I tend to get pretty cranky when I read the perceived wisdom of a new microstocker, full of pronouncements based on a couple of months' sales of a couple of hundred images.  Part of that reaction is general crankiness; I've reached that age, after all, and enjoy the privileges it confers.  But part of it is my own experience after doing this for a while, and my own understanding of the proper use and limitations of statistical methods. 

It's generally a bad idea to extrapolate from small data sets.  That was brought home to me today, when I saw that my balance at BigStock has nearly doubled since the last time I'd checked.  Lo and behold, this doubling was the result of a single extended license sale, the equivalent of between 40 and 80 regular sales there.  That was enough to vault BigStock from 7th to 3rd place for December, and to make a mockery of any attempt to estimate my sales for the month.

Fortunately, I've figured out a better way to look for revenue trends.  Instead of looking solely at sales by month, I've adopted a 12 month moving sum.  The idea is that for each month I add all the sales for the past year, with November including everything from the previous October, October going through the previous September and so on.  Not only does this give me larger numbers of sales, so a single or even a few surprises won't affect the total much, but it compensates for seasonal variations; every figure includes a result from every month of the year.  So assuming sales tank in December, or during summer holidays in the Northern Hemisphere, what I'm seeing are the success of my growing (and partially aging) portfolio, the microstock economy and perhaps the economy as a whole.

The attached graph shows two trend lines.  The blue line tracks monthly revenue, the red shows revenue for the year leading up to that month.  Blue shows an increase in revenue, but it's hard to tell with all the variation in customers' buying habits over the course of the year.  The red line makes the trend more obvious, and shows where my portfolio was gaining real traction.  Of course, I could also track the size of my port or other events that would explain the trends.  But the important thing is to eliminate as much of the noise as possible and concentrate on what's happening over the long haul.

I hope somebody else finds this interesting.  It does require of course that you have enough data, which is why my graph begins around my first anniversary with Shutterstock and iStock.  Before that point, I just didn't have enough information to form an insight.  Oh, and apologies for not including any scale information.  There is such a thing as oversharing, right?

PowerDroid,

The lower (blue) line is revenue for the month.  Due to seasonal variations, it bounces all over the place.  I don't track RPI and haven't since I started shooting with models two years ago.  That's because my RPI dropped like crazy with studio work, even as my bottom line revenue increased.  Shooting isolated objects or scenics or other subjects, I'd get a small number of shots in a day and have (I hoped) a high RPI.  But shooting with models, I can get as many as a hundred usable shots from an hour session.  Individually they don't sell terribly well, but as an aggregate they do.  So even as they lower my RPI, they do good things for my bottom line.

Hank

I tend to get pretty cranky when I read the perceived wisdom of a new microstocker, full of pronouncements based on a couple of months' sales of a couple of hundred images.  Part of that reaction is general crankiness; I've reached that age, after all, and enjoy the privileges it confers.  But part of it is my own experience after doing this for a while, and my own understanding of the proper use and limitations of statistical methods. 

It's generally a bad idea to extrapolate from small data sets. ...
...

Very true... It's called clear seeing... Before we draw quick results from monthly results, more large time periods must be analyzed and even some months foreseen. Also variations of the surrounding world has to be considered.
Also the drawn conclusion is true for a kind of set of images. Persons who have different content in their portfolios, have very different results.