request for June we get:
Yah = 156'429 ( source )
Rx = 62'911 ( source )
R1 = 4'815 ( source )
We get Я1 = 11'973
Remembering that in June there are 30 days, of which 20 are weekdays and 10 days off, and also looking at assumption 3 (which is twice as many weekdays as weekends), we get the number of hits on our site for “insurance” per day.
On weekdays: 478
On weekends: 239
Similarly, we calculate the number of impressions per day for other requests.
2. Actually CTR (share of clicks per 1000 impressions)
For each day of June and each of the requests in question, we calculate the “local” CTR = ( clicks / impressions) * 100%
Next, for each of the requests in question, we write down what positions the search engine managed to stand on for a month. If a site has stood at a certain position for less than three days on a specific request, then we will not calculate for this case (see assumption 4).
For other cases, CTR is calculated as a sample average:
CTR = (CTR1 + CTR2 + ... + CTRn) / n,
where n is the number of days that the site stayed at this position for this request.
It is important that the calculation is carried out separately on weekdays and weekends.
Results
I calculated the requests for the company name for the whole month immediately without dividing by weekdays and weekends, because the site did not change positions for these requests. It turned out the following:
Conclusions
1.The statement that is obvious and without research - CTR depends on the relevance of the search results - is confirmed here very clearly. The spread of CTR is possible within all acceptable values, from 0% to 100%. If people search for the “first star” and get the first star website in the first place, then this is almost 100% CTR (on average less than a month, but almost 100% on weekdays). On the other hand, for the highly “vague” “insurance” request, the first place of the Pervostraha in the search results gives an order of magnitude lower results - about 5% CTR, which is comparable with the contextual advertising. Thus, a promotion for osago, for example, is more profitable than a promotion for insurance. Despite the fact that the latter has twice as many impressions, the first gives twice as many conversions due to four times more CTR.
2.CTR on requests of the subject of insurance on weekdays is approximately 30% higher than on weekends. It is clear that the main interested parties in insurance seek insurance information more actively on weekdays, using the Internet at work.
3. For “vague” requests such as “insurance” or “insurance companies”, the CTR difference between the first places is insignificant. In the case of “insurance companies”, on the contrary, there is an increase in CTR from 1st place to 3rd. For more specific requests, there is a fairly sharp decrease in CTR with a decrease in position.
4. Even on competitors, you can break your penny. 7th place at the request of "secondary" gives the First Straight 100-200 visitors per month.
PS If you know published similar studies, please throw links in the comments.
Source: Oddboy
Surprisingly, with such market indicators, the SEO services themselves are absolutely not formalized. They are not so formalized that customers who, ultimately, are only interested in the volumes of attracted traffic, do not know in advance what they are buying. More precisely, they cannot a priori compare SEO services with Internet advertising (whether it be contextual advertising or a media card) according to familiar indicators, such as CPC, CTR, etc.
I tried to find at least some research on this subject. I can’t say that I tried hard :), but in the end, all the information on this subject comes down either to the dull Ashmanov shit (attempts to find the CTR of the search results regardless of the subject and type of request), or to empty-minded conversations of optimizers who are afraid to discover their “secret”.
Many at this moment will probably say that, but bullshit, SEO is a “semi-legal” service, which in some cases violates the rules of search engines, which increases the risk of the latter or some kind of anathema from the latter. And the less you shine, the calmer your life.
This is clear!
Another thing is surprising. Surprising is the position of customers who are ready to allocate solid budgets for SEO only because somewhere on the Internet they read that, they say, this is cool.
In short, it's all the lyrics. My real thought is that if it’s time to start full-fledged independent research, the result of which would be real data on the effectiveness of SEO? TNS Gallup will soon begin to monitor display ads on the Internet - perhaps search engines should confuse in response to measuring the positions of companies in search results ...
Unfortunately, there are no such studies so far, and we have to be content with a trifle.
Introductory
There is a site of the company Pervostrakh (the name has been changed for obvious reasons), which is promoted by a number of insurance topics. These requests are:
-insurance
- car insurance
- Casco
- CTP
- property insurance
- insurance companies
- pervostrah (company name)
- first fear (slightly modified name of the company)
- vtorostrah (name of the company, competing with Pervostrahom)
Promotion of and behind the monitoring was conducted during June 2007.
