This gave me food for thoughts, thats for sure.
A pal of mine, London based at the Ogilvy & Mather, AD-agency, one of the worlds biggest and most creative agencies, was appointed Creative-Director, only a few days back.
Well we got talking for ages and sooner or later slipped into pics, photography, etc. Cutting a long story short, they buy on a global basis, thousands of images per year, for ads, corporate, brouchers, annuals, the lot and unless its RM, they actually buy from independant portfolios, in the micro-stock world, there is no differance at all, he explained and when you buy in such vast quantity its a matter of cost as well.
Further more, the larger Micros are so cluttered with all similar shots and search engines so apallingly lousy that it takes a miracle to find a shot and then you move on.

Makes me wonder really, how many thousands of sales everyone in the micros are loosing, simply because they dont invest more money in expertise, trying to perfect the Search. Its the old story of wnting everything, the cake and eat it, I suppose.

Quote from: lagereek on February 10, 2011, 07:32

...Makes me wonder really, how many thousands of sales everyone in the micros are loosing, simply because they dont invest more money in expertise, trying to perfect the Search. Its the old story of wnting everything, the cake and eat it, I suppose.

Well we've seen the ugly side of what happens when a company tries unsuccessfully to perfect the search. istock has tried for years to improve their search functions, only to make the site more buggy and complicated than ever. It's a great idea and a good goal for a company to have, but I think many companies underestimate the cost of having good people on board who understand search programming and can really improve things. It takes a lot more time, resources, expertise, and money than most companies care to invest.

Thats just it, people who understand search programming!  they dont come cheap but just look at the extra revenue all around. I mean, I bet you will find many of these agencies and no matter what search-words you put in, on the premiere pages, 50% are almost identical shots. What a waste.

Just tried this search on Shutterstock "Black, Businessman, Standing" with 50 images showing 20 images are of White businessman standing.  To get black Businessmen only would I have to put the words "African Descent or Afro American"?  

Show me a non-micro search engine that works better than SS and provides more choice.

Google?

...iStock did not do everything wrong in that regards but the IT side is messing up constantly, otherwise the concept of the search is great...

Without proper functionality, the concept is useless. Sure it's a great idea, but I think buyers would prefer a working less-feature-rich search engine than a feature-heavy but extremely buggy search. Until istock gets the proper manpower and expertise to do it right, I think they're losing the search battle.

That said, it's not like any other microstock agency is making strides in search. Right now it looks like they are all just sticking with what works. But as you said, while thousands of new images hit the sites every day, the search function will ultimately become a major breaking point for some in the future.

Quote from: gostwyck on February 10, 2011, 08:44

Show me a non-micro search engine that works better than SS and provides more choice.

Google?

Google images search is next to useless for most searches other than celebrities.

I think it's possible to develop a search system that works fairly well just by tracking customer behavior and connecting the customer behavior to keywords and images.

I have developed a system (just a simple diagram on paper) how a better search could be done, without disambiguation. I won't be sharing it here, I might use it someday

I can only tell that it's a formula that calculates a "relevancy factor" for every keyword that an image has.

Scenario: 1000 people search for "flower". Image #1 gets shown 200 times but only 2 clicks. Image #2 gets also shown 200 times but gets 58 clicks and 12 purchases.
Result: Image #1 gets a low "relevancy factor" for keyword "flower", and gets pushed back in the search results in the future. The image #2�s keyword gets a very high "relevancy factor" and gets positioned better in search results in the future.

BUT, this does not mean number image #1 will be in the back of the search results, it only gets pushed back when someone searches for "flower".
This ranking could also be used for keyword combinations.

To be fair, iStock's BM2 was moving towards being very good before F5 and then facetted search.
I'm a bit concerned about how some bits of best match worked, and your system would be the same. I had a pic that was for several months on the first page of searches for 'elephant', but (at the same time) on the very back line on a best match search for "African elephant", and could never work that out. It was behind many non-exclusive wrongly-tagged pics of Asian elephants. :-(

Once you have 10 million products on your shelves about which you (the shopkeeper) basically know nothing - labelled with tons of keywords you neglected to control - no amount of search technology is going to be the answer.  It's too late for that.

