2009 Year in Review: Web Optimization

In this second segment of my series, "2009: Year in Review," I discuss issues related to managing my web presence. Some of these methods directly result in income, such as advertising dollars, whereas others indirectly affect income, such my ranking in search engines or by directing traffic towards monetizable content. Nothing discussed here addresses my actual sales and licensing methods, which was addressed in Part 1 of this series.

Web Traffic and Advertising

Traffic to my site has marginally increased by 16% from the same time last year (2008). More specifically, I averaged about 15,000 visitors a day in 2009, but the number would have been much higher had it not been for a technical mis-decision I made during the summer months that dramatically dropped my rankings, which had to do with "keyword stuffing", discussed later. Normalizing for that, my traffic has been pretty steady at around 16-18K unique visitors a day, compared to 14-15K/day in 2008. (Stats can be seen here.)

While that may sound impressive, it's not that simple. There are a number of devils in the details, and sifting through the data is only half the battle. For example, the bounce rate (the rate at which people leave my site after viewing the first page) rose to 8.5%, and the average time on site dropped by 11%. In other words, people are leaving my site sooner than before.

One would think that this is a bad thing, but there's other data that suggests otherwise. For example, advertising revenue more than doubled; in some cases (some pages and topics) tripled and quadrupled. All those people "bouncing" away without spending time on my site are clicking on ads. For 2009, advertising revenue jumped to represent 17% of total income.

One might say that I'm losing potential buyers to advertisers, but that's not what's going on. Most of the ads on my site are not for photography prints or licensing, which is the lion's share of my online transactions. That is, people are clicking on ads because they decidedly do not want anything I have to offer. I don't care that they leave; it just so happens that they're paying me a effective "exit tax." Or rather, the people who are getting my traffic are paying that tax.

Indeed, this turns out to be mutually beneficial: advertisers whose own sites don't rank well for some search terms, actually get a lot more relevant traffic from my site than they would if they paid to get onto Google's search page directly. That is, they'll pay ten cents to a dollar per click to put an ad on my page (through Google's adwords program), compared to twice or three times that much to put the same ad on Google's search results page. They may not quite get the same number of total traffic, but they'll get much more relevant traffic that converts to revenue if they place those ads on my site (or any of the other top-ranked sites). This kind of advertising-indirection costs them less, they get better bang for the buck. Best of all, I get a cut of it. :-)

I should point out that this isn't always so straightforward for advertisers, because targeting a specific site can be costly (in the form of lost opportunity, not necessarily money) if that site isn't consistently well-ranked. That is, if they target a site that appears to rank well sporadically (because their content changes), they could get a boost of traffic for a short time, and then go dark. Since my site has been around for a long time and is generally stable, this risk is not a concern.

In fact, many advertisers come directly to me and pay me to put their ads on my pages, rather than going through Google. There are advertising aggregators that have clients that pay them to do this analysis, and my site is coming up more often in their radar. My advertising rates are not based on clicks or impressions; they're flat fee rates, which advertisers like a lot for a high-traffic site like mine.

This then begs the question: what was the actual end-user looking for that they landed on my site, even though I didn't have what they were looking for? Why am I ranked so highly for them? Isn't that a problem with the search results?

First of all, the bounce rates are still quite low. Google does accurately put users on pages that match their searches. Of the low number of people who bounce, it's usually because they used the wrong search terms in the first place, and Google couldn't possibly know that ahead of time.

Take the Olympics in Vancouver, for example. If you search for "photos of vancouver", I'm currently ranked #8 on Google. (Before the Olympics, I was ranked among the top three.) So, I get a lot of people looking for olympics photos, even though they didn't use the term, "olympics" in their search query. When they don't see such images on my Vancouver page, users click on an ad that gets them where they wanted to go.

Vancouver is only one of a long list of examples. At the moment, I score very highly for phrases like:

1. "black and white pictures" (Google Rank: #4)
2. "what kind of camera should I buy" (#6),
3. "learning photography" (#2)
4. "photography business" (#1)
5. "model release" (#1)
6. "star trails" (#1)
7. "fill flash" (#1)
8. "photographing people" (#1)
9. "selling prints" (#1)
10. "photography marketing" (#3)
11. "sahara desert" (#5)
12. "stairs" (#6)
13. "photos of doors" (#1)
14. "photos of new york city" (#3)
15. "photos of san francisco" (#1)
16. "photos of kids" (#1)
17. "photos of united states" (#1)
18. "photos of patagonia" (#3)
19. "photos of cuba" (#1)

These are but a few among hundreds of phrases that Google ranks my site and/or pages among the top-five. But the key is that these terms are generic and they themselves do not bring traffic that can be attributed to a single dime of sales revenue.

While they are good for generating advertising revenue, there's an even better benefit to ranking high for generic search patterns: Non-buyer traffic out-strips buyers by orders of magnitude, and any traffic--buyers or not--contributes to the overall ranking of my site. When people search using more specific terms (for content that they do want to purchase), my site will rise in those search results, yielding sales.

So the objective is to have as many pages rank as highly as possible. One key strategy here is that I don't particularly care to rank highly for any single or small set of search terms--that doesn't necessarily benefit me. It's just having my site itself be indexed well for whatever content the search engines deem appropriate. And therein lies the question: how do they determine what search terms should send users to my site? Since they cannot determine what's inside of a photo the way a human eye does, search engines look for other clues to determine the content of a page that otherwise has very little text: metadata.

