Okay, then, skimming through the first 100, they all have the keyword "business" and all but one has flames (many blue flames). The one that is not in flames has 95 downloads. IMO, all seem relevant to business, either by concept or by content. I don't really see any that I would be annoyed to get as a result to my broad search for "business".

If the new best match is triggered by successful downloads, it would suggest that once on the front, forever on the front... but hopefully there are other factors that will counterbalance these feedback loops, and they will be revealed as things progress.

What exactly are you testing for? If it is supposed to be indicative of which images might actually be sold, shouldn't it at least be remotely close to the reality of what a buyer might search for? No designer in their right mind is going to come to a site, enter "business" into a search engine, see that their search has yielded 255,328 results, and start flipping pages.

I'm guessing you must not ever actually use the search engine for the purpose of finding a file to purchase which serves a specific need?

And would you be a buyer, photographer (contributor) or both?

Quote from: hoi ha on December 14, 2008, 22:33

Quote from: whatalife on December 13, 2008, 23:30

What exactly are you testing for? If it is supposed to be indicative of which images might actually be sold, shouldn't it at least be remotely close to the reality of what a buyer might search for? No designer in their right mind is going to come to a site, enter "business" into a search engine, see that their search has yielded 255,328 results, and start flipping pages.

I'm guessing you must not ever actually use the search engine for the purpose of finding a file to purchase which serves a specific need?

And would you be a buyer, photographer (contributor) or both?

I am both, but primarily a designer.

So what term do you use to determine the efficacy of a site's search engine? From my perspective "business" is a pretty good determinant - if the site brings up a toddler playing with a mobile phone in the first 20 images then I know the search engine is lousy ... if it turns up all relevant or arguably relevant images then I know the search engine is probably working pretty well ...

Quote from: hoi ha on December 15, 2008, 00:19

Quote from: whatalife on December 14, 2008, 23:36

Quote from: hoi ha on December 14, 2008, 22:33

Quote from: whatalife on December 13, 2008, 23:30

What exactly are you testing for? If it is supposed to be indicative of which images might actually be sold, shouldn't it at least be remotely close to the reality of what a buyer might search for? No designer in their right mind is going to come to a site, enter "business" into a search engine, see that their search has yielded 255,328 results, and start flipping pages.

I'm guessing you must not ever actually use the search engine for the purpose of finding a file to purchase which serves a specific need?

And would you be a buyer, photographer (contributor) or both?

I am both, but primarily a designer.

So what term do you use to determine the efficacy of a site's search engine? From my perspective "business" is a pretty good determinant - if the site brings up a toddler playing with a mobile phone in the first 20 images then I know the search engine is lousy ... if it turns up all relevant or arguably relevant images then I know the search engine is probably working pretty well ...

Obviously a toddler appearing in a business search is not relevant. Unfortunately, the search engine is somewhat handicapped by the potential (and likelihood) or inaccurate keywords attached to the files.

I guess my point is, I don't ever search for a term. Unless what I'm looking for is rare and hard to find, so that there will only be a few to choose from anyway, my search box will be something like:

business woman smiling "looking at camera"

or

"manual worker" "protective workwear" man industrial

or

office man business team


For me, this is a more efficient way to search because my results are narrowed down from the beginning. I do not have time to sort through hundreds of pages of results.

I think you may have mis-read or I may not have been terribly clear - sorry about that -  of course I would never use "business" as a search term if I was looking for an image - what I was saying (and I think what averil was saying as well but I would not want to put words in her mouth) is, as mentioned, we use the term "business" to determine the efficacy of the search engine only - not when searching for a particular image - it would of course be a collosal waste of time wading through the thousands of images the term "business" would produce ...

So what term do you use to determine the efficacy of a site's search engine? From my perspective "business" is a pretty good determinant - if the site brings up a toddler playing with a mobile phone in the first 20 images then I know the search engine is lousy ... if it turns up all relevant or arguably relevant images then I know the search engine is probably working pretty well ...

I have to agree with whatalife that the toddler image - unless done in a way to simulate an adult businessmasn - would have been wrongly keyworded, so it's not the engine's fault.

I would consider a limitation of the search engine typing "white dog" and getting brown dogs in white background. If the search engine looked at title and description, maybe the right ones would show first.  Try this in DT and you get a slightly better result than IS.

Regards,
Adelaide

Well, I thought I understand what you meant. I thought you were trying to determine how the best match might affect you by judging the arrangement of the results to your one word search. I was just questioning what those findings really represent when they are not typical of how most buyers search (based on my own practices, and those of other buyers who have discussed this on the forums before).

The practices you outlined are definitely the most sensible way to get good results from any search engine IMHO. 

However on Dreamstime we can see the words searched for when our images are purchased and you would be astonished at the frequency of one word and/or two word searches!   (Not to mention the conceptual keywords used to buy our images which are quite often wiki'd into nonexistence on istock) 

I can only assume this is because there are a lot of people buying micro who are not professional designers or even necessarily that tech/search savvy. 

Whatever the reason, it would be a good idea for programmers to bear that in mind when tweaking the search engines on the sites, and also for us as contributors to remember when keywording. 

Quote from: lisafx on December 11, 2008, 12:35

I wonder if the new Premium Collection for exclusive images may be a way to address this exclusive vs. non bias in the best match. 

I can't see how that would happen. My expectation is that a very small portion of content will end up in the premium collection - I don't expect anything of mine will be put there. Those people with professional setups (studio or location) are, I think, the ones who'll see their content there. I believe the notion is to provide some incentive to produce the more costly stuff, which is a great idea, but it isn't going to change how a large number of us part timers work.

So that leaves a huge portion of the content in either the bargain bin or the regular collection. I think the exclusive queue, which gives new images an edge of a few days, is probably a fine bias to keep in the best match-v2 results, but otherwise images should sink or swim based on their saleability.

What I have really disliked is watching images that sell move back in the best match ranking, especially if they just sold versus attracted a lot of looky-loos. That seems to be the polar opposite of what should be happening. What was Joe Gough's lovely analogy? Something about taking the hot selling items away from the front of the store and hiding them in a store room in the back so you couldn't find them.

just noticed I have an image which had just '20 downloads per month' not have a sale in 4 weeks, considering it has a cross as main subject its sad that they killed it before christmas. 

I think I could just about pick all the best match changes.  another one now down to 10 / month as it hasn't sold in 8 weeks.  think I'll stop looking

ditto

OK good. Best day for downloads this month
All my best sellers are back in the front where they were before my sales plummeted.
I'm guessing that once the holidays are over I should do a lot better than Oct., Nov., and Dec.
Very optimistic now.

Wow, that is so much better.  My flamers that were on the last page are now back on the first page and my 'crap' is now rigt where it should be on the last page.

Edited some keywords on a freshly uploaded stuff (just to match the images better), and they are up in the best match ranking. It seems something's really happening.

There has definitely been a change. What were at the bottom of my gallery before (top selling files) are now at the front.

EDIT: A quick search using one of the most relevant and popular keywords showed my photo on the 1st page 2nd photo. It is actually possible to find my images in the search again woo.

It's indeed cool to see sales back at the pre-shake level, but it's also somewhat frightening to see the impact on sales best match algorithm has.

One of my good seller (200+ downloads) had 0 download from begining of September to begining of December (3 months), but has been downloaded 8 times between Jan 01 and Jan 07.

One can wonder how important are keywords and picture quality on sale when you see the huge influence the best match algorith has