The Regional Cost of an Engineer, and Software Vendors

People outside of Silicon Valley (or more accurately, the whole San Francisco Bay Area) marvel at the levels of Bay Area salaries. People who’ve spent a long time in the Bay Area scoff at salaries elsewhere.

There’s some interesting things that come with the cost of an engineer. I’ll leave aside the flame war. Instead, I want to talk about vendors and SaaS (Software as a Service) solutions.

I work at a company in the Bay Area, and we desperately need a MacGuffin, which VitalVendor happens to sell a product for. This product’s cost is quantifiable in engineer hours - let’s say engineer hours per year, assuming I have a good guess at what the average engineer costs. I can also extrapolate that cost into % of engineer hours, factoring in the size of the company. Then, add a bit to account for integration work and management.

Now, on the other side, I can estimate the cost of building what the vendor provides, summing:

  • Engineer hours to build viable solution.
  • Engineer hours to operate, fix, and extend.
  • Infrastructure/material cost to build and run.

Now, I have a cost for renting, and a cost of building (and an idea of upfront vs ongoing costs). On top of that, I can also roughly extrapolate opportunity cost - IE what kind of value am I losing based on the headcount that I’m taking away? If immediacy or headcount is an issue (especially if skills are required that are not readily available), I’ll skew toward renting. If custom requirements, or major dependencies on the solution are an issue (read: it’s mission-critical and potentially risky to outsource*), I’ll skew towards building. This is simplified decision making, but the point stands.

Here’s where geography comes in: an engineer in the Bay Area overall costs 1 million US a year, if anecdote is to be believed (I have some trouble believing it, but it’s an approximation, and the point still stands even with something like a “mere” 500k). In my hometown of Smallesville, I can probably peg an engineer-year costing 100k US, to give a nice number. True, you won’t get the person worked on MacGuffins at Google, but that’s 10x the headcount. And given software’s increasing skew towards operations and glue over out-and-out building solutions, a bum in a seat is a bum in a seat.

VitalVendor can sell me a MacGuffin integration for 0.5 engineer-years. It’s important to my team’s productivity, and would be a big project to build, so I gladly sign on.

My friend Susan Lastname back in Smallesville also needs a MacGuffin. However, her company can’t justify the cost. 0.5 engineer-years to me is 2.5 engineer-years to her. She’d have to majorly reshuffle budgets then freeze hiring.

So, what does Susan do? She has her team hack together something that mostly works. It takes a few engineer months, and gets in the way of productivity and reliability sometimes, but her team keeps moving.

Susan Lastname and her team are many teams outside of the Bay Area bubble. Vendor prices target companies with a lot of money to burn. And, note that in this story, Susan doesn’t have her team build a full equivalent of VitalVendor’s MacGuffin product. They have specific needs, and the MacGuffin is just a checklist item so they can do their work. Sometimes, a company builds their own full thing. Sometimes, they hack things on top of a cheaper vendor or open source solution. And sometimes, they go completely without.

* You don’t only need to worry about reliability and security of vendors, you also need to worry about the future. They could shutter a key feature, or double your bill.

Vallery Lancey
Software/Reliability Engineer

I work on distributed systems and reliability.