Keywording

I've blogged before about keywording; it's a huge topic. I'm not going to reiterate points I already made, but to appreciate how and why I employ my keywording methods, you need to at least understand this very basic set of truisms:

1. Most image buyers use search engines first, stock agencies second.
   Search engines act like "metasearch" for all the stock sites, as well as many other image sources, including mine, yours, everyone else's. It's best to use keywording techniques advised by search engines, not stock photo agencies.
2. Search engines are intelligent about search queries.
   Unlike days long ago, they know all the synonyms that are related to a common root. So, you do not need to include the singular and plurals, all the variants of "dog" (canine, puppy, pooch, etc.), and so on. What's more, intelligent search is becoming more common, even among stock agencies. The need to stuff your images with synonyms and other related keywords to make your list "more thorough or complete" is gone. In fact, attempting to do so can backfire on you. (More about that later.)
3. Controlled Vocabularies are a complete waste of time.
   There was once a time when such lists were useful, because it made the job of image search much easier for unsophisticated (brute force) search algorithms. Controlled vocabularies helped you use a small, consistent set of words, which kept you from using dozens of similar words that might come up with different search results when the user input search queries.
   
   While that premise was useful, it only addresses half the equation: the weakest link in search is not you, it's the end-users. Or rather, the search queries they submit. These people are not going to conform to controlled vocabularies. So, in order to map their queries to your images, their input text has to be converted to root words anyway. If the search algorithm is going to do this to end-user queries, it can (and should) also do it with your keyword list. Forcing you to conform to a list becomes a waste of time.
4. Keywording should take only a few minutes and minimal thought.
   It's very easy to over-think how people might find your images, or to worry that your images might not be found if someone uses a series of queries that you didn't think of. But this kind of over-thinking can negatively affect if and how your images are found. End-users learn very quickly to be very conservative in their search queries, or they will get a lot of irrelevant results, rapidly wasting their time. They may experiment with creative, conceptual, or "refined" queries to see what they get, but it doesn't take long to learn to "keep it simple." So should you. Keywords should include only the most basic, obvious, and prominent items in the photo. Search engines also rank the quality of photos (and the sites that host them) on their brevity. More than ten keywords will diminish a photo's rank because it usually means that someone is going to stuff the keyword list with unrelated words in an attempt to game the system. This is a common technique among photographers who submit their images to dozens of microstock agencies who do not enforce such restrictions, and who use brute-force (letter-for-letter) search algorithms. Keyword stuffing--also known as "keyword pollution"--has proven to be effective for such photo sites because it allows those images to be found ahead of other, potentially more relevant results for any given search.

In fact, I fell victim to "keyword stuffing" myself midway through 2009. In my automated keyword algorithms, which normally strips redundant or "similar" keywords, I had thought I was being clever by adding in location information (city, state, country) into the keyword list. Yet, what I found was that because the IPTC data already had these keywords, which search engines tap into, and because my keyword list grew (unnecessarily) by three more words, this dropped my rankings down by several notches, which kept me out of the "top fold" of search engine results. It's a huge deal dropping from #3 to #6 or #7 for a given search term, and you can see the results of this in my site traffic data over the summer of 2009.

Needless to say, this cost me quite a bit in traffic, which affected every other aspect of my business, from sales to advertising rates.

You can imagine, therefore, that "effective keywording" (so that images and website are deemed "credible" and ranked highly) is a hotly debated issue in the photo community. It's also one where entrepreneurs try to come up with solutions--some good, some not so much.

One example is a product "imense annotator" (annotator.imense.com), which has some interesting ideas, such as an image-recognition algorithm that tries to guess keywords that might describe the people in an image. It will do a reasonable job in ascertaining the ages, sex and ethnicity of people in a photo, and then attach those keywords to your images. Clever, and possibly quite useful more to a stock agency than an individual. This is because agencies have millions of images to process, none of which have been (or will be) seen by company staff. On the other hand, original photographers that shot the images could do this task quite easily on their own. One can only shoot so many images in a day, and since one has to eventually go through a manual (if not minimal) keywording phase anyway, one can assign the keywords associated with the "people" photos as part of that process. This shouldn't be all that time-consuming for reasonably well-disciplined photographers. And human analysis on such things is always going to outperform a computer. (Yes, I say this as an active programmer.)

(Note: The annotator only does people/facial recognition.)

All other aspects of annotator look and sound cool, but are considerably less effective in practicality. Again, these include "commercial vocabularies", "crowdsourcing" and "controlled vocabularies." As noted earlier, these ultimately contribute to the perils of keyword stuffing that search engines don't like--and which only serve to confuse stock agencies' less sophisticated search algorithms.

Another thing to keep in mind is keywording is often done once, and then you never touch those particular images again. Therefore, whatever you use as keywords today are likely to stick with your images long into the future. But technology doesn't sit still--especially image-recognition and search algorithms. For these, time has a tendency to speed by rather quickly. Before you know it, most search engines will be incorporating the same sort of algorithms like the annotator above. In fact, Google's own image recognition features are rather well developed, and can be seen in action if you use their Picasa image management solutions.

In any event, the point is that keywording is a classic case where "less is more." Images should have minimal base tokens in the keyword list; the search "intermediary" interprets the uncontrolled end-user queries and maps them to the minimal keyword list in your images. This is and will always be the most effective way for images to be found.

While I don't necessarily fault software companies for coming up with creative ways to "enhance" keywording, I draw the line when companies actually recommend methods and behaviors that are wholly counter-productive. An example is Cradoc Software's latest product, fotoKeyword Harvester, a product that does a form of semi-automation of keywording your images. While I am a fan of the company in many ways because it tries to also be the photographer's "coach" on many vital business matters, it has never been on the forefront of the photo business--rather, they seem to be stuck in the 1990s with many of them. Alas, most of their advice, while applicable 10-15 years ago, is well behind the times today.