New agencies need to get started on a better footing by being more selective as to subject and composition - having high but consistent and sensible quality standards - vetting all the keywords without setting up a tedious CV nightmare - correcting minor keyword problems at submission without wasting everyone's time with rejections for trivia.

They need to keep submission time down to the point that contributors can make payback on the time spent producing an image, without needing hundreds of sales of that image.

What it all comes down to is finding reviewers who understand what they're doing, then paying them enough to keep them around.  Mindless crowdsourcing has been tried, and it worked for a while, and made some people a lot of money. But that's over.

All we need now is a way to connect with buyers on our own.  I have faith that one will emerge.

Quote from: lagereek on February 10, 2011, 07:32

...Makes me wonder really, how many thousands of sales everyone in the micros are loosing, simply because they dont invest more money in expertise, trying to perfect the Search. Its the old story of wnting everything, the cake and eat it, I suppose.

Well we've seen the ugly side of what happens when a company tries unsuccessfully to perfect the search. istock has tried for years to improve their search functions, only to make the site more buggy and complicated than ever. It's a great idea and a good goal for a company to have, but I think many companies underestimate the cost of having good people on board who understand search programming and can really improve things. It takes a lot more time, resources, expertise, and money than most companies care to invest.

Controlled vocabularies do not work outside of the agency itself.  They had their place 15 years ago, but today they are archaic.  This is why iStockphoto struggles to find placement in Google Images.  That, and keyword stuffing, which is also causes Google to lower relevancy rankings.  Of all the micros, iStockphoto is probably least responsible for keyword stuffing because of its controlled vocabulary (no need to put "ocean", "water", "sea", "sea water," "salt water", etc in as keywords).  However its controlled vocabulary is not well recognized by Google (ie, "Nobody")

Stocky imagery search is a market that Google hasn't gotten around to exploiting yet, but it's no doubt on one of their lists.  When GoogleStock finally shows up - with options like "show RF images only" - today's microstocks could find their business melting away like a popsicle in the sun.  Hope so anyway.

Or if not Google - why couldn't we sell through Amazon?  It's all possible.

A lot of the sites are trying to improve things with various tweaks, but they are mostly all hampered by the weak link - the keywords, this is especially bad at the sites that split apart multiple work keywords. The problem for contributors is different for buyers, because if there are 100 images (or 1000s) that would work for the buyer, any of them would be fine, but the contributor wants the buyer to at least see their specific image. Looking at the search terms that DT provides most of the searches are one or 2 words, but if a buyer wants to drill down to more specific images they need to have more specific searches.

Still, for example in SS you can choose various properties that are linked to model releases.

IS could have locked things up with their CV, but in my opinion the way they implemented it both encouraged spam and makes things that aren't in the CV but are similar just not work at all. They could still come up with something that works well, but I haven't seen much encouragement there lately. The few times I tried their search it just didn't work at all, so who knows what is happening with it.

You know, I have seen so many buyers, in ad-agencies, designers, God knows and the way they work when searching for a specific shot. They will first go to their favorite site, there they spend 3 minutes at the most, flicking through no more then 5 pages, that your lot mate and if they dont find it there they move on to next site.

They all like the keep-it-simple method, not having to choose between diferant search methods, not having to see any rolling lists falling down, not having to explain the search word and ofcourse the real search-killer, NOT having to wade through pages with series of almost identical files.

Ofcourse, today with agencies housing more then 10 million files,  well its a bit too late, isnt it.

Or if not Google - why couldn't we sell through Amazon?  It's all possible.

Now you're talking!

Quote from: Eyedesign on February 10, 2011, 09:48

Just tried this search on Shutterstock "Black, Businessman, Standing" with 50 images showing 20 images are of White businessman standing.  To get black Businessmen only would I have to put the words "African Descent or Afro American"?  

I think half the battle is educating the users on the features available.  Creating a good search engine is difficult if the users don't make use of any features outside of the basics.  Even Google becomes that much more powerful if you exclude a keyword or two.  The SS example, you would just check the 'People search' and select African American (or the other options they have depending what you mean) from the list...

Exactly, in addition to the black businessmen you will also get all the buisnessmen wearing a black suit or against a black background or wearing a black tie, etc etc.  Input something vague in the search field and you will get something vague back.

