Are your Getty Images sales declining? It may have nothing to do with the quality of your images, the subjects you shoot or your keywording. It could be that customers rarely, if ever, get a chance to see your images.
Currently
Gettyimage.com has 4,278,804 RM and 6,034,642 RF images on the site for a total of 10,313,446 in the creative section of the site. Getty has images from 103 different RM collections and 98 RF collections.
I recently conducted some searches to try to determine how Getty orders image returns. I chose the keyword “People” and counted the number of thunbnails delivered in each group of 100 images up to a total of 500 thumbnail returns. I picked this keyword thinking that virtually every brand would have some images with people in them. Thus, I thought that by the time I got through 500 images I would have seen at least one image from each brand.
It is my belief that most customers will not look through more than 500 images before they either find something to use, use new search terms, or go somewhere else. In the past Getty has said that most customers don’t look at more than 3 pages before they change their search parameters.
Royalty Free
I started with the RF images. Only 33 of the 98 brands represented had at least 1 image shown in the first 500. But 89% of the images shown were from only 12 of those brands. The number one brand with 14% of the images shown was E+. The images in this collection are pulled from the iStock Signature+ collection.
E+ does make up a significant percentage of the Getty Images RF collection. There are 982,725 E+ images representing 16% of all the RF images on www.gettyimages.com. In fact if we total the E+ images, and the Vetta and iStock Vectors collection they make up 24.4% of Getty’s total RF collection. It is interesting that starting next week customers will be able to purchase any of these images on iStock for around $30, but they will pay a lot more if they buy them on gettyimages.com.
| Getty RF |
|
|
|
|
|
Total |
| |
1-100 |
101-200 |
201-300 |
301-400 |
401-500 |
500 |
| E+ |
10 |
18 |
20 |
15 |
13 |
76 |
| Moment |
11 |
7 |
12 |
11 |
13 |
54 |
| OJO |
13 |
11 |
5 |
10 |
12 |
51 |
| Calaimage |
8 |
5 |
8 |
10 |
10 |
41 |
| Digital Vision |
11 |
10 |
6 |
7 |
4 |
38 |
| BrandX |
8 |
12 |
4 |
6 |
7 |
37 |
| iStock Vectors |
6 |
6 |
11 |
2 |
5 |
30 |
| Vetta |
13 |
6 |
2 |
2 |
4 |
27 |
| Image Source |
5 |
3 |
3 |
8 |
8 |
27 |
| Photdisc |
3 |
5 |
10 |
5 |
4 |
27 |
| Blend |
4 |
5 |
1 |
5 |
5 |
20 |
| Cultura |
3 |
1 |
1 |
5 |
3 |
13 |
| Photographers Choice RF |
|
3 |
5 |
|
|
8 |
| Stockbyte |
1 |
1 |
2 |
1 |
3 |
8 |
| Hero |
1 |
1 |
|
3 |
2 |
7 |
| Blue Jean |
|
2 |
1 |
1 |
1 |
5 |
| Tetra |
1 |
1 |
1 |
1 |
|
4 |
| Purestock |
|
1 |
1 |
2 |
|
4 |
| PhotoAlto |
|
|
3 |
1 |
|
4 |
| Bloomimage |
|
2 |
|
|
|
2 |
| Collection Mix Subjects |
1 |
|
1 |
|
|
2 |
| amana |
|
|
1 |
|
1 |
2 |
| Onoky |
|
|
|
1 |
1 |
2 |
| National Geographic |
|
|
|
|
2 |
2 |
| Fuse |
1 |
|
|
|
|
1 |
| imagezoo |
|
|
1 |
|
|
1 |
| Doring Kindersley |
|
|
1 |
|
|
1 |
| Maskot |
|
|
|
1 |
|
1 |
| RooM |
|
|
|
1 |
|
1 |
| Design Pics |
|
|
|
1 |
|
1 |
| Viewstock |
|
|
|
1 |
|
1 |
| Rubberball |
|
|
|
|
1 |
1 |
| ABSODELS |
|
|
|
|
1 |
1 |
| |
|
|
|
|
|
|
| |
100 |
100 |
100 |
100 |
100 |
500 |
On the other hand there are a significant percentage of other brands that never get to show any of their images near the top of the search return order.
Of course customers do much more focused searches than searching for the word people. But on nearly all searches the images from the various brands are displayed in relatively the same order. For example here are the brands displayed, in order, for the first 10 images with 4 different searches.
People - BrandX, Image Source, Moment, BrandX, Blend, Calaimage, Moment, Moment, Image Source, OJO
Trees - Moment, BrandX, Moment, Moment, OJO, Moment, Vetta, OJO, Tetra, Vetta
Computer – BrandX, Moment, Moment, OJO, Digital Vision, Digital Vision, Digital Vision, BrandX, BrandX, Vetta
Family – Calaimage, Calaimage, Blend, Moment, Moment, Blend, Calaimage, Hero Images. Calaimage, Moment (with this search the first BrandX picture comes up as number 15)?
The order is not exactly the same, but there tends to be a certain consistency. Getty seems to go through the brands in a more or less a consistent order and a consistent frequency. If the brand has no images with a particular keyword then it is skipped and they go to the next brand in the order list, but there are very few brands that don’t have at least some images with the keywords customers most often use. One exception I notices was in a search for “business.” With that search 10 of the first 12 images were Stone images, but I didn’t see images from other brands that are not on the basic short list.
