Industry Comment #20--The Dirty Secret of Ranking

Over 200,000 of the more than one million images in my stock photo agency’s library contain the keyword “people.” Have you ever wondered how we decide what image from those 200,000, to show first?

Of course, the first element in ranking generally is the quality of the match between the search string a user entered and the metadata the stock photo agency has added to its images. There are many tricks and twists available when a user has typed in a long search string. But, because 57% of all searches done on our site contain only one word, we need to use other tricks to put our images into order.

Does image rank matter? It is hard to say. There are many variables associated with each image licensing decision. However, in certain categories such as royalty-free lifestyle images, there are many competing suppliers with similar product lines. In these “commoditized” subject areas, it seems likely that a hierarchy similar to search engine ranking may exist. According to a Marketing Sherpa study, the top three lines on a text Web search get 60% of all clicks. Pushing a set of images twenty or thirty pages down in a search result probably would cut their license fees by 50% or more.

There are several ways that stock agents could rank images:

1. Random. I know of one agent who uses random ranking. I find it extremely annoying. I don’t search linearly for images, but recursively. In other words, I may try the same word several times, as I wander around an image site. On the randomly-sorted site, if I see an image I like, but neglect to put it into my lightbox, I may never find it again!
2. Newest image first. Most of the celebrity and news stock photo agencies work this way (see Newscom for a well-organized and clean example). Whatever has come in most recently, hits the top of the list. This makes good sense for news images, but doesn’t seem suited for general purpose stock photography.
3. Best image first. Doesn’t this sound like a great approach? Make the first image the one that is the most creative, most artistic, and most interesting image that matches a certain search term. Of course, the problem lies in deciding which image is the most beautiful or marketable. I know of two approaches:
   1. Human editors pick the top images. This approach gives consistent results, but requires “tuning” of the editor’s taste to match the taste of stock photo customers. Without this correction factor, the editors can get away from current market trends, and develop a view that misses the reality of their market.
   2. Customers “vote” for top images. This approach purports to give power to end users to express their views. However, it can lead to self-perpetuation (an image that is “voted up” will stay highly ranked, because its higher starting position gets it more viewing opportunities) and to style lag (what was popular yesterday remains popular for too long).
4. Most profitable image first. This it the “dirty secret” I mention in the title of this blog. Major agencies are starting to rank images based on how much of the revenue they keep from each license. That means that an agency that has created and owns its own images will put those images first (and keep 100% of all licenses). It will then follow its “owned” images with the group of images on which it has negotiated the lowest commission payout. It will show last the images it makes the least on. This policy is not enunciated on the Web sites of these agents. Some of them make it especially hard to spot, by hiding the names of the party who contributed the image, deep in their site.
5. Sell the top ranking. Some search engines are starting to charge extra fees, if you want your link to show up in the “preferred” area. I don’t know of any stock agents that have tried to extract this type of fee from their image contributors—yet!  I doubt that it would fit our industry association code of ethics, but a lot of the new players in the stock industry aren’t PACA members.

Which approach do we take, here at Index Stock? Option 3a, of course! While it is more work for us, we feel a human-based ranking gives our customers the best possible search result. Our approach favors those artists who give us their best images. That seems fair, and is easily understood. To keep things simple, we do not tell our salespeople what our profit margin is on an image. Their goal is to satisfy a customer and maximize the revenue we generate for our artists, not to maximize our gross profit.

While we endorse Option 3a, Option 4, the “most profitable first” approach, seems to be sweeping the industry. Recently, several of our distributors approached us and threatened to “demote” our images, unless we cut our commission percentage. For about fifteen years, the “standard” payout from a stock agency to a distributor had been 40%. Now, distributors are now demanding 50% or even 60%, with image ranking as their lever.

A change in ranking could easily cut the license revenue from a distributor by 50%. Therefore, it could seem rational for a stock agent to agree to a cut in payout from 60% to 50% of gross revenue (a 17% drop), instead. However, accepting this cut might invite a further cut later. Further, if the distributor is investing heavily in shooting and promoting its own material, it is likely to push up these images to the top. Then, its agent partners will see even fewer revenue opportunities.

There is a risk for our distributors, from this strategy. If our percentage gets too low, we will eventually make more total revenue for our artists by going around our distributors, and marketing directly to their customers. I estimate that the “break point” for this strategy is somewhere between a 50-50 split and a 60-40 split in favor of the distributor. We hope our partners will understand this, and not push us to the point where it makes more sense to buy email lists and do our own pricing and marketing, instead of relying on them.

Comments

[Robin Liao]

Hi Bahar,

This is a very educational read. As an image provider that rely on distributors (including Indexstock), we never would have thought of this issue if it wasn't for your post. Thanks for sharing.

Robin Liao

[Jeff Greenberg]

It is possible to offer all ranking options to buyers by having drop down "ranking" menu that allows buyers to control type of ranking. Why isn't THIS strategy sweeping stock industry?

[Bahar Gidwani]

Two reasons it is not sweeping the industry:

1. Putting the images an agent owns (if you own images, which we don't) first, makes more money for the agent. Greed conquers all!
2. Customers are not very interested in search complexity. I've seen all kinds of special search features come and go. 99.9% of customers just want to type in a word or two, and then page through the images.

We added what I thought was an interesting option to www.indexstock.com. We allowed users to decide if they wanted the default operator in searches to be "AND" or "OR." So, one person would type in "child with toy" and see ONLY images of children AND toys. The other would see all of our children and toy images, then, all of our children images, then all of our toy images. (Make sense?) The ten percent who notice it, seem to like it. The rest just take whatever our search gives them, and trust the "we know best."

Bahar

[Sean]

Perfect