In the case of the Keyword Harvester, the company sent out an article titled, "best ways to keyword images using concepts and attributes." A quote is: "You'll need to start paying attention to how images convey messages in advertising." They say:

One of the most valuable types of keywords for an image are things called Concepts. A concept is a term that describes non-concrete aspects of your image, an abstract idea. Concepts are used by advertisers to sell their product with the use of your image. They want the consumer to think of something specific when their product is thought of. (...) For example: Wells Fargo Bank uses images of cowboys, wagon trains, horses, and the wild west to promote their business. The concepts for these images are: excitement, freedom, trust, historic, strong, powerful.

There are several problems with all this. First is one I highlighted above in my bullet list: photo searchers (commercial or not) do not use conceptual search terms very often--at least, not with much success as they once did when the stock industry was far smaller, before digital images, and before the internet--a time when almost all stock sales were dominated by Getty Images. Back then, yes, conceptual keywords worked. And this was because Getty internally controlled all keywords for all images. Also, they had their own intelligent search, and they controlled the images in their databank.

Today, images are found in many places, are keyworded by arbitrary staff--or worse, photographers--and the consistency is impossible to centralize and manage. The direct result is that photo buyers don't search the way they once did. (This is an example of Cradoc seems to be stuck in the 1990s.)

It's easy to put this to the test: go to images.google.com and search for the "conceptual keywords" that Cradoc said represented the kind of themes Wells Fargo uses in their imagery. I tried every word on their list, as individual search terms, in pairs, in triplets, and as the entire group. Not one single set of results from these queries contained images that would ever be used by Wells Fargo. They are totally unrelated to all their business models. This is not unique; it's rarely ever the case that conceptual keyword searches yield desirable results. That's why most searchers don't use them anymore.

By contrast, if you search for images based on the actual elements used by Wells Fargo imagery -- cowboys, wagon trains, horses -- image search results show many images similar to those the bank actually uses.

Again, the lesson: keep it simple. Don't get clever. Do not try to anticipate what the searcher might use as search terms. Photo researchers are more afraid of you than you are of them. They are going to keep it simple, too.

I can verify this with my own statistics: My site gets about 19,000 search queries a day on my own search pages. Of the search terms I get, 99% are for very specific items. Furthermore, when someone actually licenses an image from me, and I track their search patterns that lead up to the sale, it is never the case that people use conceptual terms.

In preparation for this article, I interviewed one particular client about how he tends to search for images. He said, "I found that sites are so inconsistent about search terms, that I've learned not to use big words. Just be as specific as possible to the actual things I want to see in a photo."

When I asked him how he chose the particular photo he licensed from me, and what search terms he used leading up to it, he said he wanted a "futuristic landscape." When he tried that phrase (and derivatives, such as "future" and "cityscape") on Google, Getty and Corbis, he got nothing like what he wanted. So, he just got specific: "glowing buildings", which lead him to the image he licensed from my site, which can be seen here.

Keywording Methods

So, let's get to brass tacks: how should you keyword your images? Google has a document called, Google's Search Engine Optimization Starter Guide, which includes tips on optimizing your images for search. It all boils down to:

1. The image's filename should include the most relevant elements of the image.
   For example, if it's a photo of a boy and a dog, use "boy-dog.jpg". If you have many such images, use sequences: boy-dog-1.jpg, boy-dog-2.jpg, etc.
2. Use keywords sparsely.
   The more keywords you try to associate with an image, the more you dilute it, bringing down its "rank" and relevancy (and credibility) with search engines, or with given search queries. This is because search engines use two key metrics to determine how well a given image matches a search parameter: the ratio of matches between an image's keyword list and that of the search query, and the filename of the image. For example, if the user entered the query, "boy and dog", the search engine sees two words: "boy" and "dog." (It throws out filler words like "and.") Here, the image named, boy-dog.jpg has a 100% hit ratio of query terms with keyword terms, and the keywords were in the filename. Note that the actual photo itself may very well be that of a fish and a boat. (Google doesn't actually look at that, because, well, it doesn't know how.)
3. Avoid using synonyms and other "related" terms in keyword lists
   That is, do not attempt to be thorough in describing images with keywords. That's not your job. Search engines already know how to do that. They've got thousands of programmers with PhDs doing that for you (and for the end-user). The more you try to "help," the more you're actually interfering with the process, which reduces your relevancy and ranking.

The good news about keywording is that proper and effective use of keywords is extremely simple and shouldn't require much (if any) thought or time. Using myself as an example, my workflow involves two phases: the edit phase (where I rename all my photos so that their filenames reflect their content), and the keywording phase, where I apply individual words to images--usually in very large batches.

For example, let's say I'm on a photo shoot of a boy and a dog. After editing out the stuff that gets tossed, I'm left with several hundred images, where I then name them just as recommended by Google: boy-dog-lake.jpg, boy-dog-bridge.jpg, boy-dog-1.jpg, etc. In order to assure the highest ratio of search queries to keyword terms, I try to limit filenames to two to six words, though most are either three or four. This is a difficult decision because if I use too many words, I may "match" more queries, but the ratio will be diluted. If I use too few words, I will rank highly for very narrow searches, but may miss more opportunities. This trade-off is a zero-sum game, so rather than try to game the system, I just be honest: determine what's in the photo, and use that as the filename.

Any words that may be "in" the photo, but seems to be less relevant are then added to the keywords list in the image's metadata. And even then, I rarely add more than two or three words, usually modifiers such as "young" or "funny."

Naming files is often very quick because most are batches of similar images. One only needs to browse a given gallery on my site to see the number of similar images that are shot together. The keywording process is similarly fast, also involving mass-assignment of specific, unambiguous words to large batches of images. My rule of thumb is that keywording thousands of images should take no more than 30 minutes.

Most any image-management software can add keywords; I happen to use Adobe Bridge, which is bundled for free with Photoshop or any of the creative suite products.

