Yesterday Gordon commented on Silicon Alley Insider’s back of the envelope calculations for Google’s food costs. IMO they’re not doing the right numbers. I spent a fair amount of time fighting (somewhat unsuccessfully) this sort of backwards thinking, so here’s a copy of the back of the napkin numbers that I just did. Here’s a copy of what I posted on Gordon’s site:

I think they’re not doing the right numbers. GOOG headcount is at 18,000. If half the workforce ends up working for an extra 1 hr/day (it’s probably closer to 2) b/c of breakfast and dinner, and assuming that an FTE hour is worth $50 to Google (probably low since the fully loaded cost for the avg knowledge work is probably about $100/hr, and the Goog’s profit margin is >25% and that’s including all operating costs), that means:

9,000 employees * 251 days * 1 hour * $50 = $113M

If it does cost $75M (again, we’ll use the worst case number, even if it’s more likely 70% of that), we’re talking about a 50% return on investment. Since we’re using worst case numbers, the direct returns are probably much higher.

And that’s before calculating *any* second order benefits like: better retention, increased loyalty/productivity, increased morale/quality of life, increased reputation/easier hiring and recruiting – and of course free write ups. By almost every set of these important (but harder to quantify) metrics, they is a non-trivial qualitative improvement. Even if it were a complete wash, any knowledge company that could^* would be (and are IMO) stupid for not following suit. Any proper calculation would properly include/calculate offsets for these externalities (certainly recruiting/retention cost projections would be easy to model).

Finally, speaking from personal experience, I doubt SAI realizes how disruptive and hard it is to replace (recruit/hire/train/etc) high performing engineers (especially w/ the huge ramp up w/ the custom systems and processes at a cutting edge tech firm)? Does anyone realize how non-linearly time scales and what it means in terms of development productivity (talking to lots of people, and certainly for myself, I find it takes me a couple hours for me to ramp up into a flow state, after which I get more productive (until a certain point – but again, depending on blood sugar)). Anyway, I’ll stop railing against stupidity. I’m sure Google has run the numbers themselves. They do love doing that, so I hear.

Based on my napkin calculations, and the failure of other organizations to grasp the work environment tactics that Google employs, my conclusion is that everyone (employees, the corporation, and the shareholders) at Google are very much enjoying a “free lunch.”

(* Where “could” means a company that could successfully profit from the additional man-hours and the reduced turnover costs. While the former may be a smaller slice, I would suspect that once you add retention, almost every company would be over the line.)