The Getty algorithm may give some weight to the number of images with the keyword in a particular brands collection, or the percentage of images with the keyword, but that doesn’t seem to always follow. Even when the search is for the keyword “disability” images from DisabilityImages that specializes in this subject matter don’t seem to show up anywhere near the top of the search return. The same is true for the Foodcollection images when searching for “food.”
In addition to keywords the algorithm may give added weight to whether the search term is actually a word used in the caption, not just as a keyword.
Rights Managed
When I turned to the RM collection using the keyword “People” the differences were more stark. 96% of the images found in the first 500 were from 6 of the 103 brands in the RM collection. They are: Stone, Iconica, Taxi and Photonica, all Getty house brands, plus Moment (formerly Flickr) and Oxford Scientific.
These six brands have a combined total of 354,171 images and represent 8% of Getty’s total RM collection and yet 96% of images shown to customers come from these collections.
| Getty RM |
|
|
|
|
|
Total |
| |
1-100 |
101-200 |
201-300 |
301-400 |
401-500 |
500 |
| Oxford Scientific |
48 |
31 |
32 |
47 |
11 |
169 |
| Stone |
24 |
34 |
28 |
23 |
50 |
159 |
| Moment |
7 |
7 |
13 |
8 |
14 |
49 |
| Iconica |
7 |
6 |
13 |
8 |
11 |
45 |
| Photonica |
6 |
6 |
7 |
5 |
6 |
30 |
| Taxi |
5 |
8 |
3 |
6 |
6 |
28 |
| Riser |
2 |
3 |
3 |
3 |
2 |
13 |
| The Image Bank |
|
3 |
|
|
|
3 |
| Blend |
|
2 |
|
|
|
2 |
| Photographers Choice RM |
|
|
1 |
|
|
1 |
| Cavan Images |
1 |
|
|
|
|
1 |
| AGE Fotostock |
|
|
|
1 |
|
1 |
| |
|
|
|
|
|
|
| |
100 |
100 |
100 |
100 |
100 |
500 |
The real shocker is Oxford Scientific (OS). One-third of the images shown in the first 500 when I searched for “people” were Oxford Scientific images. OS has a total of 60,435 images or a little over 1% of the total RM images on gettyimages.com. Oxford Scientific certainly has some great images, but it is hard to explain why such a high proportion of their images should be shown ahead of all the others when they represent such a small percent of the total collection.
Moreover two of their strengths are underwater images and thermal images so a huge percentage of the images shown when you do a search for “people” are people swimming underwater with various marine life and thermal images of various human beings.
Given the dominance of OS there were an amazingly small number of images in the first 500 that showed people in a business (non-medical or scientific) environment, or that showed family life. I would think that customers who search for “people” might have more interest in images of people in business and family situations that of thermal and underwater images. Also, the people in a lot of the OS images shown were a small insignificant part of the image.
Less you think this ordering only happens with a search for people, searches for other subject often bring up an Oxford Scientific image as the first image shown. Here are a few of the words I used individually in searches.
Family, swimming, office, working, satellite dish, Africa, ugliness, home, technology, computer, couple, friends, love, wedding, bride
Some weight may be given to images that have been licensed previously, but that doesn’t seem to be the case.
Another factor to consider is that once the algorithm chooses a brand to pull an image from, it seems to always go to the “newest image added to that collection that hasn’t already been shown.” Thus, suppose that a photographer has just added a series of six of seven very similar images from the same shoot to a collection like Stone or even Photonica. All those images will appear somewhere in the first 100 images shown. The algorithm appears not to be able to say, “I’ve already shown an image from that shoot so I will skip over all similar images to the next image in line that is significantly different.”
At first glance this may seem to benefit the photographer who has just added multiple versions of the same shot. But it is a very short-lived benefit. As other photographers add new images to the collection all of the first photographer’s images get pushed down together until they are below the level where anyone gets a chance to see any of them. If a photographer has seven or eight very similar images from the same shoot and with the same keywords that he wants to upload he might be better advised to submit one a month rather than submitting then all at once. That way, they would be at several levels in the pile of newest images, and have a better chance of at least one of them being seen over a longer period of time.
What Can Photographers Do To Avoid These Problems?
If you can uncover uncommon keywords that customers use regularly, but other image creators seldom attach to their images there is a chance your images will get near the top of the search return order for a longer period of time. However, the likelihood of finding such words for the subjects in greatest demand is very slim.
Some of the brands whose images appear to be buried might want to consider setting up a new distribution network that uses an algorithm with a more systematic and fair approach to showing images from all brands represented on a more equitable basis. They might choose one image from every brand represented that has anything with the particular keyword before showing the second image from any brand. They might rotate the order of the brands on a regular basis to give those who start out at the bottom of the list an opportunity to work their way to the top.
The brand whose images are shown first might be based on the percentage of images in that brand’s collection with the keyword used. In this way brands with specialties, but with a relatively small number of total images, might get their images seen ahead of those with a marginal collection of the subject matter.
Another thing to consider is a system where a brand is allowed to prioritize a certain percentage of the images in its collection that it believes should be shown first above all its other images. In this way the brand could surface older images that have either sold well or have some other characteristics that make them more likely to better fulfill customers needs than some of the newer images the brand has added to its collection. This would require more work on the part of brands, but it would likely present a much more enticing offering to customers and encourage them to use the site.