Note that if you inspect the images on my site, you may notice that they appear to have lots of keywords. Most of these keywords aren't actually in the images that I process--these are added later by an automated post-production algorithm that generates all my static html pages. I do all this to present hints to the end-user for suggested related search terms to stimulate new search ideas.

Maps

The newest addendum to my website is the use of Google Maps. Essentially, each of my web pages incorporates a google map to represent where every photo was taken. While it may seem frivolous, there's been great advantage to the maps. (It also wasn't entirely easy; Google set up the whole mechanism for the sole purpose of presenting maps based on specific street and/or mailing addresses. I have no interest in that level of detail; I just wanted to generate maps for generic locations, like city/state/country. Well, that isn't quite so easy because there are many streets named after cities, states and countries, and there's no way to tell Google maps that I'm not interested in street addresses, just general city maps.)

Though I instituted maps onto my site late in December, the effect its had on my traffic and ranking has been a surprise. Search engines seem to give extra boost to web pages that are geo-tagged--that is, they indicate location. When people search for images where the search parameters include a location, my pages get an additional bump. I've seen about a 10% boost in traffic two months after having introduced geo-tagging onto my web pages, and I look forward to seeing more data to quantify the extent to which geo-tagging has long-term benefits.

Lying about Photo Licensing

What was your annual income from photography last year? What was your average license fee? What kind of terms do you agree to? Do you ever give away work for free, or shoot an event gratis?

If you're like most people, your answers are highly unreliable, and most likely weighted towards the kinds of answers you would like to be true, especially if you believe you can give a "bump" in the right direction for the industry as a whole.

Unreliable answers from survey participants goes with the territory in the data analysis world, but in the photo licensing world, is it enough to distrust the underlying assumptions we have about the photo industry, such as the total market size, or the role of semi-pros and consumers?

The question rises to a new level given a similar awakening within the radio industry, according to this article in the New York Times (Dec 16, 2009). New, provocative and surprising insight about people's actual listening habits, versus what they claimed they were doing, has had dramatic effects on advertising rates, and even the existence of certain kinds of broadcasting.

According to the Times article, what has propelled the industry into a flurry of self-examination was a recent conversion from "measuring ratings through surveys to monitoring listeners electronically using so-called Portable People Meters."

Among the findings, the Classical Music market dropped by 10.7%, Talk Radio by 2.6% (and consists of 80% conservative commentary), and more people listen to "light rock" and "easy listening" than they ever admitted before.

"People tended to look at it almost like an election -- they would vote for the things they liked," said Jaye Albright, an industry consultant with Albright & O'Malley, a radio consultancy.

Classical music, being one of the largest music forms and radio station formats affected by the new data, is probably most closely associated with the photo industry because of the impassioned opinions by its own advocates. According to the Times article, classical music is perceived by its advocates as being an important civilizing force, and an "art form that is extremely related and important to our cultural history," Joseph W. Polisi, president of the Juilliard School, said.

But, as the objective and indifferent truth-telling meters indicates, strong belief in the culture and the importance of the art for does not necessarily translate to people's actual behaviors.

Saying you support a point of view, even though it's not backed up by actions, is one thing. Another is that people actually engage in behaviors they wouldn't admit to.

For example, more people listen to oldies, country and "light rock" than they have admitted in surveys. Especially men. In fact, under the survey format, 34.7% of men volunteered that they listened to soft rock, but when they were using the meters, it turns out that 40.1% did -- a 16% jump. This has a huge impact on the rates advertisers are willing to pay, and what stations are willing to broadcast. And this affects where investment goes, and so on.

Indeed, these discrepancies are consistent with findings within the television industry, when it moved away from volunteers hand-writing their viewing habits to being given electronic monitoring devices. As Arbitron (the ratings company) put it, "people overstate listening to stations they felt reflected better taste."

As an objective photo industry analyst, I immediately see an identical phenomenon in the photo industry. I've long argued that most in the photo sector use unreliable data collection methods, survey models, and sample sizes that have never represented the population at large.

To wit, most pro photographers and trade organizations cite two common sources for their industry data. Cradoc Software, makers of FotoQuote, a software application that helps photographers come up with tools to help price their work based on prior sales figures they collect from the industry. However, their data is collected from pro photographers who volunteer licensing information, which, as we should have learned, is highly unreliable. And it's made worse by the unrepresentative sample size of the population of those who license images.

Other perceived reputable sources include surveys done by trade publications like Photo District News, and those from Jim Pickerell of selling-stock.com. In those cases, data is collected from either traditional stock agencies or self-proclaimed pro photographers (as defined as someone whose income from photography is more than 50% of their total annual income). The fundamental premise here is that they are the prime and statistically viable representatives of the bulk of all licensed images.

This then raises this disturbing question (one that I've been raising for years): what should one make of his analysis if it turns out that agencies only make up 60% of the market? 30%? 15%? Or Less? Would stock agencies start focusing attention on consumers? Would non-photo related media companies start eying photo agencies and social networks as a new, untapped source for potential revenue? Might trade associations and publications shift focus to the consumer market?

Perhaps so, but they can't do it just yet. Knowing that something is wrong with the old data does not draw of map of what the correct data looks like. Real numbers still need to be gathered.

And we're getting closer to that all the time. Using image-recognition technologies from PicScout and Idée, the web can be crawled and images can be examined to determine their source. PicScout has the advantage here in two ways. First, they have already fingerprinted and indexed most images from all the major stock agencies, as well as the larger microstock sites. In one fell swoop, they could examine images used commercial websites and calculate this critical piece of information:

What ratio of licensible images can be attributed to a stock agency?

When I say "licensible images," I'm referring to image uses where there is no legal ambiguity. That is, I'm not talking about social networks, photo-sharing sites, personal web pages or other sites that might host images in a manner that could potentially be permitted under Fair Use.

