Dan Heller's Photography Business Blog Industry analysis from www.danheller.com

The photography world -- the business, the culture, the art, the politics, the technology.

Site Feed

Subscribe to
Posts [Atom]

My Photo
Location: Santa Cruz, California, United States
My Books on the
Photography Business

Tuesday, May 29, 2007

The Virtual Stock Agency

So... where were we?
In my last post, I mentioned www.smugmug.com will soon allow members to license their images as stock. What makes smugmug unique (or did, at the time) was to let users set their own prices for images, and the site itself setting no boundaries, high or low, on pricing.

While I have always advocated user-specified pricing, there's a wrinkle. While "crowdsourcing" is a great way to build content and traffic, it doesn't work so well when you try to monetize that content. Or, more specifically, when you allow the members of the community to monetize their own content. This is a subtle, but important difference. That is, contributors to social networks add to the overall value of the system because the incentive to participate is found in the reward of getting praise from the rest of the community, or by benefiting from the advice or assistance from others. While there are those who may act as spoilers, their net effect is usually benign to the total value of the community, unless the number of spoilers reaches a critical mass, then chances are that the site is either administered poorly, it's of too little interest to be successful anyway.

Suffice to say, this generally isn't the case with photo-sharing sites.

The question is, what happens to user behaviors once you introduce financial incentive for them to monetize their own content? My previous article presented the Prisoner's Dilemma experiment ("PD"), which states that people will service their own needs first, even at the expense of the other members. This is the basic principle of the Prisoner's Dilemma, and has been the inspiration for several winners of the Nobel prize for economics. As such, it has been tested repeatedly since its inception in 1950. It explains many aspects of society and economics, such as why negative advertising works, why negative political campaigns work, and so on.

What this means for business is that if the conditions exist where participants benefit by bringing others down, the entire business suffers. In short, the very thing that made a social network successful is the very thing that would doom its ability to monetize the very same content.

This translates directly to user-specific pricing in this way: the weakest of all the photographers who would not otherwise sell any images if prices were all equal, will lower their prices in hopes of getting some sales. But, it's because of the closed, insular environment where images are placed side-by-side that the domino effect takes place: each player lowers their prices in sympathy with others for fear of being left out.

But, let's be clear--this type of phenomenon only happens in contexts where products are placed very closely together in the same contextual space. For example, two gas stations across the street from one another invariably sell gas at exactly the same price, even if it's two different oil companies because the contextual space is the same: buyers will go to one or the other based solely on price. Go a mile away, and you may see different prices, even from the same gas station brands. The same is true for buying cans of soda in the corner store, where things are more expensive, than in a major chain store, where they tend to be less expensive. If the buyer were truly economically-minded, no one would ever buy at the higher-priced stations, or from the corner market. As we all know by observation, if not our own experiences, this doesn't happen. And this is because of "economic lensing:" people's perceptions of price comparisons to identical products is limited to small contextual spheres. Outside of that sphere, perception of value is governed by more than just price; many other factors come into play. For example, convenience, time, and so on.

In the photo world, the "sphere" I'm talking about is a microstock site, and the perception of a buyer considering photos of very similar subjects is that the prices should be very similar, if not identical. Therefore, the lowest prices will, eventually, cause all others to drop.

This isn't new to most photographers--most believe this is the case in the broader industry as well. But, as exemplified by the gas station example (and numerous others), this principle doesn't extend beyond the small, tightly defined sphere. Buyers will accept different prices of the same products from different websites, just not from the same website. The fact that most microstocks sell photos for $1 each (plus some tiers that are virtually identical) is not because this is what the market demands, it's the psychological immaturity of the market. Microstock is still relatively new, and inexperienced players fear that if they don't meet everyone else's price, no one will come to their site. This miscalculation is not rational, and economics has proven otherwise. But this is beyond the scope of this discussion. (For a detailed discussion on this, see my blog entry on this page.)

It is because of this fundamental problem that I've always predicted the microstock business model will eventually evolve into something else. What that is must at least satisfy this fundamental law of economics: remove the incentive for members to harm one another. To build a viable business model, we must understand the ways members harm one another, or that erode broader quality or integrity of the site. In other words, what are the ways that photographers can behave badly? Once we identify these, we can then devise ways to deal with it.

