After reading my story “
Getting Images Seen” on how Getty Images orders search returns John Lund did some searches for his own images on the Getty web site. He came up with some additional insights. Check out the full text of his article on his
blog.
John discovered that one his older images in the Stone collection came up as the 8th image when he did a search. This image has been a best seller through Getty. John concluded from this that rather than showing the newest images in the collection first, the algorithm is weighted in some way to bring at least some of the best selling images to the top of the search return. I agree, and am thankful to have the benefit of John’s research.
Here’s what I now think may be happening. It seems clear that the brand order is the same as I described in
my story. Thus, the 8th image shown will always be a Stone image if all the other seven brands ahead of Stone have at least one image that fits the keyword.
However, within each brand Getty may be raising the best selling image to the top. Thus, since the 8th slot is for a Stone image the best selling Stone images in that category will appear. In this case John’s image was probably the best seller in the category. However, before it was time to show an image from the Stone collection images from Digital Vision, Lifesize, Stone+ (a separate brand), PhotoDisc, The Image Bank and Photographer’s Choice would be shown. The image shown for each one of these other collections may be the best seller within that collection, but not necessarily the best seller on the site. None of the other images shown may have sold as well as the image from the Stone collection.
Hypothetically, here’s what might have happened.
| |
Num. Times Image Licensed |
| Digital Vision |
25 |
| Lifesize |
3 |
| Stone+ |
2 |
| TIB |
8 |
| Photographer's Choice |
0 |
| Digital Vision, 2nd Best |
10 |
| Stone |
100 |
Thus, while the images ahead of it may not have sold as well the best selling Stone image will be in the 8th position. Note that the second best Digital Vision image will be seen before the first Stone image. The second Stone images will appear at slot 12 and the third at slot 26. At total of 18 Stone images will be shown in the first 300.
With this system the photographer’s image that is the best selling in a particular category in a major brand it may get shown fairly high up. But if it is only the third or fourth best selling image in most of the brands then its chances of ever being seen are not that good. Sales of the best selling images are likely to get pushed higher since these are the images customers see and the spread between the first and those that have sold a few times or never will get larger and larger.
Also, this doesn’t account for how new images that have never sold will be shown. We know from looking at search returns that in some cases brand new images are shown near the top. Thus, the second image shown may be the newest image uploaded to the site in that brand. Of course, in any particular brand if none of the images that relate to the keyword requested have ever sold then the search engine will probably deliver the newest image uploaded to the collection first.
This would explain how images that one would think are seldom if ever used manage to make it to near the top of the search return order. Anyone doing searches on the Getty site must agree that the first 20 or so images are not all best sellers. Usually, there are a few classic shots and many that only very narrow groups of customers might consider.
This is also a strong argument for the microstock strategy of letting customers choose the order in which they want their search returns delivered. The default search order on iStockphoto is “Best Match”. With this search iStock determines the weighting each image should get in much the same way as Getty determines which brand gets shown first. But, iStock also gives customers a choice of searching by “File Age” (newest first) or “Downloads” (number of times licensed). A large percentage of their customers tend to search by downloads because this enables them to quickly review images that have been popular with other customers. If they are interested in the newest images they can search by Age and if they want to see what iStock wants to sell them they can go to Best Match. In addition, customers can easily review the entire collection of a particular photographer. The returns delivered from these various search options are dramatically different from one another.