Filling out forms for $.80/hour

I read a while ago about the single-question customer-satisfaction survey, one that’s basically described here. Basically, the question is: “Would you recommend us to someone else?”. It’s simple, it’s easy to take, and it provides valuable feedback. The single question is a pleasant surprise to customers who may be used to…ah…the other kind of surveys, which I just noticed the linked document also talks about. This latter kind seems rather common in “real life”.

A week or two, I ate at Chili’s, and noticed their receipt’s bold offer to enter me in a contest–if only I visited their website and filled out their brief satisfaction survey. Although the odds may be vanishingly small, I had a spare moment, and visited the site, entered my code, and hit “go”.

A screenfull of answers later, I hit the “next” button. The survey’s progress indicator moved slightly toward completion.

A minute and another screenfull of answers later, I hit the “next” button again, and the progress indicator went up by another twentieth.

I stared in disgust at the progress indicator, closed the survey window, and got on with living my life.

I laud Chili’s for wanting to satisfy their customers. Entering them into a drawing when they fill out the surveys is a nice touch, and might raise the response rate just a bit. I question, though: why would a restaurant even want feedback from the kind of people who spend 20 minutes filling out a mind-numbing survey to get an almost-zero chance at winning a drawing? Do they really need to know that the presence of three softball teams made service a bit slower than usual, the iced tea was good (5, totally wholeheartedly without question agree), and that the experience was, on the whole…quite ordinary–but that the survey-taker has an uncontrollable urge to force a salt-shaker down the throat of the sadist that created the survey?

The next question: why would a person take the time to fill out one of these things? Wal-Mart, it seems, has followed in the blessed footsteps of Chili’s, and is printing magic codes and survey URLs on their receipts. The prize for their contest: they’re giving away five $1,000 gift cards! If we assume that the contest gets 100,000 entries (I’m guessing that number’s low, even given the torture one has to go through to enter), that’s an “expected value” of $5000/100000, or 20 cents. Wal-Mart cheerfully assures us that the survey shouldn’t take more than 15 minutes, so multiply that by 4 to get an expected value of…80 cents per hour. You’d do a lot better flipping burgers–and have a lot more fun!

It seems the dataset generated by this kind of survey has to be relatively small, and of even less worth for decision-making than its size would indicate. Only the ecstatic customers, customers contemplating Vengeance on the company, or the profoundly bored will run this gauntlet of verbiage to submit a survey. The resulting dataset must be hideously distorted. I’m not sure that one question is really all that’s needed on a survey, but to companies who want to give me virtually nothing for 20 minutes of my time: you’ve gotta be kidding!