The Scope of the Problem
Members harm the system in two ways: by artificially pumping up the perceived value of their own photos, and by artificially harming others opportunities. In the first case, one way photographers "cheat" the system by using a technique I've discussed before: "keyword pollution". They do this by assigning keywords to photos using terms that artificially cause their images come up in search results, but in so doing, cheapen the quality (relevancy) of those very searches. (This, in turns, makes the site less desirable for buyers to use.) For example, using the keyword "sexy" on a photo of a women, even though she may not be, is a common practice. Stuffing mass quantities of words into an image's keyword list is one of the most widely used methods for gaining visibility on a site where it's too hard to get noticed otherwise. This is discussed at length in my article titled, "keywording's role in the future of stock photography".

In the second case, photographers can harm other users by dragging down the "votes" of other players. The incentive to do this is due to the fact that image searches may sort results by more than just keywords, but also by "popularity." Regardless of whether popular photos actually sell better, there is now financial incentive to drag down others' image ratings so as to allow their own images to rise higher in rank. (They'll also fabricate higher rankings for themselves by creating false accounts, or by other means.)

I've spoken with many microstock agencies in one form or another over the past several weeks about a variety of topics, but for this topic, all claim this isn't happening. Since it's impossible to really measure whether image "popularity" is being manipulated, one uses indirect methods by way of inference techniques. That is, by looking at how photographers behave in other ways, such as "keywording," we can infer whether users employ counter-productive practices. If photographers are polluting their images' keyword lists, it's likely they're artificially nudging up photo popularity.

Fotolia is not any different than the others I've tested, but their site is easier to demonstrate because they publish their master keyword list in a central location. Looking at the first page of "A"s, you can see that it's inundated with bad keywords that range from misspellings to the use of entire phrases, complete with sentence structure. (Yes, these were added as keywords.) While sample searches for many commonly used terms yield results that range from "generally good" to "pretty poor," it suggests that the system needs finer control over its users' behaviors.

Then there's the question of why no one fixes this problem internally within a site. Again, all sites I've spoken to claim that they have in-house staff to police this activity, plus member-volunteers from around the world who also act as moderators. But, as evidenced by looking at their image's keyword lists, this isn't working well enough. This leads to the question of whether it could ever work. Many of these "staff" people are the same member-competitors as those they're policing, so it's in their interests not to work in the best interests of the system. There's incentive for them to permit the misspellings and poor phraseology in keywords, plus an incentive to reject keywords that may actually be appropriate for images. If you're just a volunteer, who cares? You have no financial incentive to work with the system. And besides, who's overseeing the volunteers?

The net result for stock agencies that allow contributors to control how their images are keyworded, ranked, and now priced, is that they are providing financial incentives to their members to "not cooperate." Those who are least likely to yield any sales will be those most compelled to "cheat" the system in order to get "any" sales. And this trickle effect infects the entire contextual scope.

Why isn't this happening rampantly? Because the industry is still too new. In the early days of search engines, it took a few years before people artifically stuffed their web pages' "keywords" list with bad data, but once it was clear that it worked and word spread, it was so widley used that search engines no longer look for metadata "keywords" in web pages. Then came link exchanges. It took a few years for people to figure out that this was a good way to artificially raise your site's rank, but once it happened, search engines had to be more intelligent about scientific in their analysis of whether any given links are really more part of a link-exchange program. (And if they do, the sites involved actually get a lower ranking.) So, the fact that microstock or photo-sharing sites might not yet be seeing widespread, rampant manipulation of keywords or photo ratings, this is solely an artifact of the business model being so new. Once people figure out there's money to be made, everyone will do it.

So, where does that leave us? Is user-generated pricing a bad idea? Not quite. I think it's a fine idea; it's just being applied to the wrong business model, which happens to be one that's used by all photo-sharing and microstock sites.

(I am sympathetic to the problem in this respect: what can you expect from a company with vast quantities of images uploaded on a daily basis, with pricing margins too thin to hire a real staff that can not only handle the volume, but do it well? Tough challenge. All the more reason why I feel the business model is not economically viable over the long term.)

