For a majority of my professional life, I worked at a large company that sold consumer electronics online. I did various types of marketing analytics: analyzing sales to determine how to incent salespeople, building models to pick which customers to pitch which products to, then several different jobs setting prices for different types of products. Then, just over a year ago, I left the company to take a job managing pricing for a moderately large chain of quick service restaurants.
There's been a lot to learn moving from electronics to food service -- but the biggest change, given that I deal with data and prices, has been going from working with a company that sold virtually all its products via the web or phone to a company in which sales depend on what's going on at several thousand individual locations scattered all over the country.
For someone who spends his time trying to figure out how to set prices based on sales data, this takes a lot of getting used to. At my previous company, I could put an item on sale and within hours see the sales pick up. The vast internet, and the ease with which people using shopping portals search out the lowest price on expensive purchases like electronics, made it incredibly easy to measure and predict the effect that setting the price of a given product at different levels would have.
Here things are very different. Not only is making a change harder (changing the price of a product involves getting the crews at hundreds or thousands of locations to all make the right change to their menus on the same day, getting them many materials they need to make that change, making sure they have the supply of the featured items to meet increased demand, etc.), but the results are much harder to read. If the results at a couple dozen locations are terrible in response to a promotion, is that because something was "the wrong price", or because there was construction in the neighborhood making that location hard to get to, or because the assistant manager left and things were run badly for a couple weeks while people scrambled to re-organize the team, etc. Not only can it take a lot of extra work to figure out some of these extraneous factors -- because of scale and the ability of the people at each location to explain what they're running into, you never actually know all of them.
It struck me recently that one of the appeals of the online world is that it presents a sort of streamlined mental universe, in which rivers of data flow smoothly and are little interrupted by local particularities. Having been designed by people who think in numbers, the internet is particularly congenial to those who like the world to be quantifiable. I think at times there's a tendency among those of us who try to understand the world via data to make the rest of the world work "more like the internet" simply because the internet is made to work in ways more easily comprehended by out methods -- a bit like preferring fiction to real life because fiction tends to have a comprehensible plot.