LookStat – Official Blog

Last week I wrote about total earnings and downloads and the trends associated with pricing, downloads and earnings. In this post, I looked at the same data and adjusted for active users in that year to get an average per user for the year in question. The results are shown below (the earnings per download line is carried over from last week.)

User-adjusted Earnings and Downloads 2002-2009

Key Takeaways

• Declining downloads per user after 2007
• Slowing growth in earnings per user
• Increasing earnings per download

Although earnings per user have continued to rise, the downloads per user peaked in 2007 and have since declined about 20% from 2007 to 2009. These data support some of the things I heard at PDN and UGCX about flat downloads and increased earnings driven by price increases.

Increasing Competition & Higher Standards

As more contributors enter microstock, especially in a down economy when people are looking for other sources of income, there has been an increasing sense  that it is harder than ever to make money in microstock. If you layer on tougher acceptance standards, you can make a case that new contributors will have a tougher time establishing themselves in the market. This then suggests lower downloads and lower earnings per contributor

Competition isn’t the Whole Story

Increased competition and tougher standards only account for part of what is happening. I think there is no question that new contributors and images are entering the market at increasing rates. If the growth of new users exceeds the growth of the overall downloads, then we’ll see a decline in the per user averages. If increasing competition was the only factor however, we would also expect to see a decline in earnings per user. Clearly, this is not the case. Also, one thing that isn’t clear here is if competition is actually hurting established players or whether new entrants are just struggling without creating an impact on existing users. (A cohort analysis could help illuminate this but that is a post for another day.)

On Average, Users Are Earning More

Price increases by the agencies and increased pricing of individual images as they begin selling more are factors driving up user earnings. As contributors gain experience and their images sell, they benefit from increased pricing for their images as well as better placement in search results. There is a little survivor bias at work – you only stick around if you’re seeing success. It is interesting that this effect is more than compensating for the reduction due to competition, market factors etc.

Conclusions

It is harder to break in to microstock and succeed but there is no question that the market has grown overall, through difficult times. While there are many new entrants and standards are rising, increasing earnings per user suggest there is still opportunity in the market.

I’d love to hear people’s thoughts and interpretation in the comments.

The conventional wisdom is that photographers should shoot both horizontal and vertical formats and that square images do better because they stand out in search results (see screenshot below.)

The reasoning here is that square images take up more pixels in a thumbnail grid and therefore stand out. While this may be true, it’s instructive to look at the volume and earnings per image data to see what this really means.

Relative Image Volume in 2008

As you can see, photographers are getting good at rotating the camera while shooting. There are equal numbers of horizontal and vertical images being submitted. Square images lag behind, but that’s to be expected since it takes a little extra effort on the part of the photographer to frame and crop the shot.

Royalties Per Image in 2008

The sales data tell a very clear story – horizontal images earned 2x per image what the other formats did. While square images may take up more pixel area on a row of thumbnail results, they are not driving revenue to the same degree.

As food for thought, it’s interesting to note that all computer screens are in horizontal format and given the rise of online vs. print advertising, maybe these data aren’t that surprising after all. Also, given that most of us are pressed for time, microstock buyers may just not be willing to invest the time needed to crop a square image to their liking.

The bottom line, I still think it makes sense to give buyers a choice but you have to pay attention to the numbers to really see what’s going on.

LookStat collections makes it easy to do this sort of analysis on your microstock sales stats.

We did an analysis of  2008 data for images with the keyword ‘cat’ vs. ‘dog’ to see how they did overall. It doesn’t matter what metric you look at, it’s not even close – dogs rule!

Image Volume

There were twice as many images tagged with the keyword ‘dog’ as opposed to ‘cat’ in our system. This is interesting when you consider that according to the American Vetinerary Medical Association, cats outnumbered dogs in the USA in 2007 (81 million cats vs. 72 million dogs.)

Royalties per Image

In addition to there being more images of dogs, they also earned more per image and were downloaded more frequently. Royalties earned per image for dog images were twice those earned by cat images.

Images of dogs also had 30% more downloads per image than those of cats.

Total Earnings

This translates to dogs earning over 4x more than cats in 2008.

I’m confident that dog-lovers everywhere knew that their beloved pets would come out ahead, but it’s nice to see it in the data. Let me know your thoughts and whether you see similar things in your portfolios. Also, if you’re interested in specific keywords, let us know in the comments and we’ll see what we can do.

(Full Disclosure: I am a dog owner, but since Casey has a cat at home, we are a pet-neutral company.)

