For the purpose of making a good keywording program which I and heavily involved in these days. I have had to do a lot of research on keywording and search patterns in our buyers. I have done this based on some back-end material that I cannot share in the public, but I CAN share my findings�:)
Use these results wisely.

General observations:

-   People are inconsistent and non-decisive about using plural and do so randomly. The highly ranked keyword �groups� illustrate this very well. People may often search for �girls, fun, lifestyle� and end up buying a picture with only one girl.

-   Search patterns are heavily dependent on stereotypical behaviour to a much more serious extend then first concluded. From a logical point of view it would have been good keywording to include the following keywords together: �multi-ethnic, multiethnic, mixed races, diversity, diverse, international, multi-national, multinational, multi, ethnic� but my research indicate that all these keywords combine to only one forth of the search rank of one single keyword: �interracial�. This means that leaving out essential keywords can damage the sales at a much higher level then I thought.

-   Search patterns are much more direct then common sense would dictate. People looking for a picture of three or more business people discussing can be found searching for �three, suits, business, table� and not �business people talking� or �businessmen interacting�. This means that concrete keywords like �faces, hands, table, sitting/standing� are much more important then secondary keywords. For example people will search for �face� when they are looking for a close-up and not search for �close-up�. 

-   People never use the �-� when searching. No search pattern occurrences of �close-up� or �multi-ethnic�.

-   Keywords longer then 8 characters account for less then 10% of the total search rank from all keywords.

-   The keywords: �splat, splatter, award, happy hour� are highly searched for but no one uses them.

-   The �adult� and �adults� are useless, and almost never lead to sales.
-   There is a 80% overlap between highly searched for keywords and highly used keywords by a photographer. This means that one can use the public resources available on istock, fotolia, ss and Dreamstime and do not need back-end material on search behaviour.

In other words. All the keywording I have done on my 8000 picture portfolio is wrong�

The keywording program I am trying to design will warn a user when a stereotypical keyword is left out and suggest other highly ranked keywords based on the keywords the user types in.
We have been into some really serious mathematics and some really crazy algorithms to get the program to work.

To all the beta-testers: Sorry for the delay. My programmers are working night and day -  I am waiting myself for a functional beta. The first beta will be an XP version and should be ready for beta feedback late this week.

Yeah, intuitively I think I have tried to tailor my keywords to simple and direct search terms, only to be told off for too many unrelated and repeating keywords (IS). Sucks.
But thanks Yuri for your insight.

I can't wait to try it. Key wording is ssssoooooooooo important.
It can make or break a portfolio

The MIZ

Thanks! It will be everything we have waited for� If not, we will simply keep improving it until it is perfect.   

Since we don't know what site you've managed to secure this mysterious back end information from, it may be more site specific than you think.

I'd like to hear more about the beta when it's ready.  Your research makes a lot of common sense. (Which is 95% of marketing  )

Also, how does it relate to a site like IS with the CV?  Does missing a major keyword affect results as much when all those keywords would map to the same CV? 

Yuri:


Also, your link for SS is incorrect.  When I clicked on it, it doesn't show any images.  It seems that you are using your username instead of the SS ID.

same with your fotolia link.  You need to enter your fotolia id# not your name.

Yuri:

May I ask how many searches were in the research data that you based your findings from?

This will give us a better idea of how good the results are.

I'm not doubting your research, just wondering what sort of data you are basing the result on.

My point exactly.

The �adult� and �adults� are useless, and almost never lead to sales.

I find it hard to believe a leading agency released all their searches and and cross referenced those to tracked sales for you just because you asked nicely.  Forgive me for questioning a little bit.

Quote from: GeoPappas on January 08, 2008, 09:59

Yuri:

May I ask how many searches were in the research data that you based your findings from?

This will give us a better idea of how good the results are.

I'm not doubting your research, just wondering what sort of data you are basing the result on.

My point exactly.

Quote

The �adult� and �adults� are useless, and almost never lead to sales.

I find it hard to believe a leading agency released all their searches and and cross referenced those to tracked sales for you just because you asked nicely.  Forgive me for questioning a little bit.

Are you suggesting that I made this up! Seriously.....
I got the information because:
1.   I�m using it for data in a real research project I am doing on search patterns. (I study psychology)
2.   We could not have built the relevancy data-base in our keywording software without it.
3.   We are providing the agency with the results.

The data is based on about 2�000�000 searches

I am sharing this information with you out of kindness...Hmmm...

Very interesting!  I know people on iStock have been asking for this, although I don't really see the use.

Well, is dreamstime the only place to offer this?  Since it is there for the taking, maybe that's where this information has come from.

Shutterpoint shows a list of the most recent terms used by buyers in their searches.
http://www.shutterpoint.com/Photos-SSpy.cfm
It's very interesting also to observe the five top search terms (sex, nude, sexy, nudes, erotic).  And yes, all have less than 8 letters. 

Regards,
Adelaide

Very interesting!  I know people on iStock have been asking for this, although I don't really see the use.

Well, is dreamstime the only place to offer this?  Since it is there for the taking, maybe that's where this information has come from.

No. It's not from Dreamstime and it's not publicly available anywhere. Are you by any change very sceptical natured? 
We did not leech this info from some websites front-end. Think about how long time that would take, getting 600000 clusters of keywords� and then you would still be missing the search patterns!

No. It's not from Dreamstime and it's not publicly available anywhere. Are you by any change very sceptical natured? 
We did not leech this info from some websites front-end. Think about how long time that would take, getting 600000 clusters of keywords� and then you would still be missing the search patterns!

That's not what I meant.  I meant since dreamstime just came out with data that shows a link between keywords searched and image sales, that they have been stockpiling that data.  As I said, I don't know any other company that publicly makes it known they have this data.

Vonkara,

Shutterpoint has no submission inspection, so indeed you find a lot of crap, but there are also some awesome contributors (some sell at US$100 or more).  Storage is not free, but photographers get 85% of the sales.  Prices may look expensive, but their full license is equivalent to an extended license in microstock.  They recently introduced an optional standard license, with restrictions (and prices) similar to microstock's regular license.

I can't complain, as my sales surpassed my storage fees and got me payouts.  This is one of the places where I sell my non-microstock images, although some microstock are there too.

Here is My gallery, if you want to take a look.

Regards,
Adelaide