MicrostockGroup Sponsors


Author Topic: ImStocker Keyworder - new tool to choose profitable keywords  (Read 1828 times)

0 Members and 1 Guest are viewing this topic.

« on: January 14, 2019, 15:05 »
+2
Hello! My name is Mikhail and I am developer of ImStocker :)

Keywording is hard but very important process. Your keywords should be not just relevant, they also should be competitive. To help finding such keywords I have created IMS Keyworder service (https://imstocker.com/en/keyworder [nofollow])

Enter some search phrase, pick similar pictures and the service will show you list of relevant keywords and how good they are.

Each keyword in service have special mark, called rank, which calculated based on popularity (how often customers download assets with that keyword and how often they view these files) and competition level (number of assets which contain that keyword). This information helps you choose most profitable and relevant keywords.

Other features:
   saving metadata to JPEG-files
   advanced search: type of assets (photo, illustration, vector or video), type of license, presence of model release
   combining results of subsequence searches
   spellchecking
   trademark highlighting
   synonyms
   keyword templates
   ..

This service is free and I hope you will like it  :)

Please try it here: https://imstocker.com/en/keyworder [nofollow]


« Reply #1 on: January 14, 2019, 15:43 »
+3
I was intrigued to see what your tool did, so I tried some searches including keywords that I expected would show some of my popular images.

They did show up, but the keywords shown in purple for them - if I understand it correctly, those would be the best selling keywords for those files - aren't the keywords that Shutterstock shows me as the top selling ones (using the top performers list).

I obviously have no clue as to whether what SS shows me is accurate, but as you're apparently looking at SS, AS and CanStock (looking at the thumbnails that show up), how do you know?

Images are duplicated when they are on multiple agencies - it'd probably make more sense not to show the same image multiple times, or at least to somehow indicate it's a duplicate.

I tried the synonyms feature and it was utterly wrong - lots of truly useless suggestions (very funny though; dixie and southwest as synonyms for "New England" and trendy, fashion and accessory for "New Hampshire")

It is certainly important to include relevant keywords, but a keyword is either relevant or not, and how often it's searched for by buyers doesn't make it any more relevant to an image.

What is the thinking behind showing the "competitveness" of keywords? Spamming images with irrelevant keywords that are popular is already an issue; doesn't a tool like this just encourage more spamming?

« Reply #2 on: January 14, 2019, 16:33 »
0
It is certainly important to include relevant keywords, but a keyword is either relevant or not, and how often it's searched for by buyers doesn't make it any more relevant to an image.

I disagree completely on that point.  Assuming the keyword is relevant to my image, then I absolutely want the keywords that are most often used by buyers.

If all my keywords are relevant, but no buyer uses them, then I have wasted my time even bothering to upload it. 

I frequently find that I can get more than the maximum allowed 50 keywords.  That is where knowing the "competitiveness" of the keywords comes into play.  If I find myself with 60, currently I knock out 10 that I would personally not use.  I would much rather knock off 10 that I know buyers are not using though.

ShadySue

  • There is a crack in everything
« Reply #3 on: January 14, 2019, 17:19 »
0
It is certainly important to include relevant keywords, but a keyword is either relevant or not, and how often it's searched for by buyers doesn't make it any more relevant to an image.

I disagree completely on that point.  Assuming the keyword is relevant to my image, then I absolutely want the keywords that are most often used by buyers.
That's what she said.

Quote
If all my keywords are relevant, but no buyer uses them, then I have wasted my time even bothering to upload it. 
That's a totally different issue, but a relevant one at mini-micro prices.

« Reply #4 on: January 14, 2019, 17:55 »
0
thanks. Just tried it for a couple and i think it works well if you choose the right images and delete the irrelevant tags. The colour coding is an interesting way of looking at possible keywords if nothing else.

« Reply #5 on: January 14, 2019, 19:35 »
0
I've tried a lot of these tools and though I've been using the same one for years, this one looks interesting enough to spend a little more time on. A couple observations. It is looking at recent files; a file of mine uploaded a couple of days ago popped up in my search.

Also, and I may be wrong, but the words you got if you clicked 50 and the ones you got if you click 25 and changed it to 50 aren't the same.

Per relevance, two that showed up in mine were lscape and lmark. I can't imagine anyone ever using those tags to look for an engine, so would question the relevance.

At any rate, looking forward to digging in a bit more.

