New (possible) keyword tool, need feedback

by SlavaV
0 replies
Hello, Warriors,

I've created a keyword tool, and I have a few concerns about it, that probably best be addressed to Warriors.

So, here goes. This tool is my original creation, it doesn't use any other well-known tools (or any of keyword tools at all, actually), therefore what it produces isn't actually "keywords", but still it seems pretty useful. It's more of phrases that people use (and might possibly use in near future).

Ok, let's go with an example:

semkernel.com/google.adwords.500.html
(no ads there, no selling anything, just a demo of output)
(warning, file is 5MB!)

Don't mind the layout

The example is for "google adwords" query.
The first column is what I call query's semantic kernel. It might not be related to scientific semantic kernels, but it pretty close to what is there. Basically, it's 1-2-3-word phrases (uni-, bi- and trigrams) that are most related to the query. There are more than a million of such words, but only 2000 tops is shown (see next paragraph for why).

The rest of the table is columns. Column is "groupped semantic kernel". For example the word "adwords" is highly relevant to original query, so in " 'ADWORDS' GROUP" column all of the millions of found words that contain 'adwords' are shown. It's basically filtered first column, but whereas first column has only 2000 words, these other columns show filtered millions of words.

Inside the column there are few blocks:

WHITE background are the words that are in ACTUAL usage right now. It means that these WERE actually found to be used by people. These include often used things like "adwords account, adwords campaign or adwords campaigns" up to lesser used things like "google adwords ctr" or "adwords list building" (which is weird, but was found at least on some site).

REDdish (rose) background are the words that are possible, but haven't been yet found. Like "adwords search", which is a quite possible product and maybe even exists in some form, but since the system didn't see the words used - it is marked as RED ("Spin-off").

BLUEish background are more wild guesses. "Adwords people" are wild guess, but still not to far off.

LIGHTER BLUEish backgroung ("Maskeds") are the inferred keywords. It's a bit more complicated. It tries to find language patterns inside of this ("google adwords") group and guess new words based on those patterns.

Lastly GREENish block with links is the "Switch word (for synonymous)" - it replaces main keyword in whole column with synonyms. It can be pretty smart ("cpc"="cost per click" and wildly wrong "cpc = communist party of china"). It's because it's automated and uses two of my own algorithms for finding synonyms.

The system was developed because my friends needed more keywords than Google's system can provide, since google doesn't move too far from original keyword, like if you ask Google Keyword Tool (or other keyword tools) about "Adwords" it'll suggest a lot of actual phrases with "adwords", but not things like "ad copy, conversion rate, ppc management, conversion rate, click fraud, targeted traffic... etc" (check it out - I'm quoting the actual columns that my system generated).

My friends found it incredibly useful. On the other hand, being a perfectionist, one thing bugs me quite a lot. And that a lot of those things aren't "keywords", meaning that many (90% maybe) of those phrases are used by people writing the sites, but not the searchers. On the other hand - the results (tens of thousands of relevant keywords automatically) that you can get only from "google adwords" is astonishing.

Ok, so here's what I would like to ask the people of WarriorForum:

1. Is this useful or useless for professionals? (I'm a programmer, I don't deal with keywords on daily basis, so I can't judge the value of those, on the other hand many of Warriors are keyword pros, so I'd like to hear informed opinions)

2. This can't be released as desktop tool as it requires a lot of server power and traffic to process. It takes almost an hour to produce and many gigabytes and tens of thousands of sites analyzed. Therefore there's the second question. This list was quite costly to produce, so service that produces those can't be competing with all-you-can-eat for 50$/month services, like other keyword tools. The good news is that it isn't too costly to produce one list, so the question: what would you think to be a FAIR price for ONE such list? (I'll tell at a later point the approximate cost of it, but rather I'm interested in what seems to be a fair price) or is it a pipe-dream to charge per one query?

3. Maybe there's someone among Warriors who is interesting in leading this project as sales/going-forward guy for a share of profit he's able to generate? (I'm good with programming, terrible at sales/marketing). I'm open for discussion of that too.

The questions that might arise:
yes, it's fully automated, no human intervention whatsoever after entering "google adwords";
it takes about 30-60 minutes to produce one list, that could be parallelized more (to generate faster, but not with my current resources).

And if you have some suggestions/questions - let's hear them. I'm interested in open discussion of this (possible) tool with uber-professionals that Warriors are, because right now I'm pretty much on "destroy it and forget" course, like I did with my many-many-many other products (one of which was even featured on TechCrunch, still seemed utterly useless for me)

** P.S. Forgot one thing - it's can also be used as a "new product tool", since many of the things in red/blue blocks doesn't yet exist, but are viable.

Thank you in advance for any ideas/comments!
#feedback #keyword #tool

Trending Topics