Algorithm FAQ
FAQ
Keyword Research with ClusterAi
What to do after you receive your groupings
Have more questions?
How does ClusterAi's keyword grouping algorithm work?
Example Keyword Groupings
Main Keywords & Variations
Each keyword variation has at least 3 results in common with the main keyword.
That means, there are 3 or more pages that rank for both keywords with a single page.
If 3 or more other websites can rank for both keywords with one page, you probably can rank for both keywords with one page too.
Main Keywords vs Main Keywords
Each main keyword has 2 or less results in common with any other main keyword
That means, there are 2 or less pages that rank for both keywords with a single page
If 2 or less other websites can't rank for both keywords with one page, you probably can't rank for both keywords with one page either.
What about keyword cannibalization?
The great thing about using data from Google to determine keyword rankings is you can prove there is no cannibalization.
Compare the results of any main keyword with any other main keyword.
You will find that these keywords have 2 or less results in common.
If there are 2 or less websites that rank for both keywords with one page, you probably can't rank for both keywords with one page either.
I have a suggestion for ClusterAi's grouping algorithm
The first thing you should know is that ClusterAi's algorithm is not based on theory.
ClusterAi's grouping algorithm has been battle proven over a period of three years, and has been a critical part of taking four projects from 0 to 100,000 organics/month in 13 months or less.
With one of those projects going from 0 to 270,000 in 14 months.
Have an idea on making it better?
Great.
Unfortunately evaluating an algorithm iteration is a long, fuzzy process.
Implement each proposed change
Wait approximately 12 months per change while we gather ranking data
Make a fuzzy comparison that might not be accurate due to other variables besides the keyword grouping algorithm
I have another question about ClusterAi's Grouping Algorithm....
Example
URL1 = {key1, key2, key3}, URL2 = {key1, key2, key3, key4}, URL3 = {key1, key2, key5}, URL4 = {key2, key3}
Determine, for each keyword, associated URLs
key1 is in {URL1, URL2, URL3}
key2 is in {URL1, URL2, URL3, URL4}
key3 is in {URL1, URL2, URL4}
key4 is in {URL2}
key5 is in {URL3}
Only keep keywords which have at least 3 URLs
key1 is in {URL1, URL2, URL3}
key2 is in {URL1, URL2, URL3, URL4}
key3 is in {URL1, URL2, URL4}
Optimization - if a single keyword alone does not match at least 3 URLs, you know that you will NEVER be able to use it to form a group of 2 keywords having 3 URLs in common
Generate all 2-combinations (pairs) among kept keywords
combo1 = {key1 + key2}
combo2 = {key1 + key3}
combo3 = {key2 + key3}
Determine for each such combination URLs in common
combo1 {key1 + key2} has in common {URL1, URL2, URL3}
combo2 {key1 + key3} has in common {URL1, URL2}
combo3 {key2 + key3} has in common {URL1, URL2}
Only keep candidate combos with at least 3 URLs in common
combo1 {key1 + key2} has in common {URL1, URL2, URL3}
Now, we come to a stopping point, because, let's say you end up with 2 candidate combos like this:
comboA = {key 10 + key 11} = {URL66, URL77, URL78}
comboB = {key 10 + key 12} = {URL66, URL90, URL91, URL92, URL93}
I have just one more question about ClusterAi's algorithm.....
It's secret. 