Shutterstock Best Match - How does it work

[leaf]

Ok, I thought that Shutterstock best match was a 'simple' downloads per day equation .. or at least that each image would have one 'ranking' value, however the downloads/day value was figured out, and that ranking value would apply to all searches that images showed up in.

When looking at a couple searches however, such as horse and stallion, here are the results
Search: Horse
Position #2

Position #8


Search: Stallion
Position #7

Position #36


So you can see that those two images switch order and one moves down a lot in the second search... so obviously the search results are ranked by keyword performance as well.  Not sure this proves or concludes anything.. it was just an interesting observation I wasn't aware of before.

[gostwyck]

I seem to remember SS announcing that the keywords most commonly used to buy an image were critical in sort-order positioning. It was an enhancement to improve the search results for buyers. That was 2-3 years ago I think.

[leaf]

I seem to remember Shutterstock announcing that the keywords most commonly used to buy an image were critical in sort-order positioning. It was an enhancement to improve the search results for buyers. That was 2-3 years ago I think.

Well it certainly makes sense to work it that way.  I suppose considering that they just released the top search keywords used to buy each of our photos, it is a no brainer that they use those stats in their own back-end.

[gostwyck]

Yep, I think they introduced it at about the same time that Istock did.

[Sadstock]

It was discussed a bit in this thread http://www.microstockgroup.com/shutterstock-com/ss-now-showing-keywords-used-to-find-images/

I think that if you look at images with one sale you can deduce a lot about how they make their best match work.  If one term is used, it gets 100% of the keyword weighting for a single sale.  If two words are used, each gets 50% of the keyword weighting for a single sale.  If a compound term like "central park" is used (presumably in quotes) then 33% of the weighting is given to "central park", 33% to "central" and 33% to "park".