Site optimization under Search Engines




To do business successfully on the Internet, it is best to concentrate your efforts on attracting target visitors to your site , i.e. those visitors whose problems the service or product you have placed on that site is designed to solve.

How to distinguish target visitors from random visitors ?

One method is to consider the visitor "target" if as a result of visiting a site he transits to a special target page. This may be your product's registration page, order page, etc. If a visitor is not interested in your product he will hardly go to the order form. Such a visitor is considered random.

Another variant is to define a target visitor as one who downloads the target file. This variant is good for sites distributing software, where the target files for downloading are the software distribution kits.

To set up an analysis of target visitors in Download Analyzer, it is necessary to indicate the URL of the target page or download file in the Downloaded files field.

The technology of analysis and optimization for Search Engines described in the present article was originally refined on sites distributing software, where the main goal is to increase the quantity of downloads of software distribution kits. This article is about optimizing target traffic for software sites. However, the technology also works perfectly well for optimizing visits to target pages of practically any type of site. It can be applied by any webmaster for attracting target traffic from SEs.

Mathematical model

The purpose of optimization

The purpose of optimizing a software site under SE is not so much to increase the quantity of visitors to a site, but more to increase the quantity of downloads of the software distribution kits. Therefore, the target function for optimization, in the domain of search phrases, is the value of download traffic t(p) for each phrase in one time unit. The total download traffic is the integral sum of traffic for all phrases:

T = S t(p)

The value of download traffic on an individual search phrase depends on three functions in the domain of search phrases:

t = b g w

b (p)  -

the function of demand for a search phrase, i.e. the quantity of requests on a given phrase by SE users in one unit of time.

g (p)  -

relevance, the value defining the extent to which the solution offered on a site corresponds to the problem which the visitor is trying to formulate by searching in a SE on a selected phrase. In the case of a software site, where the goal function is the number of downloads in a unit of time, relevance can be defined as the ratio of number of downloads to number of visitors from SEs on the selected phrase.

w (p)  -

the availability of a site, defines how easily a site can be found as a result of a search on a selected phrase.

Availability of a site

It would be great if our site were to appear in the first position in the search results for any search phrase. But in real life, it is not always like that. We still need to inform potential customers that our solution to their problem exists.

The value of the availability function of a site is greater, the higher the position the site occupies in the search results. As well as the position number, the page number on which the site appears is also a factor. I.e the function of availability of a site depends irregularly on the position number in search results. Furthermore, the availability function depends also on the brief description that we see as a result of a search.

The value function

It is impossible to achieve complete optimization on all phrases at once, and it is not necessary to do so. In practice, it is more important to maximize the availability function for those phrases on which download traffic will be largest.

Let's define the value function a(p) as the value of download traffic for a given search phrase, provided that the availability function for this phrase has reached a maximum value equal to 100%. Then download traffic is defined as result of multiplying the value of this function by the value of site availability.

t = a w

In fact, the value of a phrase depends on external factors and cannot be affected by optimizing the site:

a = b g

You could say that the value function a(p) is a projection of target traffic of our site onto the global function of demand for all possible search phrases. To attract the maximum traffic from SE in the shortest period of time, it is necessary to start by optimizing the site for those phrases for which the value function has the greatest value.

Under-received traffic

In practice, a software site will already have partial optimization and some traffic from SE even if the webmaster hasn't especially worked on it. The site may even appear in the first few positions of the search results in a search on the most valuable phrases. However, there will definitely be a set of phrases for which the site lies beyond the bounds of availability to the SE user, and on which appreciable download traffic could be attracted.

If the greatest possible download traffic on a selected phrase equals its value function a(p), and the traffic which the site receives at the present moment is the result of multiplying the value function by availability a(p) w(p), then under-received traffic is the difference between these two values:

d = a (1 - w )

The total under-received traffic is the integral sum of the under-received traffic on each search phrase:

D = S d = S a (1 - w ) = S b g (1 - w)

Thus the task of optimizing a software site for Search Engines can be reduced to the task of minimizing D, the total under-received download traffic.

From theory to practice

Optimization cycles

It would be great to create a report in which all target search phrases are sorted by their value function. And even better if phrases could be sorted by the value of under-received download traffic. However, in practice it is not always possible to do a quantitative analysis of the download traffic immediately.

One reason is the absence of generally available reliable data on the popularity of search phrases. There is some data, which may be appropriate for an initial calculation, available from the following services: trial/ d/ searchinventory/ suggestion/ metasearch/ index.html

But the main reason is that the relevance function is unique for each site, and to estimate this function it would be necessary to accumulate statistics on downloads for all our target phrases. And to accumulate statistics, at least partial optimization for target phrases is necessary.

