pancakes

MicrostockGroup Sponsors


Author Topic: Number of title and image keywords in best selling images  (Read 14329 times)

0 Members and 1 Guest are viewing this topic.

« on: September 20, 2014, 09:40 »
-14
Hi guys :)

I did some heavy work to figure out
-the best number of keywords to submit to shutterstock
-the best number of words to use in the title/description

Given the time I invested, I decided to offer the information for a small fee here:
https://www.fiverr.com/manmademagic/reveal-best-number-of-keywords-for-shutterstock-submissions

Check it out and if I see enough appreciation, I will do more research for other websites as well.
I found the results quite interesting.

Have fun

einstein

19.03.2015
Just an update for the haters: Istock confirms how much title and description matters. While you waisted time hating me, I have done it for several months now. By now I am also hitting many first pages, especially on bigstock and 123rf, many of them are single keyword searches. 50 description keywords is btw not the optimum for istock, it's not surprisingly between the numbers they tested (assuming that a description is accurate as they noticed themselves).

I wonder why they need to rewrite 20000 descriptions and plan a rewrite for another 100000 to find out what I already know. I guess they want to increase sales for current images after that realization. Otherwise it's really not the most efficient way to get the data. Here is the link:

http://www.istockphoto.com/forum_messages.php?threadid=365797&page=1
« Last Edit: March 18, 2015, 23:06 by einstein »


« Reply #1 on: September 20, 2014, 10:14 »
0
how many sales do you have by day? in how many files ?

Dook

« Reply #2 on: September 20, 2014, 11:33 »
+5
This statistics matter IF number of keywords and length of title makes any difference. And we don't know that.

Beppe Grillo

« Reply #3 on: September 20, 2014, 11:34 »
+3
Hi guys :)

I did some heavy work to figure out
-the best number of keywords to submit to shutterstock
-the best number of words to use in the title/description

Given the time I invested, I decided to offer the information for a small fee here:
https://www.fiverr.com/manmademagic/reveal-best-number-of-keywords-for-shutterstock-submissions

Check it out and if I see enough appreciation, I will do more research for other websites as well.
I found the results quite interesting.

Have fun

einstein


Well, for a not so big fee I could read your information

« Reply #4 on: September 20, 2014, 12:07 »
+3
I'll wait for somebody whos been in Micro for over 6 months to make these claims. You want money to know best number of keywords, based on what? Title description. How did you come to these heavy work decisions.

« Reply #5 on: September 20, 2014, 12:24 »
+10
The implication is that you put the right number or keywords and title info and you will rocket in sales. Perhaps it is more important to have a great image and relevant keywords, title and description. I don't think this info have a great deal of relevance.

« Reply #6 on: September 20, 2014, 12:27 »
+6
I'm inclined to think number of keywords is less relevant than the keywords themselves and the image in question :-\

« Reply #7 on: September 20, 2014, 14:04 »
+6
If all it took to having a best selling image was the magic number of keywords and words in the titles I would think the author would be making a mint w/ best sellers on SS instead of trying to get us to pay for this maybe not quite so valuable information.

« Reply #8 on: September 20, 2014, 15:17 »
-3
I started my microstock adventure in the beginning of this year. I increasingly rely on software I have written for myself. Analysis of the keyword length and title length is my newest addition. I have build in those results into my software, but since it was so much effort to figure that out, I thought this information should be valuable to others. At least I would have expected some excitement because it's something I always wanted to know.

By now I only have around 300 images online which is by far not enough, since I spent more time writing the software than making photos and submitting. I have at least one sale a day (average 2, sometimes up to 5). Recently my number of the "more expensive" sales went up but it had nothing to do with the new algorithm yet. It was because I'm learning how to do better images, which images sell and I'm filtering out "bad" keywords depending on stock website.

I'm not promising it will rocket your sales but I can imagine that knowing the number from other popular images should give you a small extra boost. And even if not, why to look for a couple extra keywords when it turns out you don't have to. It can be a lot of work, especially if you have serious amounts of images to submit. 100 images times 3 keywords - you saved yourself 300 keywords. I see a lot of people saying that too many keywords are diluting the good ones and other people saying that the more keywords the better the chance that someone finds your image. I just wanted to get closer to the truth, not the speculation, and I wanted to know it for each website separately. Plus I always wrote monster descriptions and I wanted to know if it's worth the effort.

The data is not field tested yet, you have me there, but if you want, I can make a report in 6 months (just write me a message to remind me then). The only problem with that is, I will optimize more than just the number of them. For example I'm getting into questions like, where does plural make a difference or if the word "red" is better than "cherry red", given there is cherry red in the image.

If you have couple of thousand of images, I would expect some impact, that's all I'm saying, I can't prove that. I will use that data from now on whenever you will as well or not.

I also admit that the biggest surprise so far was not shutterstock but dreamstime. They seem to have two sweetspots with a large gap in between. But I haven't yet done all the research to fully confirm it.