« Reply #6 on: January 14, 2019, 22:44 »
0
It looks interesting, but it need a lot more developing before to trust and use it. I searched for bitcoin and it shows me the word "bitcoin" with rank 0

« Reply #7 on: January 15, 2019, 03:07 »
0
Thank you for all your reviews! :)

Idea of keyword ranking is in exactly what mindstorm said:

I frequently find that I can get more than the maximum allowed 50 keywords.  That is where knowing the "competitiveness" of the keywords comes into play.  If I find myself with 60, currently I knock out 10 that I would personally not use.  I would much rather knock off 10 that I know buyers are not using though.

When you have described your file with more than 50 relevant words you have to choose which keywords should be removed. And ImStocker shows you additional statistic information to help you make your choice. It is not about spamming. It is about optimizing your metadata

I was intrigued to see what your tool did, so I tried some searches including keywords that I expected would show some of my popular images.

They did show up, but the keywords shown in purple for them - if I understand it correctly, those would be the best selling keywords for those files - aren't the keywords that Shutterstock shows me as the top selling ones (using the top performers list).

For now keywords rank is system-wide property, it is calculated based on information of entire database (about 7 000 000 analyzed files). Therefore, rank of same keyword in different queries will be same. Thats why it is important also to take into account frequency of keyword how often this keyword appeared in selected files (showed in parenthesis)

I obviously have no clue as to whether what SS shows me is accurate, but as you're apparently looking at SS, AS and CanStock (looking at the thumbnails that show up), how do you know?

Keyword rank is based only on open information. To calculate competitiveness (C in popup window) ImStocker uses number of analyzed works with such keyword on ShutterStock, Fotolia, Dreamstime and CanStockPhoto. To calculate download (D) and view (V) rank data only from Dreamstime and CanStockPhoto are used, because Shutterstock and Fotolia are not showing this information. Market of Dreamstime and CanStockPhoto are very small relative to Shutterstock, that is why ImStocker uses not absolute values but relative ones. While customers are downloading and viewing pictures with same themes, it is possible to extrapolate received data for other microstock agencies

In future, I plan to make ranking algorithm more smart that is why it is very important for me to receive such comments like yours. I will really appreciate if someone will share with me information about searched keyword from top performers list. It will be very helpful for me

 
Images are duplicated when they are on multiple agencies - it'd probably make more sense not to show the same image multiple times, or at least to somehow indicate it's a duplicate.
 

I know about this issue and plan to fix it in future updates

I tried the synonyms feature and it was utterly wrong - lots of truly useless suggestions (very funny though; dixie and southwest as synonyms for "New England" and trendy, fashion and accessory for "New Hampshire")
 

Synonym lists are computed using AI. They are not manually curated lists. Thats why I have called them Synonyms & semantically related words in popup window and only AI knows how exactly these keywords are semantically connected:)

Sometimes it is helpful, and I agree that sometimes it is totally wrong. It will be great if some native-speaker will help me to fix it.

Also, and I may be wrong, but the words you got if you clicked 50 and the ones you got if you click 25 and changed it to 50 aren't the same.
 

Thanks for noticing. I think it was because now these buttons works as switchers. When you click 50, you get [150]. Then you click 25, you turn off first 25 words because they are already selected, and you get [2650]. Then you click 50, and you will get [26..50, 1..25[. because last 25 are already selected and ImStocker is not changing position of already selected keywords. I will try to make behavior more obvious

Per relevance, two that showed up in mine were lscape and lmark. I can't imagine anyone ever using those tags to look for an engine, so would question the relevance.
 

Keywords in suggested list firstly are ordered by number of repeats in selected files. If you have seen lscape and lmark it is mean that authors are often using these keywords for some strange reason :)

It looks interesting, but it need a lot more developing before to trust and use it. I searched for bitcoin and it shows me the word "bitcoin" with rank 0

Zero rank means that this keyword is appeared very small times in database of already analyzed files by ImStocker. This keyword will receive real rank when ImStocker will analyze enough files with that keyword. Use ImStocker and it will happen sooner :)

Thank for all of you again for your comments!

« Reply #8 on: January 15, 2019, 09:35 »
0
...To calculate download (D) and view (V) rank data only from Dreamstime and CanStockPhoto are used, because Shutterstock and Fotolia are not showing this information. Market of Dreamstime and CanStockPhoto are very small relative to Shutterstock, that is why ImStocker uses not absolute values but relative ones. ...

DT and CanStock are not just smaller, but arguably increasingly irrelevant in the marketplace and doing anything on the basis of their sales patterns has major risks.

I just looked at my last 20 sales at DT and at SS to see if there was any overlap at all (and that has issues as a bad day at SS is now about the same as a good month at DT in terms of volume) and there wasn't any. Not one image.

I went back to a day just before Christmas at SS to see if that helped line up a bit better with DT and that gave me one file in common to both lists.