Is this a closed loop? No, not quite. There will be information on rare, poorly optimized phrases in your site's log files over a long period of time. You can achieve a higher level of optimization by using this information skilfully.

Actually, the process of optimization develops in a spiral: analysis -> optimization -> waiting for traffic -> accumulation of statistics -> analysis -> etc .. At each turn of the spiral, new traffic is attracted and new statistics are accumulated.

Rating of traffic

By analysing the log files of a site, you can define functions on the number of hits h(p), number of downloads t(p) , and relevance g(p), for all phrases on which statistics have accumulated. While the log files of a site do not provide information on functions of value a(p) and demand for phrases b(p), an exact view of these functions is not so necessary. In practice it is sufficient to have an approximate rating of the under-received traffic on each phrase.

According to statistics, only 23 % of SE users go as far as the second page of search results, and only 10 % turn to the third page. It follows that, if a site appears after the first page of search results on one of the target phrases, it definitely receives less traffic than if it were on the first page. To rate this under-received traffic, the number of downloads for the second page of results should be multiplied by 4, and for the third page by 9.

So, in Download Analyzer, to create a report on phrases for which there is under-received download traffic, you calculate the position of your site in the results of a Google search for each phrase, filter out phrases which result in your site appearing after the 10th position, and sort the report by the quantity of downloads.

Directions of optimization

If the initial data are taken from log files over a long period, say half year, the report may contain thousands of phrases with under-received traffic. From all these thousands of phrases it is necessary to select a few basic phrases for the primary directions of optimization.

The most appropriate phrases to serve as basic phrases are those which consist of two words, and, in rare cases, of one. For example, the phrase address book can occur in the report as a component of other phrases such as the following:

simple address book
easy address book
free download address book
phone and address book
electronic address book
address book database
how to use address book

To select a direction of optimization on the basic phrase address book, and to rate the under-received traffic on all phrases in this group, set the phrase filter in Download Analyzer at: address book.

If the value of under-received traffic on all phrases in the group (take the total number of downloads on the status line in Download Analyzer and multiply the minimum by 5) seems to you worthy of attention, then for optimization in this direction you should set up a special advertising page and submit it to the SE.

How to get to the Top of search results

When you have selected a number of directions for site optimization , you should create a special advertising page for each of them, crammed with those phrases which bring the greatest traffic for the chosen direction of optimization. Choose the 10-20 most valuable phrases, and use them as a basis for writing an article. In fact, our target visitors have decided everything for us. We just have to materialize their wishes in the contents of the advertising article. After reading this article, the visitor should have no doubt that our solution to his problem is the best!

To improve the position of our page in the search results, the 5-6 most valuable phrases should be placed in headers (<h1>, <h2>..) between fragments of the text. A bold font or highlighting can be used to pick out search phrases inside the text. It is possible also to insert phrases in picture legends. It is also very useful to create links to other advertising pages, including in the text of these links the most valuable phrases of the page to which the link leads.

But perhaps the most valuable field for catching visitors is the Title of the page. If you under-receive traffic on a phrase that was not in one of the Titles, you can be sure that if you put it in the Title of a newly created page your site will show up in one of the first positions of search results in Google and several other SE. The Title of the page should be a sentence, correct according to English grammar, but at the same time composed of several of the most valuable search phrases.

What do you do if even after optimization you still can't achieve one of the first positions in search results on very valuable phrases for which there is a lot of competition? In this case it helps to insert these phrases in other special pages. You have to convince the Search Engines that not only a single page, but the whole site has direct relevance to the topic of these phrases.

How it works

Generation of a report on under-received traffic

Optimization is a multi-stage process - it should be done regularly, at least once a month. I shall give as an example what I do to increase the quantity of downloads from visitors who come to the site from a SE.

In the program Download Analyzer I load the log files for one month. No, for a serious analysis it's better to take the log files for half a year. In my case, for half a year, the phrase report shows 10583 phrases, 43450 hits and 14989 downloads.

To understand on what phrases I under-receive traffic, I need to determine the position of the site in the results of a Google search, the most popular SE, for each phrase. I launch the communication unit and I wait. Naturally, this process takes time.

But the benefit is obvious and should enable positions for the majority of phrases to be loaded. As a minimum, I am interested in phrases on which at least one download took place. I set the minimum quantity of downloads filter at 1. I don't have 10 thousand phrases anymore, but only 3849. The random phrases have been sifted out - now let the communication unit work.

Now I have a report which shows in the fifth column the position of my site in the Google search results on each phrase. I save the report. Then I have to begin a search for those phrases on which further optimization of the site is most attractive. There's no particular reason to optimize for phrases for which the site falls on the first page of search results. But those phrases for which the site occurs on the second and third page of search results, and for which the quantity of downloads is greater than 0, are exactly the phrases on which I under-receive traffic.

