Endless emails! Get 80,000 Direct Email Leads A Month Guaranteed! Every lead will already be expecting your message. If you are promoting any sort of internet business, you cannot afford to miss this opportunity.
Click here!

Powered by:

Estimating the amount of traffic your site should receive will always be difficult because any prediction is by definition a "static snapshot" of a dynamic and fluid stream of searchers:

1. Traffic to any SE varies greatly hour by hour, day by day (Internetters like their week ends too!), week by week, month by month and year by year. Seasonal keywords related to "holidays", "travel", "gifts" or "christmas" will show even greater variance during certain periods.

2. Your ranking will vary in time according to your and your competitors' efforts to keep a high ranking.

3. The number of sites ranked under your keyword(s) will vary in time.

4. As SEs themselves battle it out for traffic, your traffic will be reflected by their standing with searchers.

5. What appears in your listing will also affect click thru rates. From now on we'll assume that the "display" of your listing is at least "as good" as that of sites surrounding yours.


Enough! Send me estimates of my web site ranking results!

In spite of these difficulties, a meaningful model of how traffic will be distributed for a given keyword, a given site ranking in a given Search Engine is not entirely impossible. Traffic prediction models are invariably based on Zipf's Law:

c_i = k i^-a (1)

Where c_i = the 24 h traffic (count) for a site ranked position i

k = a distribution constant

a = an exponent ranging from 1 to "small"

Zipf's Law was dicovered when studying the relationship between the occurrence (word count) of English words in the English language and their ranking. The ranking attributed was 1 for the word with the highest occurrence, 2 for the word with the second highest occurrence, etc. Plotting occurrence v. ranking, Zipf found a good fit for a=1. In plain English this means that the word with ranking=10 occurs 10 times less than the top ranking word, the word with ranking=100 occurs 100 times less than the top ranking word, etc.

The exponent a has recently been estimated from search engine logs (searches vs keywords) to be close to 0.2. Find an example of such a plot here. The graphs show that the top three rankings seem to lead a "life of their own", more about that later.

This is quite significant because in practice it means that position 10 on average will still receive about 63 % of position 1's traffic (not barely 10%, for a=1), and even position 100 would receive about 40% of position 1's traffic (not barely 1%, for a=1).

We feel the apparent discrepancy between Zipf's Law (a=1) and search engine logs (a about 0.2) reflects the fact that the ranking is not the only thing that appears as a search result (if that was the case then a would probably be 1 also for searches v keywords plots) but instead the searcher is presented with a small bit of information on the listed sites as well. This little bit of "advertizing" is the "great leveller" that makes the ranking less important than would seem at first sight.


Enough! Send me estimates of my web site ranking results!

Nonetheless, ranking has still a profound effect on the traffic you might expect, so here goes.

From (1) the meaning of k can be understood for i=1.
(1) then becomes: c₁ = k 1^-a

And because 1^-a=1:

c₁ = k

In other words, k is the traffic received by the top ranking (i=1) site listed under that keyword!

k be obtained by calculating the total traffic c received on the keyword as the sum of traffic received by all N sites listed under that keyword:

c = SUM(c_i) for i=1 to N
= SUM(k i^-a)
= k SUM(i^-a)

Or: k = c / SUM(i^-a)

It can be shown that the summation SUM(i^-a) from i=1 to N, can be replaced by an integration of i^-a from i=1 to N, i.e. INT(i^-a)di between i=1 and i=N:

k = c / INT(i^-a)di with integration to i between i=1 and i=N.

Solving this determined integration between the boundaries 1 and N yields a simple relationship between k, c, a and N:

k = f(c,a,N) (2)

If we abbreviate the Keyword Effectiveness Index (KEI) to K and because:

K = c² / N (3)

It is now possible to combine (2) and (3) to:

k = g(K,c,a,N) (4)

Where g represents a simple relationship.

Here is an example of traffic for some hypothetical search results for i=1, 10, 20, 30, 100 and 1,000 and N=10, 100 and 1,000. The traffic is presented as a percentage of the total count (24 h) c for the key word.

K(EI) and k for various c and N
for a=0.2