Wednesday, 19 January 2011
After the last 4 posts on the subject (1, 2, 3, 4), this is the final one
I had a few chats with respected and "bearded" analysts in the field, and realised that my unorthodox calculations would be fine as long as they'd make sense - the average analyst is only interested in earnings per share (EPS), it seems
I have a confession to make: my entire working life was spent as an expert in Systems Integration for a company commonly labeled as System Integrator. Of course I started off as a tester / programmer, but I ended up integrating any system with any other system - regardless.
System Integrators, or Consulting firms, make money by selling people: consultants. Those have hourly rates and thus can easily calculate their revenue - very few do by the way. Even if it's not on a strictly Time & Material basis, but rather project-wise, still their rates affect the project price so it's easy math.
Of course, in the past few years, cheap labour from India a.o. has played an ever increasing role, and that's exactly the reason why the Indian players stand on top.
A few years from now it might all globally merge and break-even, or shift to China or Africa and just keep on sucking the margins out of the classical SI's - who knows?
So, in short, comparing per-employee revenue and profit was a perfect measure I used to compare countries within the global group, local versus global service lines, and what more. It also was a perfect measure to establish overhead percentage: if the average revenue of most people around you was 200K, but headquarters reported an average of 100K among the same-ish, you knew that overhead was somewhere in the range of 50% - give or take.
I rest my case therefor with Wipro, Tata CS, Accenture, Atos Origin, Logica, and Capgemini - in that order of the foodchain. Whatever EPS they hand out, my per-employee revenue and profit show who's boss.
It also raises a few eyebrows at IBM, who's in the services business for 57% of their revenue, and HP, who makes 32% of their revenue out of services - is it so costly to develop hardware and software?
It's not a perfect measure for firms dealing in hardware or software - I'll admit
But I don't believe in stock and share telling analysts everything about a company - if that situation lasts for a while, companies will simply game the system and provide the analysts with fodder: there's always a few places you can withdraw money from if need be. The budget / realisation game is well-known too, as is expectation management
But, if anyone feels to do the calculations, please be my guest - I wrestled my way through 82 financial reports, comprising an estimated 15,000 pages, and have them all "right here" - 300 MB raw. No pun intended by the way
One thing is for sure: all companies "in my order" are ordered top-down by R&D per employee. Coincidence? I don't think so. Apart from effectiveness and return-on-R&D of all that money, I think it simply shows the mindset: invest in the future of the firm at all cost.
Another thing that is clear, is cost: the more hardware you relatively sell, the bigger the cost per employee. Look at Apple, HP - and Google. The latter doesn't sell hardware but sure as heck has an awful lot of it running smoothly.
If you look at software, Microsoft is the odd one out at cost compared to Oracle and SAP, but they're in entirely different markets and the profit-per-employee is excessively higher as well
Anyway, this concludes these series. I'm not an accountant nor professional analyst but merely a figure-fetishist, and these stats are here to stay - and I'm sure some of it will be used for some purpose: I certainly don't expect e.g. Larry Ellison trying another Open World R&D suggestimation in this year (but made sure of that already)
But if you're an analyst, just let me ask you this: if you're playing this fixed game with static rules, are you the gamer, or gamed?