So are you saying not to include the keyword "background" for a background image so as to not compete with hundreds of millions of images or not make those images at all or just stick it somewhere down after the first 10 or more keywords?  How about for an image with a white (or other color) background - should "background" not be one of the keywords or should it be included but just know that other keywords are more important?

When I look at what DT says was searched to find images sold it is usually a very simple one word search. Same for back when SS showed that info.

Good reminder to put in the words to find the image, not just describe it.

L e a r n    B y    Y o u r s e l f

The only thing you are achieving when you add a keyword "background" is that most likely you are wasting 1 keyword. Microstock agencies are smart enough to understand your image is a background nowadays, without you adding an explicit "background" keyword there. What they are not able to understand - concepts like "sad" etc. Also from the ranking perspective using keyowrd "background" is very questionable, judging from the sheer volume. So I would conclude that with high probability it does some harm (waste) and no good at all.

Quote from: BelieveInStock on February 10, 2023, 04:51

Could you please elaborate ?

Indexing for search is quite more complex than just taking taking all the keywords you entered, creating tens of thousands of "lists" (per each keyword) with all the files that have that keyword and then when customer entered "mangrove forest background", checking if the given file in the in the list of "mangrove", in the list of "forest" and in the list of "background" - then showing it.

The way it works in reality is that all of the these keywords are essentially replaced to one of the concepts (in the so-called "concept tree") that agencies search engine understands. Amount of concepts is smaller than amount of the keywords (there're just too many different words, just the first 100 pages of search results contained approximately 49 thousand different keywords, that's a lot). It could be that the concept is exactly equal to the keyword (e.g. "background") or might be that it's not (e.g. keyword is "paris", but the concept is "city").

The typical mistake everybody are doing - batch-editing pictures and considering keywording "boring" directly trumps your sales.

Quote from: Big Toe on February 10, 2023, 08:38

Some search engines will also include pictures where Paris is only in the title, but not all do that and it would be stupid to rely on that. Even if an image shows up, when the search term is only in the titles, you would have no guarantee that it stays that and also the image may get a better search placement when the word is in the title and the keywords, instead of only in the title or only the keywords.

Only in the title? Where and then why do we worry so much about keywords if they are irrelevant? Maybe I don't understand, but using the Paris example, why would some search ignore the keywords?

True about title and keywords, we know that Alamy and Adobe do that, as their assumption is, we put the most important words in both. Or maybe I should say, we should recognize that only the most important words, should be in both?

Title/Description: Blue sky with clouds Paris France street scene. Might be better found if it just said Paris France, Street Scene as the sky is usually blue and clouds aren't anything remarkable.

But the question is, where do you find a search that only uses Title or Description?

And along with the thread, my main question wouldn't be what keywords do I have in my images, but what keywords do buyers actually search for. I think too much goes into,  creative finding of many words, instead of finding the useful words that buyers actually use. As in the blog post, paraphrased, 50 words may be too many and 18 might be the perfect number.

That all goes back to, if I'm looking at an image, what words would I use to describe it and what's actually in the image? I say those are good keywords.  All the similar, related and different definitions are a waste of time, because almost no one, uses those words to find an image. I'm not buying into the what if someone uses that word, because 98% of the time, no one does and no one will, so why is it worth the time and effort to add something that's ineffective and unnecessary?

Pete, the problem with irrelevant keywords is that you can't tell beforehand for sure if they are irrelevant to the buyer.

My best selling wheat grain field was allways found by keywords "Celiac disease" 

That all goes back to, if I'm looking at an image, what words would I use to describe it and what's actually in the image? I say those are good keywords. All the similar, related and different definitions are a waste of time, because almost no one, uses those words to find an image. I'm not buying into the what if someone uses that word, because 98% of the time, no one does and no one will, so why is it worth the time and effort to add something that's ineffective and unnecessary?

Quote from: Big Toe on February 10, 2023, 08:38

Do you have any shred of evidence that anything like this happens?

Well, as I said, I do not have access to their systems so I cannot say exactly how to they store/index data. However, this part, as I mentioned, is well-researched so I would imagine they use an industry-level solution which I can relatively precisely describe.

If you are interested to read about concepts and search, I'd recomment to start here https://en.wikipedia.org/wiki/Compound-term_processing. Note, this is also a bit of generic hand waving in the article, but at least, if you are interested, it will give you a better direction of the search.

As I mentioned, what is not standardized is how they sort outputs of the search results, while storage and indexing more or less is.

So you are just guessing. I looked into the link you provided and also into wikipedia article on concept search that is linked there, and I do not see any evidence that the search engines at the major microstock sites work that way, except that they may address issues like various forms of the same word (strike vs. striking was given as an example) and sometimes minor misspellings, but the issues with polysemy and synonymy.

That means that a search for "fast" will not show images that only have the keyword "quick", even though the words stand for the same concept.