Large Agencies
On the other side of the street, we have the larger stock agencies, who rightly claim they don't suffer from any of these problems because photographers don't have any mechanism to dupe the system. They only use in-house editorial staff to not only manage the entire keywording and ranking process, but they also cherry-pick photographers and photos. And, of course, they set their own pricing. The only influence the photographer has is his initial stab at keywording, which is tightly managed after submission. (Cheating is futile.) For the photo-buying client, he gets a consistent set of images and themes that the company stands behind.

Yet, as evidenced by their struggle to compete in the open market, this well-intentioned strategy is insufficient. With the exception of specialized market segments or photo collections, the bulk of their inventory is not getting the kind of visibility, distribution, marketing, or pricing appropriate for today's market. (Acquiring other stock agencies doesn't help this problem.) And because they only have just so many people on staff, they limit the photographers and their photos by denying entry to many otherwise-qualified candidates. They, in turn have no other outlet than to go to the microstock sites that have essentially become the major thorn in the sides of the large agencies.

Yet, the business model of a large agency is not inherently broken; it just suffers from two major problems. It cannot be scaled up to what is required for today's economic conditions, but more importantly, they don't have the one and only advantage that the microstocks have: the grassroots swell of crowdsourcing. They not only miss the opportunity of having a much larger cache of photos to sell, but they also lose the social-networking advantage that attracts both users and buyers. Only one datapoint is necessary to prove this: visitor traffic. Flickr's is higher by more than than an order of magnitude over Getty's, and Corbis is hardly in the running. (Jupiter Images is barely even on the chart.) (For fun, my site, www.danheller.com, is just behind Jupiter by only a fraction, and my graph crosses over every once in a while. :-)

There's also the question of whether a centralized editorial staff is sufficient in this day and age anyway. Essentially, it's still "one vision." And buyers are more than just small pockets of the media world and large corporations; they are everyone, everywhere. More than just one particular (albeit consistent) marketing partner is needed.

So, the state of affairs today is that we have two business models for stock agency sales, neither of which work very well in today's economic climate: one is exemplified by larger stock agencies, where all control is centralized. The other is the microstocks, where there is no organized control, and everything is left to the good will of the individual contributors, which is economically doomed, as predicted all measures of how economic models of human behavior works.

The Virtual Stock Agency
I have an idea for solving this problem: the virtual stock agency. I envision this as being a hybrid of both the larger stock agencies, and the microstock sites. In essence, I see it as a mechanism by which an editorial staff that's normally found within a stock agency can do the same work from inside of a microstock site. They would use the site's existing photo asset base, and sales would be done using exactly the same mechanism in place today. But the key here is that this "staff" is not employed by the microstock agency, but rather, is a highly-incentivized class of user that's already a member of the community.

This new class of user is what I call, The Editor. Unlike existing users, who are currently photographers competing against one another, The Editor is different in that his financial incentive is to make any image sell, not just his own. In fact, their job function would be almost identical to that of an in-house marketing staff at a larger agency, but scaled down to a micro model. It's the cumulation of many such individuals that would comprise of a larger, more effective team. In essence, it's crowdsourcing the marketing and sales of images, which is quite different than just allowing the photographers to upload images, and hope that buyers show up.

The more editors, the better. And by giving this class of user financial incentive, it removes the incentive for photographers in the older (existing) model to work against the system in an effort to boost their own sales. While the photographer can still sell his own work, the new virtual agency model provides more incentive for photographers to work with Editors--as many as possible--who, in turn, will do more (and better) work to sell all images they can. As with any business, the smart ones will rise and the weaker ones will drop off. We'll get to the details of that later.

To understand why this class of user solves the problem, I cite an extremely simplistic and non-economic example of it in a feature of Amazon.com called "Listmania!" It lets registered users create their own music selections and ratings or reviews. If I search for "Ronnie james Dio" on amazon.com's music section (this choice inspired by the fact that my iTunes player happened to be playing a song by him at the moment), I can opt for two kinds of results: those from a generic "amazon" search, or by going to lists created by users that match this criteria. In the first case, Amazon's list is just a discography in chronological order. But, on the "list" side, there are several choices by users who've made their own selections. I chose a list by a guy named (I'm not making this up) "Lebronze James Dio." (Amazon lists stats about this member, including his "reviewer rank" and other data.) He's got some really, really good picks here, including an interesting collection of heavy metal songs sung by none other than Pat Boone!