Quote
I'm inclined to think number of keywords is less relevant than the keywords themselves and the image in question
Quality is extremely important, no question there, but I think quantity is important as well. Especially for fotolia, istock and dreamstime (clearly below 50 if anyone noticed ;) ). Just do a random search yourself. Actually I think it's important for lots of them. At 123rf I accidentally used the right number of keywords already and my images sell twice as good as on fotolia (and now I know why). At least I strongly believe that after seeing some preliminary results. On 123rf I get quite often into the first page because of that, especially if someone uses 2 search words in my niche.

Plus, let me express my unrelated enthusiasm, I love 123rf for their uncomplicated submission process :D

« Last Edit: September 20, 2014, 15:23 by einstein »

« Reply #9 on: September 20, 2014, 15:45 »
+1
Seriously, this may be one variable that can influence search placement (it's not rocket science to see that, given 2 similar images, the one with 3 times the number of keywords is likely to contain spam and, therefore be less relevant). But only variable one of many that can include dumb luck in terms of placement.  In any case, 300 images over a few months is statistically insignificant given the volumes in MS.

dpimborough

« Reply #10 on: September 20, 2014, 16:11 »
+5
Roll up! Roll up get your snake oil here!  ;D

« Reply #11 on: September 20, 2014, 18:02 »
+4
I would guess (without any investigation), the right number is 7 - isn't it?  8)

« Reply #12 on: September 20, 2014, 18:16 »
-1
I would guess (without any investigation), the right number is 7 - isn't it?  8)

Either that, or it's eleven. Or maybe it's lucky 13!

Whatever.

ShadySue

  • There is a crack in everything
« Reply #13 on: September 20, 2014, 18:36 »
+3
I would guess (without any investigation), the right number is 7 - isn't it?  8)

Either that, or it's eleven. Or maybe it's lucky 13!

Whatever.

It's always 42.

OM

« Reply #14 on: September 20, 2014, 18:55 »
+8
This is free and the major factor is 'luck' combined with an HCV image. Your image goes into the Shutterstock 'new' lottery along with 350,000 other images that week. Your image, as with any other of those 350,000, is dependent on a potential buyer looking in 'new' for a particular subject because maybe they don't see what they want in popular. Unless your image at thumbnail size does not immediately look appealing....fuggedaboutit. Within a few days, in the more popular categories, it will be lost forever if not sold multiple times within the first week.

Beppe Grillo

« Reply #15 on: September 21, 2014, 02:27 »
+2
[] 100 images times 3 keywords - you saved yourself 300 keywords. []

If you have 100 different images representing 100 different subjects you will have, anyway, to write different keywords for each image.
If you have 100 images of the same subject you will copy the keywords from one image, then select the remaining 99 images and paste the keywords in the 99 images in one time
Maybe you should spend less time to do useless researches and more time to improve your workflow?

I would guess (without any investigation), the right number is 7 - isn't it?  8)

My friend from the CERN in Geneva tells me that the right number is 17.54698
« Last Edit: September 21, 2014, 02:30 by Beppe Grillo »

« Reply #16 on: September 21, 2014, 02:40 »
+2
My friend from the CERN in Geneva tells me that the right number is 17.54698

Hmm, maybe I should delete the last letters of each title? Or write it this way: beauti, ful?  8) 8) ;) ;)


Beppe Grillo

« Reply #17 on: September 21, 2014, 03:33 »
0
My friend from the CERN in Geneva tells me that the right number is 17.54698

Hmm, maybe I should delete the last letters of each title? Or write it this way: beauti, ful?  8) 8) ;) ;)

Yo,u go,t i,t!  ;D ;D

« Reply #18 on: September 21, 2014, 06:35 »
-6
Interesting, as far as I can see their explanation, I think they have been contacted by shutterstock to pull my offer down. I asked them to clarify, I think the answer will be interesting, because I don't see how I was violating the rules they mentioned in their explanation. But here you have it. Everyone should be happy. Me (I can keep the information for myself, it seems its value is not recognized here anyway), you guys (you didn't see the chance in front of your face and you don't appreciate me with your -11 votes) and shutterstock (who probably like google has interest in people not knowing how their search works).

Quote
It's always 42

I'm not an expert in cryptography but I wouldn't be a computer scientist if I didn't get the joke.
It's actually more insightful than you probably intended :D (but it's not shutterstock).

Quote
If you have 100 images of the same subject you will copy the keywords from one image, then select the remaining 99 images and paste the keywords in the 99 images in one time


And have you sold any of those images? Sounds to me like the worst idea ever. There are multiple things wrong with that.
Firstly: if you submit a big bunch of similar images, most of them (if not all) get rejected for that very reason (unless you sell your photos on some unknown website in which case they might go for your images anyway). Secondly, if all images are keyworded the same you will compete with yourself in the search results and the missing variation will extremely limit the number of keyword combinations your images can be found under. But I think you just want to tease me to reveal more of what I learned or know.

Seriously, this may be one variable that can influence search placement (it's not rocket science to see that, given 2 similar images, the one with 3 times the number of keywords is likely to contain spam and, therefore be less relevant). But only variable one of many that can include dumb luck in terms of placement.  In any case, 300 images over a few months is statistically insignificant given the volumes in MS.