(For the record, using someone else's photo on a photo-sharing site is not easily defined as "infringement" because it depends on how the image is displayed, or other claims made by the individual that put it there. Many such uses are protected under Fair Use, as they involve critique, demonstration, education, or other kinds of factors that may not constitute infringement. our goal here is to examine only sites where images use are not legally ambiguous.)

This quick snapshot of information might also give us a sense of which agencies are taking which slice of the pie. Are Microstocks really eating the mega agencies' lunches? What about the Creative Commons?

Obviously, this is not going to tell us about license fees, or whether the photos are sourced from pro photographers or consumers, or whether images were stolen or licensed. But, we can get a far more reliable picture of what percentage of commercial images are actually from stock agencies.

While PicScout is currently in the best position to do this analysis, and that the data is useful, there are caveats, as it suffers from two major setbacks: 1) it only examines "commercial" sites, and 2) it does not track real-time use of editorial images sufficiently to have a reliable effect on analysis results. These caveats are important because they cannot be used to draw conclusions about the industry as a whole--only about the use of commercial images buy commercial websites.

And while commercial images and uses are very important, it should be noted that the editorial market is far and away much larger than the commercial market for images, largely because more content is used, sites publish more frequently, and in larger volumes. It is also more common to use images from sources other than major stock agencies, since the abundance of such content is higher, license fees are lower, and liability risk for infringement is negligible. Gathering data about image use for editorial uses requires more frequent crawling, more frequent updates of editorial imagery served by both agencies and photographers and underlying technologies that PicScout does not say they perform.

But again, these caveats don't invalidate findings in the commercial sector. In fact, I think it'd be more like the quiet, soothing alarm one uses to wake up than the blaring buzzer of a dime-store clock. But either way you look at it, the industry does need to wake up, and this data can have the most sweeping effects on the general understanding we have about the photo industry like nothing we've seen before.

If so, what happens next?

In my last blog entry, titled, Weathering Climate Change within the Photo Industry, I posed the question, "How would the industry behave if it turned out that their assumptions about the industry was entirely wrong?"

Since it is fast becoming within our technological grasp to actually uncover this information, I strongly suggest that the pundits within the sector consider that question. Take a long, hard introspective look at such beliefs and consider how strategies would change if it turns out that its core understandings and assumptions are misdirected. It won't be long before even more advanced research methods will uncover even more detailed information, such as actual license fees, the role of search engines in the licensing path, effectiveness of keywords and other metadata, and so on.

This will raise the volume of that alarm clock even more. And there's a reason for such a clock: you don't want to miss the plane.

"There are Lies, Damn Lies, and Statistics"

There's another old saying about statistics: "The biggest liar in the world is a politician, and the second biggest liar is a statistician."

The root of these witticisms comes from the fact that statistics can be used to support most any argument. It all boils down to what your message is, and what details within the statistical data set support that message.

In marketing, this is called positioning.

An illustration of this can be found in two recent news items. One from microstockdiaries.com reports iStockphoto having publicized their expectation to reach $200 million in 2009 revenues. By comparison, the official figures for 2007 were $71.9 million. A huge gain, clearly.

Secondly, a survey from Shutterstock shows that 40 percent of graphic designers are using more stock photography in 2009 as compared to last year, and 32% of them say they will be using even more stock images one year from now.

In each case, the respective companies want to present data that not just supports the notion that their sales are growing, but that the ingenuity of their business strategies is paying off.

From a marketing perspective, the stats are genuine. However, it's the broader interpretation by the analysts and the press that makes the data deceptive. Is stock photo pricing on the rise from the $1/image bottom spearheaded by microstock agencies a few years ago? Is the industry really growing by the same proportion represented by the increased revenue growth? One needs to churn through the numbers and deconstruct surveys to determine what, if anything, can be learned from this.

I contend that most all photo industry surveys leave out too much critical information to draw good, if any, conclusions. In fact, one could draw entirely opposite conclusions for some surveys. For example, one main unstated detail about the Shutterstock survey is the company's subscription service. Here, a buyer can pay $1 for a photo, or he can sign up for a subscription, where he'll pay $250 a month and download 25 photos a day. On a per-image basis, the economics of the math is quite attractive. Yet, chances are likely that most subscribers never use all the images they download. If one were to calculate the images actually used, they're most likely paying far more than the $1/photo price they could have gotten if they stuck to pay-per-use model.

Of course, the picture gets muddier still, because not all image prices are entirely $1 either. (Prices go up with size.) The fly in the ointment for those analyzing the statistics is that it's impossible to know how much buyers use what they've downloaded. One fellow blogger told me, "It's like an all-you-can-eat buffet: Do people really eat more than if they just ordered off the menu? How you can you tell?"

Of course, with digital imagery, it doesn't matter--you never run out of inventory. Shutterstock makes $250 regardless of how much the customer downloads, or how much he actually publishes. The only thing the company wants to do is optimize the amount of money the customer forks over. And the more attractive the "deal" appears, the more likely he'll pay.

So the surveys that Shutterstock does only has that one objective: to enhance that marketing message. Everyone else, however, is using this information to get a pulse on the health of the stock photo industry. And to do that, we really do need to know how many photos the client is publishing.

While we can't know directly from buyers, we can look at broader industry data on advertising purchases and editorial print to get a rough idea of the health of the industries that use stock photography. And, as we all know, ad sales and printed editorial pages are shrinking. So we can naturally assume that, even though "traditional buyers" may be paying more for subscriptions--as Shutterstock and iStockphoto have told us--buyers aren't using more images than before.

These contradictory trends (higher revenue from stock agencies against lower image usage) should be a strong hint to industry watchers that what they thought (and have reported and blogged about) isn't what's really going on.

