Those might play a role, but wouldn't it be more likely that most software startups are optimizing for something other than revenue? Most take funding specifically so they can not worry about revenue, and instead focus on user growth.
I suppose a third possibility is that software compensation may have a larger equity component, which would allow software employees to take similar overall comp at lower revenue/employee levels.
Advertising and Media firms frequently have very high outside expenses, so that they appear to have much higher revenue than they really do. It's not just actors, it's all of the other contractors that are used to produce an ad or a show. So the real revenue is just their markup on outside costs. Plus much Ad firm revenue is from media buying - the real cost of a TV campaign is not in producing a spot but in buying placement, and the ad firm only sees a relatively small markup on the cost of placement.
Then you have the issue of conflating consumer software firms - Twitter, Facebook... who capture minimal revenue while they try to build out their userbase and network effects - with enterprise software firms who can and will charge their day 1 customers large amounts of money.
This exercise would be interesting if it were conducted by someone with a strong understanding of the structure and strategic choices of the software industry and how to compare firms with inherently different gross and net margins.
This analysis is geared less at understanding profitability/valuation metrics and more at operational and modeling decisions around growth, like headcount needed to support revenues in high growth companies.
The 2mm/employee cutoff is an interesting insight. Sidenote - is that what the grey line in http://blog.joshlaurito.com/wp-content/uploads/2013/10/Scree... is supposed to represent? It wasn't fully clear to me.
Is there anything actionable that you found? This all seemed to me to be interesting but non-actionable to a founder.
>I assumed that software companies would be more scalable and generate more revenue per employee, but the numbers don’t bear this out.
Any chance that's a bias from your data? Software companies can span the gamut from startup that's optimizing for their monthly-active-user-count to enterprise-software firm that's established with a high margin and low capex.
Also, a little discussion around why you used log-log graphs would have been welcomed by me.
You are right- the range is wider than I mentioned, and the true minimum is lower (the numbers in the post are industry-by-industry averages).
I used log-log graphs because they reduce the visual impact of outliers. You can play with the graphs yourself at http://blog.joshlaurito.com/inc5000.html if you want to see alternatives.
There are definitely biases in the sample: these are only fast-growing, 3yr+ old companies that want publicity badly enough to open their books to Inc.
If you are working in a company in a company with a similar profile or compete with any of the companies here, I think the data is useful for deciding how quickly to hire and benchmarking against any competitors that might be in the sample. Also, if you're writing a business/evaluating a business plan this might be useful data. For the rest of us, it's just fun to play with!