Thank you. I have to agree with everything you said. Probably the main reason why they punish you for too many keywords. They often get spammy once you run out of ideas. I'm still not sure if they detect spam keywords automatically or if they just assume them to be present when a certain level of keywords is reached. The first apporach is increadibly cheap to implement. The second one costs more but it would be more just to contributors who put great effort to make sure their keywords have high quality. Since the second approach is more expensive (CPU time) you can guess why I think most of them will implement the first one. In other words, I have more reasons to believe, you shouldn't use all 50 keywords, but use that optimal number that has proven to be successful instead. If you do more, you might actually get punished for quality.

That my 300 images are of very little statistical significance, I said that myself. I still think it's enough to have a rough idea about things.

Unless your image at thumbnail size does not immediately look appealing....fuggedaboutit.

True

Hmm, maybe I should delete the last letters of each title? Or write it this way: beauti, ful?  8) 8) ;) ;)

Actually you might have more sales if you do it that way (not really joking) On most websites the keyword beautiful is a bad word that doesn't sell (with few exceptions). And I'm not the only one who came to that conclusion. If you google around you will find http://research.picworkflow.com/
They say beautiful has a rating of 0.005 (good keywords are beginning at 20.0). However beauti has an incredible rating of 116.922 (probably because no one uses it but foreign people mistype often) and ful has a rating of 42.8 :D


Batman

« Reply #19 on: September 21, 2014, 07:16 »
+4
Anybody have some Wellingtons it's getting deep in here. Why not tell us for free since you now took the 5er offer down. SS would not care what you write because it's horseshit.

« Reply #20 on: September 21, 2014, 07:56 »
0
They say beautiful has a rating of 0.005 (good keywords are beginning at 20.0). However beauti has an incredible rating of 116.922 (probably because no one uses it but foreign people mistype often) and ful has a rating of 42.8 :D

@einstein: beautiful, of course, was just an example, it could be any other w,ord 8) Have a nice sunday!


« Reply #22 on: September 21, 2014, 11:01 »
-2
They say beautiful has a rating of 0.005 (good keywords are beginning at 20.0). However beauti has an incredible rating of 116.922 (probably because no one uses it but foreign people mistype often) and ful has a rating of 42.8 :D

@einstein: beautiful, of course, was just an example, it could be any other w,ord 8) Have a nice sunday!


Yeah, sometimes I take what people say too seriosly ;) Sorry about that, don't blame me. I'm a nerd. Have a nice Sunday as well :)
Btw, I couldn't resist but check it again. Guess what they say: "ord" scores a 305 :D You might be into something there :D

http://en.m.wikipedia.org/wiki/Correlation_does_not_imply_causation

Maybe I was implying a causation too strongly, I guess it was the enthusiasm that tricked me ("emotionally compromised" as Mr. Spock would say ;) )
However I can still defend my cause: Speculation about causation may lead to the truth anyway. You need a hypothesis first.

Anybody have some Wellingtons it's getting deep in here. Why not tell us for free since you now took the 5er offer down. SS would not care what you write because it's horseshit.

If that's such a horseshit, why do you want to know? ;) No one wants horseshit, why would you then?

« Reply #23 on: September 21, 2014, 12:40 »
0
Quote
Firstly: if you submit a big bunch of similar images, most of them (if not all) get rejected for that very reason (unless you sell your photos on some unknown website in which case they might go for your images anyway). Secondly, if all images are keyworded the same you will compete with yourself in the search results and the missing variation will extremely limit the number of keyword combinations your images can be found under. But I think you just want to tease me to reveal more of what I learned or know.

1. similars DO get accepted at all agencies
2. I'd love it if a search only generated a bunch of MY images for the buyer to choose from!  why put all your efforts into one perfect image when a variety of images will create more buyers?

saying there is a perfect number of keywords for ALL images is just silly - some images only need a few, some can have 20 or 30 completely legitimate keywords - eg an isolated pencil vs a complicated street scene near an historic landmark

 

dpimborough

« Reply #24 on: September 21, 2014, 14:39 »
+4
Quote
Firstly: if you submit a big bunch of similar images, most of them (if not all) get rejected for that very reason (unless you sell your photos on some unknown website in which case they might go for your images anyway). Secondly, if all images are keyworded the same you will compete with yourself in the search results and the missing variation will extremely limit the number of keyword combinations your images can be found under. But I think you just want to tease me to reveal more of what I learned or know.

1. similars DO get accepted at all agencies


I guess you never tried uploading to Dreamstime then :)

They hate similars  ::)


 

Related Topics

  Subject / Started by Replies Last post
3 Replies
6717 Views
Last post May 11, 2006, 16:01
by admart01
10 Replies
6105 Views
Last post April 03, 2008, 03:09
by Valaaami
11 Replies
4707 Views
Last post February 23, 2012, 18:13
by Mantis
5 Replies
2838 Views
Last post June 07, 2013, 21:34
by Ron
5 Replies
4209 Views
Last post July 22, 2013, 18:28
by cascoly

Sponsors

Mega Bundle of 5,900+ Professional Lightroom Presets

Microstock Poll Results

Sponsors

3100 Posing Cards Bundle