I had this argument with Dreamstime; they would not allow geographic terms in my keywords and I told them they were essential for the identification of the animals with respect to location.  Do a search for tiger or Siberian Tiger; you will get thousands of zoo images.  Do a search for Primorski, Tiger, Russia and you will get mostly Tigers from the wild.

Another example, go a search for Eastern Great Tit (a bird) and you get thousands of images of pretty girls from the east or looking east or that have been  on a trip to the east, etc.  Do a search for Eastern Great Tit, Amur and you will get photos of a small bird with a colorful pale yellow belly.  But, DT will not allow the words Primorski, Russia or Amur in the keywords.

I, the seller, am screwed because the buyer can not find my work and the buyer is screwed because he/she must wade through junk to find the images

the searches are a problem in IS because it is not focused on the buyer at all. It is focused on pushing chosen higher value content and a bias towards exclusives. The buyer wants the best picture not IS margin grabs or its attempt to keep exclusives. Only educated (in IS systems) buyer find it easy.
What is disturbing to me is there seems to be looser standards for exclusive content. Multiple similar shots get in clogging the searches and at times happy snaps that would not make it in for independents - check out the critique forums. Some shots are poor to the point you shake your head and then hear "it is fine it got it". I know there are extremely good exclusives and I am not doubting that their content is good.

This all means worse and more expensive images first?

I respect tyour view but are you an exclusive and not a buyer to feel that way?

I had this argument with Dreamstime; they would not allow geographic terms in my keywords and I told them they were essential for the identification of the animals with respect to location.  Do a search for tiger or Siberian Tiger; you will get thousands of zoo images.  Do a search for Primorski, Tiger, Russia and you will get mostly Tigers from the wild.

Another example, go a search for Eastern Great Tit (a bird) and you get thousands of images of pretty girls from the east or looking east or that have been  on a trip to the east, etc.  Do a search for Eastern Great Tit, Amur and you will get photos of a small bird with a colorful pale yellow belly.  But, DT will not allow the words Primorski, Russia or Amur in the keywords.

I, the seller, am screwed because the buyer can not find my work and the buyer is screwed because he/she must wade through junk to find the images

Sorry but that's a poor example. If you want to sell images of wildlife then it is absolutely unforgivable to not use the binomial nomenclature (Latin/Greek names) to identify them. If you want to sell an image of a Great Tit then you should have the words Parus major within your keywords __ which you don't. The location then becomes largely irrelevant as birds have a tendency to fly to different places but obviously their name remains the same. It also differentiates you as a professional supplier of wildlife images from the happy snapper at the zoo or in their garden. You could really do with improving your keywording. It is also essential to state what the subject is doing __ is the bird flying, perching, eating or what? Whenever I sell a wildlife image on DT the buyer ususally keywords the action and very often the Latin name.

If you have a crappy title and crappy keywords, the search engine will deliver crappy results.  The reviewers need to do their job and they need some reviewers to go through the old stock and cull out the bad stuff.

+1

plus if someone enters a crappy search of one or two words, they will get stinky results for everything that has those two words. What did they expect? The search engine is not intuitive and doesn't know what someone wants, anymore than a computer can guess that you wanted it to do something else when you typed in a wrong command. GIGO

All the people pointing out that black is also black and there are black suits... have covered the whole problem that seems to get ignored every time this keyword and search engine riot breaks out in a new thread. Try walnut, is it wood, a tree, a color, a part holding a wall or something else?   

Words Have More Than One Meaning!

As for the theory of views, clicks and zooms which would rank the pictures for future searches, by how buyers looked at them. Alamy does that, using their top buyers, and people complain about the search there because it finds the words they searched for! I always ask, did you want it to find words you didn't search for somehow?

No search engine is going to work if the words and descriptions are full of irrelevant words and spam. NONE. No search will work when words have multiple meanings, it's language and impossible to know which meaning someone wants, except maybe with CV and we can all see how limiting that is, especially when the word we need, doesn't exist because they haven't added it or it's spelled differently in Canada than other places.

Yeah, when is the secret code book for IS keywords going to be published.

There's no pleasing anyone... that's my answer.

Quote from: visceralimage on February 11, 2011, 01:08

I had this argument with Dreamstime; they would not allow geographic terms in my keywords and I told them they were essential for the identification of the animals with respect to location.  Do a search for tiger or Siberian Tiger; you will get thousands of zoo images.  Do a search for Primorski, Tiger, Russia and you will get mostly Tigers from the wild.