Statistics of transitions to the company’s website Pervostrach was calculated using one of the free attendance counters.
The purpose of this study- empirically, obtain data on the CTR of individual items of Yandex search results for these requests.
Assumptions in the research model
1. The ratio of impressions of the first search page strictly for the query in question to the total number of impressions (all pages of search results) for queries containing the query in question, for Yandex is equal to the same ratio for Rambler.
2. The distribution of requests on weekdays is uniform throughout the month, that is, it does not depend on the specific date and day of the week. A similar assumption has been made regarding the distribution of requests on weekends.
3. The ratio of the average number of requests on a weekday to the number of requests on a weekend is assumed to be 2. This number is calculated based on statistics on the total number of hits in Yandex search for June - the number of hits on weekdays is twice as many as on weekends.
4. The CTR data of a certain search results item is calculated based on data for at least three days. That is, if the site for some request stood at the position in question for less than three days, then the CTR for this case is not calculated due to a large error.
Settlement Procedure
1. Number of impressions per day.
Let:
• Yah - the number of impressions (all pages of search results) for the query in question in Yandex for June,
• Rx- the number of impressions (all pages of search results) for the query in question in Rambler for June,
• R1 - the number of impressions of the first page for the query in question in Rambler for June.
Then, the required number of impressions of the first page for the considered request in Yandex (that is, the page on which the Pervostrakh website was placed) is calculated, remembering assumption 1, as a proportion:
Я1 = Ях * R1 / Rx
For the “insurance” request for June we get:
Yah = 156'429 ( source )
Rx = 62'911 ( source )
R1 = 4'815 ( source )
We get Я1 = 11'973
Remembering that in June there are 30 days, of which 20 are weekdays and 10 days off, and also looking at assumption 3 (which is twice as many weekdays as weekends), we get the number of hits on our site for “insurance” per day.
On weekdays: 478
On weekends: 239
Similarly, we calculate the number of impressions per day for other requests.
2. Actually CTR (share of clicks per 1000 impressions)
For each day of June and each of the requests in question, we calculate the “local” CTR = ( clicks / impressions) * 100%
Next, for each of the requests in question, we write down what positions the search engine managed to stand on for a month. If a site has stood at a certain position for less than three days on a specific request, then we will not calculate for this case (see assumption 4).
For other cases, CTR is calculated as a sample average:
CTR = (CTR1 + CTR2 + ... + CTRn) / n,
where n is the number of days that the site stayed at this position for this request.
It is important that the calculation is carried out separately on weekdays and weekends.
Results
I calculated the requests for the company name for the whole month immediately without dividing by weekdays and weekends, because the site did not change positions for these requests. It turned out the following:
Conclusions
1.The statement that is obvious and without research - CTR depends on the relevance of the search results - is confirmed here very clearly. The spread of CTR is possible within all acceptable values, from 0% to 100%. If people search for the “first star” and get the first star website in the first place, then this is almost 100% CTR (on average less than a month, but almost 100% on weekdays). On the other hand, for the highly “vague” “insurance” request, the first place of the Pervostraha in the search results gives an order of magnitude lower results - about 5% CTR, which is comparable with the contextual advertising. Thus, a promotion for osago, for example, is more profitable than a promotion for insurance. Despite the fact that the latter has twice as many impressions, the first gives twice as many conversions due to four times more CTR.
2.CTR on requests of the subject of insurance on weekdays is approximately 30% higher than on weekends. It is clear that the main interested parties in insurance seek insurance information more actively on weekdays, using the Internet at work.
3. For “vague” requests such as “insurance” or “insurance companies”, the CTR difference between the first places is insignificant. In the case of “insurance companies”, on the contrary, there is an increase in CTR from 1st place to 3rd. For more specific requests, there is a fairly sharp decrease in CTR with a decrease in position.
4. Even on competitors, you can break your penny. 7th place at the request of "secondary" gives the First Straight 100-200 visitors per month.
PS If you know published similar studies, please throw links in the comments.
Source: Oddboy