Shutterstock show a smaller market-share (by unique users) than they would if it showed market-share (by revenue) because of their strong subscription-base (for obvious reasons).
Defining and detecting a unique-user is pretty complicated (it’s a whole big web-analytics thing) since everyone measures it differently and in many cases it can be gamed (a couple of agencies gained 1% on their market-share in that chart because they’ve gamed their usage stats, or have better social-media-integration than others). It’s a small enough margin though that it didn’t need to be factored in for the sake of an infographic where 1% on a pie is pretty meaningless.

The iStock numbers are pretty accurate (give-or-take a bit of wiggle-room)… some details in this video of my StockInRussia presentation (~606 photographers make 80% of iStock’s revenue iirc).
iStock has just over 37,000 contributor accounts with at least 1 image for sale (getting excel to chart that many datapoints was a challenge I can tell you ).

~19 million images (~ = approx, in this case a ballpark of around 16-22m) uploaded (not accepted) is based on how many images I KNOW were uploaded to each of the 40+ agencies on picWorkflow, extrapolated for the rough market-share and number of agencies (and micro-priced individual collections) I know of across the industry, then reduced to account (this is the least certain part) for the number of duplicates (also gathered from picWorkflow data). It’s not ‘hard fact’ (hence the section is fact’oids’, not facts ), but it’s a pretty close estimate.
Don’t forget, on the scale of the web, 19 million is not that much… flickr was getting 1 million per day in 2006, and facebook gets 200 million per day now. There are a LOT of photos on the web, and a LOT of photo-selling sites. I know fb/flickr don’t equate to stock but as an indicator of how many people are capable of producing images it’s good to know a comparison since at such large scales numbers like this don’t make much sense

I also ‘discovered’ that many of the biggest contributors to iStock already do collaborate with each-other, or negotiate better terms. I’m not important enough to know for a fact what those terms are, but preferential search-placement is certainly on their radar. The iStock search-engine know who produces images which convert views-to-sales, and reward them.

The list of search terms are not from iStock (no microstock agencies currently make their actual search volumes available, despite my constant and ongoing requests)… the search terms are gathered from my own microstock search apps, addons, feeds, plugins, widgets and various other services/proxies I run. My dataset is skewed towards these split (so it doesn’t make 100%) demographics:
~90% technically-capable users (at least enough to install an addon/plugin or run a blog/website)
~90% individuals or freelance/self-employed
~40% free-users (want free images, I’m working on the upsell)
~10% spam-users (kids looking for naughty pictures)
The searches are accurate for all of my received searches, it’s close to agency data, but not 100% what they would see since I have very low corporate-reach (mainly means I see a drop in business or news related terms, I also have a low-side people search-list but that’s improving).

As MikLav said, I am referring specifically to growth when I say “outpace the growth”, not volume. Volume on stills will probably always outsell footage, though total revenue for footage will probably beat stills within 3 years.