Another example, go a search for Eastern Great Tit (a bird) and you get thousands of images of pretty girls from the east or looking east or that have been  on a trip to the east, etc.  Do a search for Eastern Great Tit, Amur and you will get photos of a small bird with a colorful pale yellow belly.  But, DT will not allow the words Primorski, Russia or Amur in the keywords.

I, the seller, am screwed because the buyer can not find my work and the buyer is screwed because he/she must wade through junk to find the images

Sorry but that's a poor example. If you want to sell images of wildlife then it is absolutely unforgivable to not use the binomial nomenclature (Latin/Greek names) to identify them. If you want to sell an image of a Great Tit then you should have the words Parus major within your keywords __ which you don't. The location then becomes largely irrelevant as birds have a tendency to fly to different places but obviously their name remains the same. It also differentiates you as a professional supplier of wildlife images from the happy snapper at the zoo or in their garden. You could really do with improving your keywording. It is also essential to state what the subject is doing __ is the bird flying, perching, eating or what? Whenever I sell a wildlife image on DT the buyer ususally keywords the action and very often the Latin name.

Point taken, understood, will revise my keywords

I've just done a google image search with interesting results
I put in 3 of my common keywords which I use at all sites.  I got over 10 million results and the first 4 images were  mine.  Anyway the first 2 were from 123rf, the third from Bigstock and the 4th from Dreamstime.
I didn't get a fotolia, or istock image until the 2nd page after spiderpics whatever that is and SS didn't appear until page 8.
It makes me think that if 123 and bigstock get such good google results but such bad sales compared to the big 4 then very little buying traffic comes through google.

Search your name and see what happens. "[fname lname] photo" I was very surprised to see an avatar from a flying website my first (wow strange?) followed by istock, istock, mostphotos (my avatar?), 123rf, 123rf, (123RF is a closed account!), Imagekind (one photo account that I opened in 2005), iStock, private site, private site (a drivers webpage), icon gif from one of my websites, istock, another one of my websites (not the same one), Alamy logo from one of my websites... Not one BigStock on the front page. Kind of amusing.

I wonder why the Google Image search find avatars and gifs from my web pages, before it finds web images. And by the way, the IS images are the oldest uploads from my application set. All very odd. Yes, 123RF apparently has the best identification for getting picked up by the Google Image search for artists name.

Searching for some of my subjects by drivers name picasaweb.google.com and flickr come up and nothing on my websites in the first page.

I guess I'd better work on that? 

http://www.danheller.com/blog/posts/do-buyers-use-search-engines-more-than.html

For example, if only .1% of google searchers would actually license an image they found for a given search, the fact that the site itself gets hundreds of millions of searches a day, even the tiny .1% that might convert into buyers would still yield a higher total than what all the photo stock agencies yield...

...To use my site as strictly anecdotal evidence, I get 20,000-30,000 visitors a day, and of those who license images, only a small handful even knew that "stock photo agencies" existed in the first place.


If search results weren't lucrative, Google would have gone out of business a long time ago.  But if you think nobody uses Google to search for things to buy, then I completely understand why you feel only an agency can represent you.

Quote from: fotografer on February 13, 2011, 03:43

It makes me think that if 123 and bigstock get such good google results but such bad sales compared to the big 4 then very little buying traffic comes through google.

That hardly seems reasonably considering Google's business model is based around search results leading to commerce.  Otherwise nobody would care how they ranked in Google.

I think what is more likely is that very few people sign up with microstock agencies after searching for an image in Google.  I don't have any data to back it up, but if we assume someone is looking to purchase a single photo, they will have a hard time doing so once they click through to the microstock site.  As soon as they do, they are hit with a myriad of credit options or subscriptions.  The buyer is probably saying to herself, "I just want to purchase this one photo.  Why can't I just purchase this photo without buying all of these credits?"

And the search either goes on from there, or stops all together.  Either way the microstock site probably doesn't sign up the buyer looking for a single photo to fit their specific need.

That's a good point, but I've seen credit packages starting at $5, others start at $10. 123rf and Bigstock start at $15, even if they use this to buy one image, it's not a big amount. I've only seen Cutcaster allow individual purchases, without buying credits.