In the future, I would probably use this guy's recommendations as a guide for picking albums, because he's informed (more so than me, and probably to the point of being a nut), and his choices allow me to find something that I might not have found on my own. Even if I don't know what I want in advance, his ideas helped me. What's the chance I would have found the Pat Boone record? And what's the chance that this record would be rated or reviewed highly by others? Very low. Yet, this _one_ guy liked it, and because my interests suggest that this record might be something I would like too is a value that supersedes whatever the "crowdsourced" opinion (rating) is.

So, let's start with this common scenario on a microstock or photo-sharing website: search for images that have the keyword "romantic." This is a conceptual keyword that, currently, relies on individual photographers to have added to their images' keyword lists. But, photographers being what they are, some of them may not think to use the word for an image that should have it, while others may apply the word to images that shouldn't have it, in hopes of having that image come up in search results. As such, you see what the site returns as photos that match that keyword, the quality of which may vary.

Now imagine along with those results, the user is also given a list of "featured editors." Each link takes you to that editor's own section of the site, where he has put together what he believes to be good candidates for this keyword. An editor may even choose images that weren't originally keyworded with "romantic." Or, if the photographer used that keyword, but shouldn't have (at least, in this Editor's opinion), he can remove it. He can even correct misspellings of keywords. It's as though he has full editorial control over all the images in the entire microstock site. It's not just limited to keywording, too--it may include image ranking, editorializing, captioning, even translating to other languages or locales. In essence, the Editor is acting exactly like a virtual stock agency. "Virtual" because he doesn't acquire images from photographers, or even actively sells them. He simply prepares, packages and promotes images as "lists", collections, or whatever he likes, exactly as a traditional stock agency would.

What's important to note about this model is that this user's edits are entirely contained within his section of the site. Think of the Las Vegas tagline: "what goes on there, stays there." Whatever an Editor adds, modifies, or deletes about an image is not exported to the image itself, or to anyone else's list, or even to the site itself. It's his own dataset that sits on top of the general dataset on the site. When visitors enter into an Editor's area, they get images that the Editor would have them see, how he would like them to be seen, or any other information he chooses. It's as though he owns all the photos.

Why would someone bother to sign up to be an Editor to do all this work? Money. Editors (a.k.a., virtual agencies) earn commissions on image sales. The more images they keyword and "package" in a fashion that attracts buyers, the more potential for financial return. It's the unique and individual talents of this editor that can really begin to move images. And, as I mentioned at the beginning of this article, "any" image can sell, provided it's packaged properly and sold to an appropriate audience. (I also discuss this in greater detail in this chapter of my first book.)

The business relationship between the parties is unidirectional: to the site itself, not to one another. At least, not formally. Once an image is purchased, the photographer gets his commission, and the Editor gets his too, each from the hosting site. Neither the photographer nor the Editor need to ever know about each other, let alone work together. (They can if they want to, but they don't have to. I get into more of this later.) Because there's money involved--potentially, a lot of money--people will invest the time and intellectual capital by being Editors. Again, this is no different than someone starting a real photo agency.

Still skeptical? Think about how many people contribute to Wikipedia with absolutely no financial incentive. And what makes it even better than wiki, is that Editor-class users are not hampered by other editors or photographers. Their edits are entirely theirs, giving even more incentive to produce. Similarly, Flickr has many users who tag all their "favorites" (pictures they like) from other members' photo streams. Consider how much more they would do this if they could make money from this effort.

Alert readers may note that Flickr has programming interfaces that can be used to access its photo repository and present information in many ways. www.picturesandbox.com is one such example. You can't buy or license the pictures, but one could see how, if such a thing were permitted, that one could essentially build a virtual stock site using this method to access image databases. What would really keep it from working, however, is the glaring problem I've been talking about for months now: flickr has made no suggestion that they are going to get into the photo licensing business.

Unlike Flickr, existing microstocks do have those business relationships established with its members, requiring no fundamental alteration to its legal obligations. Indeed, very little is actually necessary to build a virtual agency from within any existing microstock site: a policy decision to do so.