I set the filter for position of site in Google at the value 11, and thus eliminate those phrases for which my site appears on the first page of search results. I get 1307 phrases. This is also a lot, but it is these phrases that I need for further analysis.

According to statistics, only 23% of users go to the second page of search results, and only 10% go as far as the third page. I do a little arithmetic and calculate that, as a minimum, 80 downloads per day are under-received. There is definitely something to work on.

Set of words for optimization

The next stage is to define the word set. For example, when I input the word "read" in the phrase filter field, I see on the status line below: Hits: 771, Downloads: 461, and that 184 phrases with this word did not get on the first page of search results. At the same time, if I remove the filter on position in Google, and set the phrase filter at "reading", I see a set of phrases with the word "reading" which have brought me a thousand downloads. And in most cases, my phrases that include the word "reading" are not only on the first page of search results, but also in the first position. This is because the word "reading" is in the Title of my special page and occurs several times in the page contents, whereas the word "read" is not in the Title, and the majority of phrases with this word put my site on the second or third page of search results. It turns out that if I take one of my special pages, substitute the word "read" for "reading" everywhere it occurs, and add it to the site as a new page, I get an additional 12.5 downloads a day!

But certainly it is better to create another page with correct English, and to take the opportunity to add some more words. In my case, I found 14 words on which there is significant under-received traffic. Now I have to make up various combinations of these words and rate which combinations give the greatest under-received traffic. For example, I input in the filter field: read + file (conjunction), I see 106 phrases and 296 downloads, i.e. about 8 under-received downloads per day.

Having rated the traffic on various words and their combinations, I make up an English sentence including the most valuable word combinations. Unfortunately, I can't put a lot of words into a page Title, and I may have to make another page and another Title to include phrases which can bring additional traffic. But for now I shall try to place the remaining word combinations in the page text.

Perfection knows no bounds

So now the page is ready, it's already on the site, and links from other pages lead to it. It's a miracle! In only two days Google has indexed the page. In the past it took two months. The result is another 60 downloads per day.

There is always something to work on. In a month I shall have new statistics, new phrases and new under-received traffic. Consequently, the optimization process can be continued.

Optimization of page contents

The secret of one anomaly

Download Analyzer is a very valuable tool for analysing not only search phrases, but also the page contents of a site.

Once, when I had created an additional special page for phrases on which I was under-receiving traffic, I discovered something surprising after Google had reindexed it. The relevance of the newly created page was almost twice as high as the average relevance of the site as a whole, 57% compared with 33%.

One might think that I had used super target phrases and as a result, the pages report showed such an anomaly. But it wasn't this. Several phrases on the new page overlapped with phrases on another of my pages, and the two pages competed with each other in a Google search. In the detailed report for these phrases both the old and the new special page were seen. It turned out that the relevance of the new page for the set of common phrases was twice as high as the relevance of the old page.

The secret of this anomaly was in the page contents. It was clear that a more structured text, one in which I had been generous with headers (tags: <h1>, <h2>.. ), and where the text was highlighted and supplemented with screenshots, was more successful in convincing the visitor to make use of the solution offered on the site.

The importance of the visitor's opinion

A very simple conclusion follows from this: the site should be optimized not for the Search Engines but for our beloved visitors. And you can be certain that they will value this.

If you managed to shoot to the top of search results using cunning spam methods, you have only one opportunity not to get lost amongst the leaders and to remain in the top - increase the relevance of your page to the maximum. This means making an attractive page design and ensuring that the page contents contain information that the target visitor is looking for, and will find interesting and convincing.

Modern techniques of ranking take into account the user's opinion. This means that the probability of shooting to the first positions in search results and, more importantly, remaining there, is significantly higher for pages containing a clear, precise article, corresponding maximally to the subject of the search phrase request.

In addition, this type of page has more chance of being mentioned in usenet and subject sites, and of being recommended to mailing lists subscribers.

If people have a problem, and you have the solution to it, your special advertising page should become the bridge which joins the problem to its solution!


Download Analyzer 1.5 / 547 kb /  

Copyright 2003 GEO Ltd
All Rights Reserved

Download Analyzer


Download Analyzer

Search Engine Optimization FAQ

Introduction to Relevant Analysis

Technology of site optimization

  Mathematical model

  Practice of

  How to get in top of
  search results

  How it works

  Optimization of
  page contents

Quick start

  How to start








  Calculate position
  in Google


  File sizes

Tips and Tricks

  Menu commands

  Keyboard Shortcuts

  Text filter Rules

Registration and licensing


  Limited warranty

  Technical support

  How to register