Does anyone know how stock agency search engines rank content?

Last year I took a brief vacation from the world of stock and tried uploading to YouTube which was interesting but didn't work out.

However, it did get me wondering about how stock agency search engines rank new uploads and if they work in a similar way to how the YouTube algorithm monitors 'signals'.

I know the YouTube algorithm is a complex beast that uses many factors to determine if your content is good, and then apply that data to decide whether or not rank or promote your content higher within the search results. But do stock agencies use a similar approach to ranking content every time a customer uses a specific search term?

Does anyone know how stock agency search engines rank content?

The rating of your content depends on the number of views and sales. For example, you added 50 spam tags, your content is viewed by many buyers for some time, but no one buys your content because your tags are spam or your content is poorly made. Accordingly, your content will automatically receive a low rating and buyers will not view it anymore. It's simple.
 

Therefore, your content should have only relevant tags and be of good quality if you want your content to have at least some rating.
 

Unless you actually work for one of the agencies, and are a coder, I doubt anyone here actually knows for certainty how the algorithms work.
Plus - each agency would be unique - because some are more advanced, some are not. There is not one 'search engine program' for all the agencies.

I wouldn't really worry about it - because on of the off chance you do figure out how to 'game' one of the agencies - once they change their algorithm you potentially would be 'devastated' because of your change in sales. Plus - if it becomes obvious that you were gaming the agency (hypothetically say you were keyword stuffing, or figured out how to say repeat keywords that gave you a higher ranking) - it would most likely be frowned upon and you potentially would have some 'xplaining to do if you wanted to continue to sell there.

Yes, apparently some agencies have been 'gamed' to some extent - because some people have bragged about it here. Maybe its true, maybe it's not. But if it was, do you think they'd share their 'secrets' here?

I'd say just do the standard stuff - in that - you be as accurate as possible with your title & keywords, & provide useful content. Hopefully you get resulting sales from that, if not - move on to the next asset(s), and hope for the best here.

No point in getting worked up about how you 'might' be able to 'trick' an algorithm into giving you a (potentially temporary) increase in sales.

Quote from: stocker2011 on August 07, 2024, 16:16

Does anyone know how stock agency search engines rank content?

Last year I took a brief vacation from the world of stock and tried uploading to YouTube which was interesting but didn't work out.

However, it did get me wondering about how stock agency search engines rank new uploads and if they work in a similar way to how the YouTube algorithm monitors 'signals'.

I know the YouTube algorithm is a complex beast that uses many factors to determine if your content is good, and then apply that data to decide whether or not rank or promote your content higher within the search results. But do stock agencies use a similar approach to ranking content every time a customer uses a specific search term?

are you willing to share your youtube experience?  i've been dabbling for a little while and after a few years got my channel monitized a few days ago (not photo related channel).

I do YouTube as well and hopefully to monetise mine too. But it's a hard game. Consistency is the key.

Stock agencies do not not reveal their search algorithms. There's no standard - each agency does their own thing. Algorithms change over time.

Anyone who tells you how it works is either making the best sense of what they've seen - which given their sample size of one portfolio is unlikely to mean much - or they're in the know from an agency about to terminate their employment

If you were an agency, how would you design your algos to be of maximum service to the buyers?

Personally I would add the following mix

- Biggest preference to files that customers view, lightbox and buy. Basically the customers would be the curators on the fresh content, I would collect customer interactions for 3 months to give a "baseline" of how interesting a file is.

- I would give preference to files that are produced in the same region as the buyer, or language of the buyer. A German slice of apple cake looks different to a US apple pie. A German mix of ethnic groups for diversity, is different than in the US or even the UK or France.

- I would prefer files from experienced, reliable sellers, that have a good ratio of sellable files to portfolio size

- I would prefer content from accounts that are actively uploading, if people don't produce actively, their files would slowly drop in rank

- I would let reviewers add a little star or favor for content they find exceptional. It should only increase visibility for 6-9 weeks and then gradually that favor should fade and only customer interactions should matter. Reviewers have seen a crazy amount of what goes in. If they favor 1% of what they review a day, it could be an interesting mix

- I would prefer the ranking of individual files over a general ranking of the port

- I would limit the number of files visible by individual producers on a given search. Not by newest of course, there you can see the 60 fresh files of the same cat playing...but on the major search engine I would set limits. From what I am seeing Adobe limits producers to a max of 5 files per page of 100, irrespective of file type, Dreamstime seems to have a limit of 10.

- I would keep changing the mix of how old content is in top level searches. There might be some bestsellers that have been proven bestsellers for over 10 years. As long as people keep byuing I would keep them on pages 1-3. But i would keep mixing it up with very new content.

- I would give a preference to current search keywords and especially trend keywords

- With the help of ai I would give a preference to unique content. A lot of content is just duplicates of duplicates. But i would look use ai to filter automatically for something that is visually new and interesting and show that to customers. Really unique files are very rare. At least give them a push for 3 months to let the customers "rate" a file by lightboxing it.


What I would do, but no agency does at the moment:


I would allow the customers to build their own personal feed.


- Customers should have the option to follow an artists gallery

- to follow individual artists

- to follow a general keyword search

- to follow a keyword search inside an artist portfolio

Then give the customer a special landing page for "My Feed" that automatically updates with fresh content.

And allow the customer to do a general search of the entire database or a specialised search in the "My Feed"

I think the latter will imrpove the search experience a lot

Because at the moment customers must build a link list and bookmark their favorite producers for a genre.

Agencies should make that process a lot easier.


Whatever it is that you personally can imagine, is probably what is in the algos.

And they keep mixing them up, which is good.

What we can do is keep uploading, be very selective, quality over quantitiy

[...]the best sense of what they've seen[...]  is unlikely to mean much

I do not agree. At least they saw something, not everyone did. Do you really think there is no knowledge within our reach ? Having been a developer and a reviewer may not be a handicap

We must act on the elements on which we can act for an adapted and sustainable strategy.
There is no point in trying to know the details of internal recipes, but we can bring the right ingredients and avoid useless ingredients.

- Optimization of the number of keywords, order or prioritization of keywords, maximum limit (in some places, too many keywords are harmful, in others, too few keywords is a hindrance).
- Use or not of plurals, genders, verbs, spaces...
- Even capital letters are sometimes important
- Use or not of generic words (animal, beautiful, no people, travel destination...), which can dilute the effectiveness of words more closely related to the representation of the image, but sometimes correspond to a search mode
- Use of conceptual, abstract or emotional terms
- Importance (indexing) or not of keywords in the title and/or description

You also have to observe the results of the queries
- proportion of new images
- study for certain keywords of the catalog content to detect shortages and find niches
etc etc...

I'll stop there, I realize that I'm going to spend too much time, there is so much to write... and I also prefer to keep some things to myself, we are in an era of ruthless competition. This forum is not always the ideal place for warm friendly relations either

Stiff competition is certainly true.  Keeping things to one-self is matter for discussion.  It depends what one believes in as general approach in life.  Being kind / sharing because good things are returned with good things, or survival of the fittest.   Not advocating one or the other, just pointing to different sides of the coin.

With regards to Shutterstock, since the big changes which occurred in 2020, i've heard many contributors mention that SS seems to more heavily weight your ranking depending on your upload frequency. Can anyone confirm this from experience? Or did SS publicly announce to contributors?

Also, I think many may agree that their sales are far more erratic than what they used to be pre 2020, which would suggest to me that they jumble the search results on a regular basis.