While stock photo industry executives can be applauded for making more money, the fact that clients are actually paying more to publish fewer images would suggest that these companies have always under-priced their inventory--that buyers have all along been willing to pay more. It's like the all-you-can-eat buffet: if the price for a full dinner on the á la carte menu is $50, and the buffet price is $5, what affect does it have on the buyer if the price doubled to $10? Probably not much. And this is precisely how most image buyers feel about most stock photo prices, too. They're already too low to care if the price happens to double.

This is yet another illustration that most all stock agencies never put any science behind their pricing strategies. I've blogged before about this before in my article titled, The myth that microstock agencies hurt stock photo pricing.

The question still remains, just what is the sweet spot for pricing? That is an entirely different matter, and the aforementioned article above discusses the science of pricing in much more detail. Ultimately, the article concludes that the problem is that no one does true, statistically viable surveys on who the real buyers and sellers of stock imagery are. Almost all price surveys out there are retrospective--what happened in the past--and are limited to a very small and unrepresentative sampling of the stock industry on the whole.

And that's why Shutterstock's own surveys can ultimately lead to more self-destruction than the temporary uptick they may have gleaned from the marketing message.

Viable Sample Sizes

Unfortunately, there's a more disturbing fact at hand. If the Shutterstock survey is representative of anything, it's that it is typical of the kinds of surveys usually done by everyone in this industry. And the major errors are that their sample sizes are far too low, and the people they survey do not represent a random distribution of buyers. In this case, Shutterstock asked only 300 "successful art directors and graphic designers" (according to the survey report) about their buying habits.

Why only "successful" ones? Unsuccessful buyers are still buyers, and there's plenty more of them than successful ones. One can even argue that "unsuccessful" art directors probably overpay for images. (Another quote, this one from P. T. Barnum, "There's a sucker born every minute.") And if there are more of them, it would suggest the market will bear higher prices than what people currently see. The real question is, which group represents a more accurate sampling of industry buyers and their behaviors?

The PR firm that contacted me about this survey (Morton PR) said in response to my query on the matter:

...the survey doesn't include the average consumer ... many professional photographers obsess about the importance of the design community as an image marketplace.

And this type of thing is at the root of a systemic problem with the stock photo industry: The obsession with the "traditional" buyer.

What is it that constantly reinforces this misperception about the stock photo industry? To illustrate, ask yourself this: What percentage of the cell phone market is represented by the iPhone? Most people think it's rather large. All you hear and read about these days is the iPhone and applications for it. Everywhere you look, people are hunched over their iPhones, thumbing away... often while driving.

Now, let me broaden the question to include all smart phones, such as the Blackberry and all the Palm phones. What percentage of the cell phone market do all smartphones represent? While most people think these devices account for 80% of phones, the reality is that they all collectively add up to only 10%.

The false impression that people have of the cell phone industry is a byproduct of several factors: hype, consumer buzz, and the news media. The psychological effect all these have on human perception translates to our subconsciously looking for smartphones: because that's all we hear about.

The same thing is true of the photo industry: all the "news" gravitates around stock agencies. And by consequence, everyone thinks that's the only important part of the industry. And this belief creates an echo chamber and feedback mechanism, resulting in even more news coverage, bloggers, photo forums, industry trade groups, and of course, survey methods.

A classic (and consistently repeated) example of how this misperception is perpetuated by press-analysts can be found in the article, Will Hobbyists Take Over? Here, Jim Pickerell addresses this question by examining only stock agencies. He concludes:

An examination of iStockphoto's top sellers reveals that those making the most the sales are very active stock producers, not part-time shooters. Over 70% of those with images on iStock have had five or fewer downloads. They are the hobbyists.

The assumption that iStockphoto is a viable representative of where most people buy stock images is more than just incorrect, the agency represents such a very narrow niche of buyers that it is probably least representative of any single buyer demographic. Yet, the constant focus on iStockphoto (because it's the largest of stock agencies) is like the iPhone of the cellular world. Accordingly, his statement perpetuates the misinformation that the company sells the lion's share of images worldwide.

The evidence against this is overwhelming. Even traditional research in the year 2000 showed that only a third of licensed images came from stock photo agencies. The rest were purchased directly from photographers. In those days, only pros were selling stock images--consumers weren't really that engaged in online photo sharing to a sufficient degree to affect the market. But has that ratio changed towards or against agencies over the past ten years? What with the growth of photo-sharing social networks and high-quality digital cameras, why would one think that the ratio of stock image sales would remain in the hands of pros, or even the stock agencies?

Pickerell's observation of iStock's sales and the distribution of pros/hobbyists is quite easily explained: most pros heard of stock agencies, and have traditionally use them, so it's natural that they gravitate there. The same with the photo buyers who use them: they go there because they always have. They all happen to be the native species to that particular economic ecosystem -- it shouldn't surprise anyone that "hobbyists" won't go there, or do well there. That iStock may have many "hobby-like" members is nothing compared to sites like like Flickr which have have millions upon millions of users. And many of them sell stock quite well.

As for the buyers, the story is much the same. We don't expect to see many non-traditional buyers go to agencies either, since they traditionally never have. And since agency's websites rarely come up in general search results for most search patterns, there's no sign that this trend will change much in the future.

Ok, But So What?

Indeed, fear may be at the heart of all this. For if one to were actually do a proper study proving my point, what then? How would that change things?

More to the point: What's the alternative to a stock agency? Though many photographers sell on their own sites, it's not necessarily easy to do. It's a saturated market, and it's hard to get instantly noticed. Most photographers either can't, or don't want to build their own sites, nor have the patience or wherewithal to wait for their site to increase in google rankings so their images are "discovered." Which brings them all back to agencies (not that they do any better with them, by the way--it's just easier to fail with an agency than it is to fail entirely on your own).