Let's get back to the basic point: the intent of my proposed model is not only provide an economic incentive for members to continue to participate in the social network, but to do so in a way that does not provide a mechanism for users to harm or interfere with the system. Indeed, because of how the financial incentives are structured, Editors that try to cheat won't get much buyer traffic anyway, making their effort irrelevant. Since none of their actions are seen or have any effect outside of their own sections, the effects of spoilers are virtually nil. This makes the landscape fertile for genuine "intellectual property," which itself is the ultimate gold mine for any company.

Evolution and Growth
Over time, as better editors emerge, the value of their selections may eventually mimic (on some scale) the patterns of those by the "editorial staff" at larger stock agencies. As the number of Editors grow further still, they will naturally segment into hierarchies and specialties, which is what happens with all economic and social networks (as any sociologist will tell you). There would naturally be specialty editors for images on all topics, ranging from cars, fashion, glamor, corporate stock, sports, travel, lifestyles, and you-name-it. This is exactly what happened within the Flickr community, and virtually all others. That the virtual agency model involves financial incentives, this would only increase the rate at which all these phenomenons happen.

At some point, buyers who are avid Getty customers may ask, why search for images from Getty with their singular staff of talented people, when a microstock site with a much larger army of similarly-talented editors can be used?

Barriers to Entry
How hard is it to build such a system? From the technology site, it's nothing more than expanding the existing database to define a new user type, and adding a user interface to give him functionality--the same functionality that already exists for photographers, but in a way that separates all this data into smaller contextual stores. In fact, nothing really "new" needs to be built, just expanded among a new user class. It's more a matter of building the business model, which is really where the work would be. Even then, it's more a matter of policy decisions on how to monitor and compensate the various players in the system. There are also needling points such as whether a "sale" took place in one Editor's account or another? What if the visitor noticed the same image in both spaces? Things like that. The details of the business model really have to be thought through carefully, but this is just a matter of filling in the blanks. Once done, the technology is brain-dead simple.

To those who might feel there's not enough margins in the microstock business to support disseminating additional commissions to editors, This is arguing a point that I feel is already off the table: it's my position that the $1/image microstock model is already a non-viable one that cannot be sustained. Some sort of change is imminent. I've suggested that $9.99 is most likely the sweet spot for any sort of real pricing baseline (this being the lowest price for the lowest resolution image for a one-time use license), and I believe anyone that would pay $1 for an image, would also pay $10. This ten-fold increase in revenue would be more than enough to absorb the additional expense of compensating a new class of user.

Once you grasp the concept that the Editor acts like a virtual stock agency, a truly free-market model is able to flourish. Photographers' main incentive is for all Editors to pick their images, since the most likely point of sale will be through Editor's sites. As such, photographers now have incentives to employ "good" keywording techniques, as they don't want to annoy Editors. Similarly, member "ratings" of photos become irrelevant as well, since Editors make their own selections for what they think is marketable, and how to present them.

Other likely actions may include user-generated incentives: an Editor could offer to discounts to buyers to generate business. The more downloads he gets, the higher his ranking, which turns into more visibility. It won't be long before Editors learn that giving up some early commissions by re-investing in their own growth is a good use of funds. Similarly, photographers may also partner with well-established Editors to gain a footing in the same way. The buyer wins because he saves time by getting high-quality results, the Editor wins because he gets more points and traffic, the photographer wins because his images are being sold better than he could have done himself, and the central hosting site wins because they get more on both sides without giving up anything more than they already have.

Photographers may be Editors as well. They can incorporate images of other photographers into their own if they like--it all depends on how they want to spend their valuable time. The "Photographer" class of user is different from the "Editor", and there are no restrictions on which account types one can have, as there's no need to. No policing is necessary--the economics and the mechanisms allow for Editors to find and prepare the best images to be used in a virtual storefront that attracts buyers.

Another example for how incentives aid growth: Editors may band together to create groups of larger virtual agencies, where they all share in the rewards of the collective. Because such groups would have oversight over one another using mechanisms to remove users from the group (and undo or even exclude their edits), this becomes more like a real-life workgroup or a real company. (Remember, in the Prisoner's Dilemma, you don't know what the other players do, which is why you can't trust them. "Editor groups" would not have this problem, since everyone's actions are transparent, and the leaders of the groups have oversight.)

