pancakes

MicrostockGroup Sponsors


Author Topic: Keyword research and science.  (Read 16901 times)

0 Members and 1 Guest are viewing this topic.

Yuri_Arcurs

  • One Crazy PhotoManic MadPerson
« on: January 07, 2008, 06:52 »
0
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.
« Last Edit: January 07, 2008, 06:55 by Yuri_Arcurs »


« Reply #1 on: January 07, 2008, 09:01 »
0
I can't wait to try it. Key wording is ssssoooooooooo important.
It can make or break a portfolio

The MIZ

« Reply #2 on: January 07, 2008, 10:22 »
0
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.

RT


« Reply #3 on: January 07, 2008, 12:57 »
0
In other words. All the keywording I have done on my 8000 picture portfolio is wrong :(

Now you tell me, what a waste of time it was copying your keywords then  ;)   (I'm joking by the way)

Yuri_Arcurs

  • One Crazy PhotoManic MadPerson
« Reply #4 on: January 07, 2008, 14:20 »
0
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.   

« Reply #5 on: January 07, 2008, 17:27 »
0
Hi,

very interesting  research. Now, I'm in big dilema about my KW's. Please put me on beta tester list and I hope will bi one of the first buyers... hahaha

Regards,
.shock

« Reply #6 on: January 07, 2008, 17:45 »
0
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.

Yuri_Arcurs

  • One Crazy PhotoManic MadPerson
« Reply #7 on: January 08, 2008, 09:29 »
0
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 sense a slight scepticism and irony in this reply. If you dont like the research feel free to not use it.

My Data is from an agency and are based on searched performed by real buyers over the last year. Please respect that I cannot say any more. :)

It was not easy to find an agency that would share these information and we had to guarantee that the program was done in such a way, that there would be plenty of room for custom and original keywording and that people would not just start keyword spamming again. I would love to publicly thank the agency at hand but I cant because this would make it very complicated for us to get other agencies to sell the program. :(

« Reply #8 on: January 08, 2008, 09:53 »
0
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? 

« Reply #9 on: January 08, 2008, 09:59 »
0
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.

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.

« Reply #10 on: January 08, 2008, 10:00 »
0
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.

« Reply #11 on: January 08, 2008, 19:34 »
0
Yuri, if you need a designer betatester, Im your man.

« Reply #12 on: January 08, 2008, 20:40 »
0
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.
« Last Edit: January 08, 2008, 20:46 by sjlocke »

« Reply #13 on: January 09, 2008, 05:25 »
0
Keywords longer then 8 characters account for less then 10% of the total search rank from all keywords.
Yuri, your first post contents only 8.8 % of words longer than 8 characters. This kind of information just means... nothing. Maybe words longer than 8 characters simply count for less than 10% of all english/american words... ?
The keywords: splat, splatter, award, happy hour are highly searched for but no one uses them.
Here we are, and I really hope that this kind of information would be shared in public by this unnamed site, for the good of both contributorS and agency, and will not be kept as a secret of gods ;)

Btw, this is a very cool idea, and I'm also ready if you need some more beta testers ! :)
« Last Edit: January 09, 2008, 07:21 by tilo »

Yuri_Arcurs

  • One Crazy PhotoManic MadPerson
« Reply #14 on: January 09, 2008, 06:12 »
0
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.   Im 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 2000000 searches

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

michealo

« Reply #15 on: January 09, 2008, 07:29 »
0
Just wondering whether you have looked at

http://www.google.com/trends/hottrends

and http://www.google.com/press/zeitgeist.html

as you may find some additional frequently searched keywords (obviously you would have to ignore all teh pop star references) but they have a large sample group

« Reply #16 on: January 09, 2008, 08:42 »
0
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.

steheap

  • Author of best selling "Get Started in Stock"

« Reply #17 on: January 09, 2008, 12:42 »
0
This is a very interesting thread and taught me a lot! Thanks.

I'm new to the Forum, and apologize if this has been answered before, but it appears to be relevant to the topic. What is the relationship between the categorization that many stock sites require and success in having a photo appear in a search? The selection of the various categories (which are all different from site to site) is one of the most time consuming part of this business.

Steve

« Reply #18 on: January 09, 2008, 15:42 »
0
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.  :D

Regards,
Adelaide

« Reply #19 on: January 09, 2008, 16:06 »
0
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.  :D

I doubt those words are coming from buyers though ;)

Yuri_Arcurs

  • One Crazy PhotoManic MadPerson
« Reply #20 on: January 09, 2008, 18:19 »
0
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!

« Reply #21 on: January 09, 2008, 20:27 »
0
I doubt those words are coming from buyers though ;)

They possibly are!  When seeing "what's selling" in SP, there is a lot of nudity (some very tasteless, IMHO).

Regards,
Adelaide

« Reply #22 on: January 09, 2008, 21:06 »
0
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

« Reply #23 on: January 09, 2008, 21:12 »
0
Just curious, what is that Shutterpoint. They seem to sell pictures between 30$ to 80$ each...the interface is poor like Bigstock is and I feel like I was in a bad picture gallery site. Dull pictures are in front page? Who is buying there...? Do that's call macrostock?

« Reply #24 on: January 10, 2008, 11:37 »
0
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
« Last Edit: January 10, 2008, 11:39 by madelaide »


 

Related Topics

  Subject / Started by Replies Last post
7 Replies
3706 Views
Last post March 16, 2016, 10:37
by wordplanet
12 Replies
3381 Views
Last post October 18, 2013, 13:12
by Starzipan
7 Replies
2644 Views
Last post November 08, 2013, 02:37
by cobalt
4 Replies
2749 Views
Last post December 31, 2013, 19:50
by Uncle Pete
6 Replies
5582 Views
Last post November 17, 2016, 06:09
by rjung

Sponsors

Mega Bundle of 5,900+ Professional Lightroom Presets

Microstock Poll Results

Sponsors

3100 Posing Cards Bundle