I left CanStock years ago (because of slow and increasingly low value sales) so I can't compare.

I did a similar last 20 at AdobeStock and likewise, no overlap. Went back to December - one file overlap (and it's the same file as at SS).

I understand you can't get the data you want from the agencies that are actually selling but that doesn't improve the relevance of the data you can get.

« Reply #9 on: January 16, 2019, 07:42 »
+2
Hello Mikhail!
Wow, you have developed very interesting tool to be honest. Certainly it has a potential to become very valuable resource. Have the best of luck to improve and tweak it further, looking forward to it's development!

Uncle Pete

  • Great Place by a Great Lake - My Home Port
« Reply #10 on: January 16, 2019, 07:49 »
0
Hello Mikhail!
Wow, you have developed very interesting tool to be honest. Certainly it has a potential to become very valuable resource. Have the best of luck to improve and tweak it further, looking forward to it's development!

I'd second that thought, just about what I was going to write. I see loads of potential and promise, I hope for continued development.

« Reply #11 on: January 16, 2019, 09:34 »
0
Really useful tool

« Reply #12 on: January 17, 2019, 18:11 »
0
Nice. :)

jovannig

  • I love to travel and make pictures
« Reply #13 on: January 24, 2019, 04:25 »
0
Great tool!
I will try to use it instead of the SS one.
I think the first keywords should be the purple ones, then the blue ones, regardless of the number of appearances.Thanks and I hope you'll further develop this tool!

« Reply #14 on: January 24, 2019, 06:02 »
0
Thank you for your kind words!

steheap

  • Author of best selling "Get Started in Stock"

« Reply #15 on: February 01, 2019, 16:19 »
0
I've been playing with it and while I may not understand everything about your color codes, I'm finding it helpful to get me started on my keywords. What I found would be helpful is that when you click the arrow to show more images, at the moment you just show one more. It would be better if you showed a new row of images to choose from - ie replace the current set by a new set.

I plan to do a blog post at some point comparing these different keywording systems - it would be interesting to see how they all compare.

Steve

« Reply #16 on: February 02, 2019, 23:54 »
+1
Looks interesting, maybe you'll want to post a short info on why it'll work better than lets say opening up shutterstock and using their tool.

« Reply #17 on: Yesterday at 13:11 »
+1
I've been playing with it and while I may not understand everything about your color codes, I'm finding it helpful to get me started on my keywords. What I found would be helpful is that when you click the arrow to show more images, at the moment you just show one more. It would be better if you showed a new row of images to choose from - ie replace the current set by a new set.
Thank you for suggestion! I will think about it

I plan to do a blog post at some point comparing these different keywording systems - it would be interesting to see how they all compare.

It will be very interesting research, please, send me a link when you publish it (if not yet). And if you will need some help I will be glad to help you :)

Looks interesting, maybe you'll want to post a short info on why it'll work better than lets say opening up shutterstock and using their tool.
I believe that ImStocker Keyworder has at least two strong advantages relative to SS tool.

Firstly, it shows additional information about keywords, based on statistic of their usage, which helps you to choose them more wisely. It is some sort of research project, where I try to find answers to questions how to make keywording process easier? How to automate it? How to help customers to find your images/videos?.

Secondly, IMS Keyworder is more flexible tool: you can combine different search queries, you can search thought different markets, you can customize tool with a lot of settings, you can save metadata into jpeg files online and etc.

:)



« Reply #18 on: Yesterday at 13:28 »
+2
New update of ImStocker (https://imstocker.com/en/keyworder [nofollow]):

1) New advanced search
Now you can choose where to search images and specify which keywords should be excluded

2) Skipping assets with similar keywords
Now ImStocker removes images from search result, which have same set of keywords as already existent ones (usually it is duplicating pictures)

3) New Dark theme
Now you can select dark theme of interface in settings window (to work at the night :) )

4) Highlighting of duplicating keywords
When you enter keywords by yourself, you can accidentally write already existent keyword, and now ImStocker shows such words.

5) + many small improvements

https://imstocker.com/en/blog/23-update-imroved-search-and-interface [nofollow] - release notes with pictures

Hope you will like this update :)



 

Related Topics

  Subject / Started by Replies Last post
2 Replies
2007 Views
Last post May 03, 2009, 02:40
by antoniodalbore
3 Replies
1396 Views
Last post November 17, 2011, 20:14
by Lizard
3 Replies
2046 Views
Last post January 17, 2012, 22:32
by RacePhoto
8 Replies
5126 Views
Last post April 14, 2016, 14:00
by Anyka
4 Replies
1624 Views
Last post April 11, 2017, 01:39
by gambolt

Sponsors

Microstock Poll Results