As for the relationship to the hosting photo site, there's a dual risk/benefit. On the one hand, Editors would be essentially "locked in" because image editing requires time and brain-work that cannot be transported to another stock agency or photo-sharing site, simply because the photo assets differ. Editors have no incentive to go elsewhere or they'd lose the opportunity to monetize their investment thus far. They "could" move around, but they'd dilute their focused impact on any given site if they diverted their attention elsewhere.

On the same token, their value must be rewarded highly by the microstock because Editors perform the one service that microstocks haven't yet done well: Edit. Such users, if properly compensated, can add far more value as a crowdsourced group due to the re-assignment of incentives (and the byproduct of removing disincentives to cooperate) than any in-house staff could do. (And they can do it better.) So, microstocks have a much greater incentive to treat Editors especially well, which is a critical component to success. Microstocks themselves may eventually find they no longer need their own in-house editors, who themselves may find it more profitable to be member-editors of the very companies who once employed them.

Let's be clear: the value is less the photos themselves, which are all but commodities these days--the value is the creative accumulation of the efforts of the editors, which would include keywording, prioritizing, pruning, and final packaging that help these photo assets sell.

Economic Integrity and Pricing Stability
Free-market economics also maintain stability on a number of factors. Obviously, some stock agencies succeed, while others fail, but it's rarely the case that an agency can have a direct effect on another agency's success (without breaking laws). A virtual agency is no different in this regard. First, there's not a whole lot of ways to cheat the system because Editors act entirely independently, and only the buyer is in control of which Editor's lists are used. "Cheating" doesn't apply in this context--if you were an Editor and simply copied images into your list unaltered (just to get a lot of matches), you've done nothing more than the baseline results of the global system itself, which won't result in much activity for your account. To really succeed, you have to actually "do work," and nothing can improve to degrade your work other than you.

Also, the relationship between the editors and the photographers is far more reciprocal than has been seen before in traditional stock agencies. That is, larger agencies have always controlled photographers by virtue of their monolithic editorial staffs that control all image marketing and presentation. (Buyers only see the images that the agency permits.) This type of overt control doesn't exist in a microstock site because all virtual stock agencies act independently. The ease for photographers to move to multiple image hosting sites (and multiple editors within them) allows them to be freer to control their own destinies than before. Again, it's the free-market model working naturally.

So, let's now come back to the subject of user-specified pricing of photos. Fortunately, to everyone's benefit, the wider distribution of photo assets and creative editing removes the incentive to drop prices for the sole reason that it doesn't attract buyers because they (may) never see those photos. That is, if someone tried to lower his prices in order to induce sales, it would only be effective if an Editor picked up his photos in his presentation portfolio. But, because the Editor makes a commission on the sales, lower prices is a disincentive to pick up that photographer's pictures because his commissions would be lower. Worse, even if no sales occur, the perception of value from lower prices may degrade the Editor's business legitimacy.

Now, what about poorly performing Editors who also carry low-priced photographers in order to gain their own sales? Unless this Editor's work in keywording, editing, captioning, and all other work that make his section of the site worthy of customer visits, he's not going to get any visibility anyway, and certainly no download credits that would raise his rank in the overall listings. So, his lower prices would, frankly, never be seen. (This ties back to policy decisions that the hosting site would have to consider.)

This will have virtually no effect on higher-priced Editors much the same way that one store's prices isn't directly affected by another store's. Again, remember this is more like different agencies rather than head-to-head photographers in a single microstock site.

Ultimately, the buyer holds the purse strings, but at least the presence of multiple players with incentives to keep prices up will keep the buyer in check.

The virtual stock agency is what I consider to be the "middle ground" between a microstock and a larger agency, and is one that satisfies all the objectives of a sustainable business: social-networking, photo-sharing, economic viability, and no built-in incentives for members to erode the system.

Crowdsourcing content (photography) is good--and my model takes it one step further by crowdsourcing the editorial tasks in a way that meets economic viability minimums. Microstock as a business model would evolve from being more than just a photo-housing center of billions of images. They would be hosts to virtual stock agencies that act as catalysts to sales.

Labels: , , , , , , ,