A lot of attention is being given to finding a better way to search for photos. Those who believe technology can solve all the world’s problems are trying to build algorithms that will instantly find exactly the right image to meet the needs of each paying customer.
With 1.8 billion photos being uploaded to the web each day and even professional sites like Shutterstock uploading more than 260,000 new photos each week there are more good pictures on any given subject than any professional user has time to look at.
Keywording
Among professional photographers and stock agents there is a big emphasis on better keywording, but with many common subjects keywords alone don’t seem to be the answer. On many sites there are limits to the number of keywords they will accept. On the other hand using lots of keywords often causes images that are marginally relevant to be shown to buyers that are looking for something totally different.
If the keyworder (usually the photographer) leaves off the word a particular buyer happens to use when searching for images, the image that may have been perfect will never be seen.
Knowing what words buyers and using to search for images, and how frequently, would be helpful but that information is never shared with the people doing the keywording. Do the people in Germany, France and Japan use different words than those used in the U.S. when they are searching for images? Probably, but those words are never shared with keyworders.
We assume Shutterstock, Getty, iStock and others know exactly which words are being used in which countries, and how often, but that data is closely held and not shared.
In addition, buyers often don’t know exactly what they are looking for until they see it. They have trouble describing what they are looking for in enough detail, to call up images they have previously seen. Let alone describing something they have never seen.
Visual Search
Visual search that compares pixels works if the customer has a copy of the image she is seeking. But, most buyers are looking for something they haven’t seen before.
Most Popular
Many professional sites offer, and customers tend to use, “Most Popular” or “Most Downloads.” The big advantage is that the searcher gets the benefit of seeing what other paying customers looking for the same subject matter found most useful after going through a lot of images on the site.
The disadvantage is that that they will be using the same image many other customers have used, but for many buyers that doesn’t seem to be a problem.
This works better for microstock and subscription sites that have lots of downloads. It doesn’t work as well for traditional sites (and none of them offer this search option) because they make many fewer sales (at higher prices) and very few of their images tend to be used multiple times.
Likes
Social media is focused on “Likes” as a way to sort out the best from the also-rans. From a professional point using Likes have a few flaws.
A few weeks ago, Aditya Khosla and two of his colleagues from the MIT released a study that tried to predict the
popularity of photos.
They set out to determine if it is possible to predict the popularity of a photo by looking at its objective aspects like color, texture, gradient and photographer. They tested their algorithm on 1.4 million shots taken by 100 photographers. They were right about one-third of the time. Thus, asking a blind person to choose photos would yield better results.
When they added tags (keywords) to the above elements 40% of the most popular images were selected. They then went on to check views and found that the number of views can predict if an image will be liked 62% of the time. (But likes are often an instant reaction shortly after the image is first posted. What if the number of images returned in a search is so large that most of them never get looked at and thus get no views? This becomes self defeating and an algorithm based on likes makes it unlikely that images that don’t get a few likes quickly will ever get any likes.)
Altogether, social and contextual cues can predict the popularity of a photo 2 times out of 3. When they added image features with social context, that number goes up to 72%.
The researchers found a strong correlation between the size of a photographer’s social network and the number of “Likes” they received. The concluded that getting more visibility might be the key to making your images more popular.
Do Likes Have Anything To Do With Sales?
The corollary is that in order to make more sales you need to build a larger social network and get your photos seen by more people.
But, does that mean more people will pay to use a well-liked photo, or just enjoy it for free. The
Stolen Scream is a good example of what happens to a well-liked photo.
If your goal is to communicate with friends about what you’ve been doing “Likes” may be important, but if your goal is to earn money they could be worse than irrelevant.
The first big question is who is “liking” your photos. Are they people who actually pay from time to time to use photos, or just friends? Are your friends clicking “Like” because they want to encourage you as a friend, they admire your creativity, the photo is something they wish they had taken themselves, or they would have liked to be there with you when you took the picture?
If you simply want affirmation that the images you are producing are “good” or “great” in the eyes of your friends then likes are important. The larger your social network the greater the likelihood that you will generate more likes.
However, if your goal is to get someone to pay for your pictures then likes may be of little value. They could encourage you to shoot more of the kind of thing no one wants to buy, but they will make your friends happy.