One of the important biases in business is the sunk cost fallacy – the tendency to throw good money after bad. For example, you’ve spend tens of thousands on developing a new product, but it’s still not working. A common thing to do is go on with the development simply because “you’ve already spent so much on it”. However, what should matter is the future: is more money likely to make it happen? The past is irrelevant – that money has already been spent.

Surely, watching over employees should reduce this problem?

Unfortunately, not necessarily. While some research tends to show is that being accountable for your choices makes you less susceptible to sunk cost fallacy, sometimes accountability makes the effect even worse! Research is mixed on this, but for now I’ll accept that accountability is not the magic bullet. Well, there have been other ideas for reducing the fallacy.

Discussions around decision making often tend to lead to the question “How can I leverage this in my own life?” Unfortunately, behavioral results are not the easiest to apply in the everyday. Sure, knowing about biases is good, especially when you’re making that big decision. But in all fairness, loss aversion or the representativeness heuristic are not usually the biggest worries.

For me, personally, the biggest worries revolve around one question: Is this really worth it? And no, I don’t mean that my mode of being is an existential crisis. What I mean is that I often find myself asking whether this particular activity is worth my investment of time and energy. This meta-level monitoring function is a direct result of the two following concepts.

Opportunity cost


If you’ve studies economics or business, you’ve surely heard of this. If you haven’t – well, you might be missing one important hammer in the toolbox of good thinking. As a concept, opportunity cost is really simple. The opportunity cost of any product, device or activity is what you don’t get instead. For example, if I go to the gym for an hour, I’m giving up the chance to watch an episode of House, for example. Of course, there are all kinds of activities one is giving up for that hour, but ultimately what matters is the best opportunity given up – that’s the opportunity cost.

Why I consider this to be important is that it’s the ultimate foundation for optimization. When one thinks about activities in terms of opportunity costs, it makes concrete the constraint that we all experience: time. No matter how rich or powerful you are, there’s always going to be that nagging limit of 24 hours a day. So it pays to think about whether something is really worth your precious time.

Marginal benefit

Marginal benefit (or utility) is also quite simple. The marginal benefit of something is the benefit you get by consuming an extra unit of that good. For example, at the moment of writing this, the marginal benefit of a hamburger would be quite high, since at the moment I’m pretty hungry. What’s important is that the marginal benefit changes over time – it’s never constant. One burger is good, and two maybe even better, but add more and more burgers on my desk and I’ll hardly be any happier. In fact, anything over three burgers is a cost to me, since I can’t possibly eat all that – I’ll just have to carry them to the garbage!

What makes marginal benefit powerful is the idea that even though I’m enjoying something, it doesn’t mean I should take in all that I can. A night out is great fun, but perhaps after a few pints the marginal benefit often plummets quite fast – you can try this by staying in the bar for extra two hours next time. Just remember to evaluate the situation next morning! ;)

These two concepts help you to ask two things. How much are you getting out of this? What could you get instead? And if the answer is that there’s something more you want instead –well, that’s a wonderful result! At least now you know what you want! :) Or, well, until the marginal benefit decreases, at least…

So, back in September I wrote about one basic bias: context-dependent preferences. At its core, context-dependent preferences mean that consumer choices are based on relative value, instead of absolute value. So, instead of assessing which option is best on its own, we tend to look at which option is the best, relative to others in the choice set. With strategic selection of the items, a marketer can thus heavily drive us to choose a middle option that sits between a cheap and an expensive one. This is known as the compromise effect.

Now, Simonson is saying that the compromise effect isn’t real anymore! I picked up his claim from his talk at TEDxBayArea. Here’s the video:

So, Simonson has replicated with Taly Reich his previous experiment that showed the existence of the compromise effect. He shows that the new study was pretty similar to the old one, and that the compromise effect just wasn’t there. Unfortunately, since the study hasn’t been published yet, we just have to take Simonson’s word of this. But let’s do that, and see what the implications are.

His main point is that the compromise effect doesn’t surface because modern technology helps us estimate absolute value instead of relative value. Twenty years ago, we would proxy quality with attributes such as brand, country of production and so on. These days, we have access to a host of online services that look at the product quality. Some of them directly assess products, while others provide just customer reviews. Irrespective of the way of assessment, both are pretty decent ways of assessing quality.

So what does it mean? Well, Simonson comes up with a list of commonly believed statements about marketing that he proceeds to debunk:

1.       A company’s brand is more important today than it has ever been

Quite a few people say that precisely because there’s so much information out there, branding is even more important today. Because of information overload, consumers can either process all the information, or ignore the information and rely on brands. However, Simonson argues that we can combine the approaches: quickly glance at reviews to assess quality. Brands will still exist – for example as status markers – but they will not be effective proxies for quality.

2.       Nurturing customer loyalty is the best investment

Customer loyalty is usually thought to be a sound investment. Satisfied customers will buy your products again, and thus loyalty will act as a barrier against competitor. Simonson thinks this is less likely to be true now. He claims that each product is evaluated on its own: he’s not going to buy a Toyota just because he liked the previous car. I think the example is correct, but the lesson’s not. Sure, when you buy a big pure product, you’ll evaluate the attributes with care. But with smaller purchases, brand surely matter. I seem unable to find the paper, but I remember reading that when buying groceries, customers tend to go for the products they usually buy (classic System 1 way of action – status quo).

3.       Market research can predict what consumers will want

Simonson says that since our choices are these days driven much by social media and we rely on opinions of others, then market research should focus on that part, instead of looking at wants and preferences today to predict the future. Simonson argues that the prior wants are really bad predictors – what matters are the sentiments in the social environment.

All in all, I think Simonsons ideas are pretty thought-provoking. There’s a lot riding on the replication failure of just one experiment, which seems a pretty shaky justification for a new theory. However, it is just one TED Talk, so I hope there’s more to it. Apparently, Simonson has also written a book about this. Well, that’s one more book for my pile of books to read...