The same is true for the traditional photo buyers: they've used stock agencies for years as well, and typically choose the kinds of images produced by those particular kinds of photographers. Tracking the behaviors of buyers who don't use agencies is, admittedly difficult. How do you find them? True, they're everywhere, but that doesn't necessarily help.

Difficult though it may be to target the consumer-photo-buyer, it doesn't mean they don't exist, or aren't worth trying to track. And it certainly doesn't mean they don't have a very large tidal effect on the market as a whole. They deserve attention. Ignoring them is, and always has been, the greatest reason why the stock photo industry has been struggling. There's more to the stock universe than agencies.

It reminds me of the quote from the Blues Brothers movie, where Elwood (Dan Aykroid) asks the bartender, "What kind of music do you usually play here?" To which she responds, "Oh, we got both kinds. We got country and western!"

Turning a blind eye to the rest of the stock photo universe affects decisions about where to put marketing dollars, where to do research into buyer behaviors, pricing, and business development. If it were generally accepted that the market were larger, agencies could form partnerships with other media licensing agents, social groups and legal networks that reach that larger market. (Getty's relationship with Flickr is in name only--there's been no actual change in how business is done. They only want to get access to rights-managed images from a uniquely small and select group of photographers. In short, it's a marketing ploy, not a true embracing of the consumer market.)

Discovering Buyers and Sellers

If the wider stock industry is so large, who are those photographers? To name a few: Wedding photographers, sports shooters, and many other "pros" (as Pickerell defines them) who have begun selling their inventory as after-market stock over the past five years. Do they get more than 50% of their income from stock sales to qualify as "pros" in Pickerell's definition? No, but they're not hobbyists either. And they certainly account for a great deal of sales.

An internal survey done by a magazine that I once wrote for showed that over 75% of pro photographers that don't "normally" shoot stock sell over $15K of stock licenses annually. This isn't a lot of money by Pickerell's definition of a "pro stock photographer" -- or, for a photographer that focuses solely on stock -- but it certainly represents a healthy chunk of stock sales that he dismisses from his calculations. The circulation for that magazine was over 50,000 at the time, which means that 37,500 people earn $562,500 annually from stock sales. While not huge by itself, it's one (small) magazine. Extrapolate to all pro photographers everywhere, the numbers get impressive: 500,000 photographers would generate $5B in stock sales annually, and Pickerell's data doesn't take into account.

And those are the pros. What about non-pros, who outnumber all pro photographers by orders of magnitude? We don't know specifically because no one has done that kind of survey. (Doh!) But, we can look to other data that might suggest what that might look like. For this, I revert back to the question of who the buyers are:

The Small Business Administration says that small businesses account for 80% of the american workforce. If so, this segment of the US economy must easily overshadow the "traditional photo buyer". And it would be naive to assume they don't license imagery. So, where do small businesses get their images?

A survey I did in 2007 of individual stay-at-home graphic designers and page-layout contractors (many of whom were part-time) shows that they purchased an average of $10K of images in the past year for their small-business clients. They create ads and brochures for professional services (doctors and lawyers) and family-owned businesses (drug stores, hardware shops, restaurants and cafes), among others (sometimes including larger corporations).

In fact, smaller companies were five times more willing to pay higher license fees than larger companies. This should also be no surprise, as big companies have weight and negotiating skills. WalMart negotiates for lower wholesale prices from suppliers, for example. On the other side, consumers are willing to pay higher per-unit prices as a general rule anyway. Consider ATM fees, which are usually $2-3 per withdrawal, regardless of the amount. Most consumers withdrawal $20, making the fee 10% of the transaction. Richer people withdrawal up to $200, reducing the fee to just 1%.

If stock agencies really wanted to improve per-image pricing, target the consumer. While they say they do, over 75% of the buyers I surveyed never heard of "stock photography" as a business. (Several thought I was referring to "stock footage", as in film reels.)

Of those who were familiar with stock-photography as we know it, only 60% heard of Getty. And only 10% of them used Getty. (In follow-up interviews, they heard of Getty because those are often the by-lines for newspaper photos; not because they purchased from them.)

So, where do they get their images? The #1 answer: "Google". Most people use the internet to find images, and extremely few image search results are from stock agencies. They bought directly from the photographer. Those photographers were not necessarily pros either. One woman said she expressly avoided buying from pros because they make the purchase process too difficult; they would require her (and sometimes her client) to sign lengthy license agreements. "It's just not worth the time. The cost of the image isn't the problem, it's the hassle."

If they're buying images and avoiding pros, one can only assume these are "hobbyists."

The #2 answer for where people buy images was specifically characterized as "Local Photographers." In follow-up interviews, the pattern is the same: friends, wedding photographers, art photographers (found from local galleries), referrals, and even their own photography.

The average purchase price per image from these hobbyists: $150/image. A price even Shutterstock would envy.

Was my survey representative of the industry at large? Not according to the rules of statistical sampling and random choosing of participants. I make no claims about the science of my methods. I conducted my survey by tracking down and interviewing the people who created the brochures that I receive from local businesses in my mailbox everyday. I also find those who create tri-folds that you find in hotels--the kind that advertise helicopter tours over the Grand Canyon. Things like that.

It's also not the first time I've done this sort of thing. But I get roughly the same results every time. It's admittedly not scientific, but the consistency of the responses certainly suggests that the "traditional" surveys done by industry analysts and trend-watchers is questionable.

Another fact my survey continues to support is the already-established truism that most stock sales are done on a peer-to-peer basis. That is, the buyer goes directly to the photographer. And this is also why it is hard to truly nail down the exact size of the stock photo industry. Unlike cell phone sales, where data is gathered by organizations that mine quarterly reports from publicly traded companies, photo sales are not done entirely through known entities who are required to release this information. (Heck, they may not even know themselves, because it's too "casual" as an income line item.)