Our vision with LookStat was to create a web platform for microstock contributors that not only tracked sales but made it easy for contributors to really analyze their performance and ultimately use that information to make their microstock activities more profitable.

Over the next few weeks, I’ll be writing about ways to think about your microstock activities as a business and various metrics that are interesting in that regard. This post is focused on Return per Image (RPI).

Acknowledgements

James, Laurent, Lee, Matt & Yuri all have great posts about RPI. I would urge all of you interested in this topic to check them out. One of the things I love about microstock is how willing the community is to share their knowledge.

Return per Image per Month

One of the more commonly quoted metrics in the microstock blogosphere is RPI. This is not revenue per download, but rather it gives you an indicator of how much each image (across all sites) will generate in revenue for you every month. This number is important because it measures the earning power of your images.

While it’s a useful top-level indicator, it does have some limitations. The most glaring one is that it doesn’t really convey what happens over time. Matt has written a great post about this and the core issue is that older images sell less and RPI at the aggregate level obscures this since it only looks at the aggregate portfolio and doesn’t account for age.

Calculating RPI:

Consider the following two scenarios:

Scenario 1:

You have 10 images in your portfolio and they are listed at 5 sites. You earn $100 in sales at each site every month. In this situation, your return per image is:

Scenario 2:

In this scenario, everything is the same as in Scenario 1, except you have 100 images at each site. In this case, your revenue per image would be:

As you can see, the images in Scenario 1 have 10 times the earning power of the images in Scenario 2. Please, keep in mind, these are made up numbers. RPI varies quite a bit and the numbers above are very high based on what I’ve seen published. In general, based on conversations I’ve had, if you’re at $1-$2 RPI/month across all sites you’re in pretty good shape.

Limitations of RPI

RPI is a useful number but as such doesn’t give you enough information on how to proceed. Matt’s thought about selecting the top 100 images in a month and looking at their RPI is a good one. Another option is to look at age. Compare how your latest uploads perform in their first month with how your last batch did in their first month, i.e. cohort analysis. Each image has a revenue curve that varies over time and looking at batches is a way to better compare apples to apples. Laurent has a post discussing this that is well worth your time. In his post about Advanced Stock Theory, Yuri also discusses some of the factors that skew RPI.

Analyzing Sales over Time

The chart below was generated from LookStat data and shows the total revenue curve of a few of our images across multiple sites. Each curve tells a different story. A constant slope means steady sales over time. A flattening curve means declining sales over time etc. RPI at the aggregate level obscures this data. Tracking RPI at the same point in an image’s life cycle would yield more interesting data. For example, looking at RPI for an image for the first 30 days that it is listed would be fairly instructive.

(the data above is total sales hence the curves that rise and flatten)

Other Useful Slices

Another useful way to use RPI is to group images by category, by location, by model, by site and to see if certain groups stand out relative to others. By segmenting your revenues in this way, it’s possible to tease out some of the meaning that’s obscured by the aggregate RPI number. For example, you might find that within the business category, your RPI for isolated images is much higher than your RPI for images shot on location. Again, it’s important to compare apples to apples so looking at RPI at similar stages in an image’s lifecyle is important.

How LookStat can Help

The main problem with increasing the granularity with which you look at RPI is that it’s a giant pain to calculate. If you’re doing it by hand, you need to track sales by image and by day and then you have a dataset that you can slice and dice. This is fine with a handful of images, but it quickly becomes unwieldy. Unfortunately, this analysis is most valuable when you have a larger portfolio spread across multiple sites.

LookStat tracks sales at the individual image level and we aggregate data across sites. As a result, it’s easy for us to get at the details of the data. Basically, for every image in your portfolio, we know when it was uploaded and how many sales it had each day. We also store all the metadata associated with an image so in the future, we will be able to use that information to analyze sales even further.

Ultimately, we want our users to be able to look at their portfolios and see which models in their business images generate the most revenue for them at Dreamstime vs. Shutterstock.

Ultimately, we believe that better data for contributors means they’ll have a better sense for what buyers want which in turn will lead to them producing better-selling images. The beauty of microstock is that this should benefit all participants in the chain.

A Request

Please let me know which metrics you like to track and what you’d like to be able to track if you had access to the relevant data. Hopefully we can start a thread that would be valuable to the community and I’d love to be able to build in some of the most useful pieces of data into LookStat as the service evolves. Because of the granular way in which we build transaction histories, creating new data slices is relatively easy.