Spending Habits

Having a better sense of who photo buyers and sellers are is one thing. But another important element is the analysis of their spending habits. This is another area where most traditional surveys fall quite short.

For example, surveys should break down buying habits by industries: which buy more, and which don't? Certain sectors do quite well in recessionary times, such as consumer staples, beverages, entertainment, to cite only a few. Would these companies show a trend towards one kind of photo source for imagery than those from other industries that don't do as well, such as construction, transportation, energy, travel? You can imagine that the old GM suppliers are probably not buying a whole of ads or spending marketing dollars, but companies that sell to grocery stores are probably buying a heck of a lot of imagery for their marketing programs.

Another data point about spending habits that's missing from the Shutterstock survey is "patterns." It showed that photo buyers "used more stock photography than they did a year ago." But that presents more questions than answers. The statement says nothing about total dollars spent on photography, or the distribution of how those dollars were divided. One naturally assumes that money has been diverted from one kind of photo expenditure (such as assignment) to stock. But that's just an assumption -- the survey didn't ask that.

For instance, if buyers spent 200% more on photos than last year, but only 10% more of that went to stock photo purchases, then it is still true to say that they "bought more stock than last year," even though their spending on other photo sources was even higher. (Of course, this is just a hypothetical to illustrate flaws in the survey methodology. I doubt this was actually the case here.)

The Business Paradox of Doing "Real" Research

In a follow-up interview I did with Felicia Morton, the president of the PR firm that helped orchestrate the survey and who represents Shutterstock, we discussed the difficulty in conducting what I would consider to be a viable study on the subject. The challenge is that a broad and detailed survey is expensive, so someone has to fund it. I asked whether someone would conduct the survey and sell the data to photo buyers. She said it was unlikely -- that companies like Nielsen Research wouldn't do it because they wouldn't think there's a sufficient number of buyers.

Seems obvious. Coincidentally, I had just such a client and would have been willing to pay for this research. So, I followed Felicia's lead and tried to contact someone at Nielsen to see what it would take to do the kind of research necessary to get a much more accurate handle on who really buys and sells stock images in the USA?

But, no one would return my call. I called two other national research firms, and one returned my call, and we spoke at length. But I was asked not to quote them by the time we got to the end of it. What I can say is that there is a general perception that the stock licensing industry probably is as large as I've always said ($20-25B), but the research firm doesn't believe existing players in the industry have any desire to either purchase the data, or (more importantly) to have the data out there if it were true. (This is why the firm didn't want to be named.)

Why would existing companies want to maintain the status quo if it were known that the stock industry were 10x larger than people think? Because it would require them to rethink their entire business model, which they would not be able to afford, which itself could compromise their current position. Therefore, with no one to buy the data, there's no financial justification for doing the research.

We also priced it out: it'd cost about $200,000 to do a real, nationwide research survey of consumers in determining what actions they've taken that might constitute a financial transactions involving stock image licensing.

Needless to say, when I reported this back to my client, he balked. His main concern wasn't so much the price (not that it would have happened anyway), but something more surprising: if the market really is $20B, everything changes. The big risk: one has to actually penetrate that market, or it might compromise the company's existing market position. It's hard enough to compete against companies that are already bigger than they are, but if it were known that there's far more fruit on the tree, then the bigger companies are already in a far better position to grow, ranging from financing to partnerships, and so on.

The question seems to have shifted now. It's not whether the market is so big, but rather, is it too big to tackle? Unless and until someone comes up with a business model that can successfully service the consumer-oriented stock photo buyer and seller -- which itself would require a monumental investment for a small company -- no one's going to try it. And if no one tries it, no one's going to spend money on a survey that shows the true market size. Moreover, no one wants that data to known, or it'll force the issue, and everyone's at risk. The industry is locked in its own self-imposed a stalemate.

This is not only understandable, but history has shown a similar phenomenon: again, in the cell phone industry. For years, not one of the major cell carriers was willing to "open up" handsets to allow consumers to download applications, not because they didn't think there was opportunity, but because it would disrupt their existing business models, which had been fine-tuned to a science. Furthermore, no one was willing to do the research to determine if there might be growth opportunity. Anything that might cause them to change their existing models was deemed as "risky."

When Apple finally disrupted this assumption, the model changed, and with it, so did the industry. Well, sort of. Cell carriers really haven't lost the stature they had before, but it sure has cost them considerably to weather the transition. But the real reverberations are being felt in the handset market -- all carriers (except for Apple) are suffering unprecedented losses. (The financial section of the July 16 edition of The New York Times had sequential headlines of reported losses by cell carriers.)

So, what will it take for the stock photo industry to change its ways and realize the potential of the consumer as both buyers and suppliers of stock photo content? It's unlikely to be spearheaded by an inside photo trade organization or stock agency, and the stock industry analysts, bloggers and pro photographers are locked into the position politically that the universe "is and always has been about agencies and pro photographers." (See my blog on the Economics of Controversy.)

It'll most likely be an outside player who sees the untapped opportunity of the global opportunity with stock imagery, much as Apple disrupted the cell phone industry. (An event that no one could have predicted just a few short years ago.) One thing for sure, whoever does shake the ground for stock, it will start by funding a statistically viable study.

I've predicted on my blog and in interviews that I think it'll be an existing media licensing company that merely expands its content library to include still images. They're already big, and they already have business models that include "consumers" as both producers and buyers of content. The stock licensing market is potentially enormous, and the existing stock agencies are drastically undervalued (because they don't recognize the industry as large.)

then again, I've heard of no rumors so far.