Innovation Science

Read This Post for Free!

People often make irrational decisions when money is involved. They’ll go out of their way to save 25 cents, say by driving around and around looking for a parking meter with time remaining, yet immediately after blow that much and more without a thought ($4.50 for a latte, anyone?).

Dan Ariely explores such irrational behavior in his book Predictably Irrational. Ariely is a business school professor at MIT, and much of the book describes simple yet ingenious experiments he’s conducted that demonstrate over and over again the consistently illogical behavior of people when making choices. This field of research, known as behavioral economics, stems from the ground-breaking research of Daniel Kahneman and Amos Tversky’s that won Kahneman the Nobel Prize in economics (I’ve been citing Kahneman for years in my talks on cognitive attention and effort, but it seems that every week for the past year I’ve read Kahneman’s name referenced in a different news story, book, or magazine article. Behavioral economics seems to be the hot topic these days). Ariely’s experiments demonstrate the fascinating ways in which people repeatably make decisions that appear to be contrary to common sense.

Ariely details, for example, the powerful allure of something being free. People’s choices change dramatically when items change from being effectively free (costing one cent, for example) to being actually free (costing zero cents). For example, he tells how Amazon’s customers bought more books on average once Amazon started shipping for free on purchases greater than $25—the increase in customer spending far exceeded the amount that they were saving on the free shipping, but they gladly spent more on books to get shipping for free. More puzzlingly, Amazon saw this behavior exhibited worldwide except in France. Was this because the French were more logical in their purchasing behavior? No. The country manager in France decided not to make shipping free but to reduce it to one franc, or 20 cents, which wasn’t a small enough shipping amount to change people’s purchasing behavior. Is 20 cents really a significant amount when buying a $20 book? No. Is a shipping cost of 20 cents effectively the same as 0 cents? For most people,yes. But when the manager in France reduced Amazon’s shipping charge from 20 cents to Free, French customers suddenly increased the amount of books they purchased per order by the same amount as the rest of the world.

The reason that I am writing about Predictably Irrational is to contest an explanation that Ariely gives for his first example, and one of the most interesting, of irrational behavior. In the example, Ariely describes an offer that I’ve seen before but never understood, which made me even more interested in his explanation (and my alternative one).

Ariely noted an ad in the Economist magazine that offered three different types of subscriptions:

1. an electronic subscription for $59;
2. a print subscription for $125;
3. a print and electronic subscription for $125.

That’s right, the price for a print-only subscription was the same as the price for a print&electronic subscription. Why would they bother offering the print-only option when it was clear that no one would select it because they could get both the print&electronic versions for the same price? Being the good researcher that he is, Ariely decided to run some experiments to determine if there was a reason why the Economist made this offer (Ariely’s ability to set up simple experiments to provide insight into unusual behavior is fascinating).

Ariely asked two different groups of students to make a selection among a choice of Economist subscriptions, but each group were given a different list of options. One group was given the same three selections that the Economist offered. The percentage of students who selected each option was as follows:

1. electronic: 16%
2. print: 0%
3. print&electronic: 84%

No one chose the print-only option (not surprisingly). This begs the question of why one would offer that option at all (the same question that motivated this experiment).

For the second group, the unchosen print-only option was removed and only the electronic subscription and print&electronic subscription were offered (still priced at $59 and $125, respectively). One might expect results similar to those from the previous group (16% for electronic and 84% for print&electronic) since there wasn’t any interest in the print-only option that was removed. The results for this second group were as follows:

1. electronic: 68%
2. print & electronic: 32%

Without print-only as an option, the number of people interested in the print&electronic option dropped from 84% to 32%! In other words, by offering an option that no one wanted, 52% of the people changed their preference from electronic-only to print&electronic at a $66 higher cost. What’s going on here?

Ariely’s explanation was that people need a basis for comparison when evaluating whether a deal is a good one or not, and from this comparison they make their decisions. With a similar item for comparison, they can assess the value of an item and determine whether that item is a good or bad deal. He gives a couple fascinating examples where the choices that people make are biased by the presence of a similar and inferior alternative. If people are choosing between A and B, they will choose A more often if there exists a third alternative similar to but inferior to A, and they will choose B more often if there exists a third alternative similar to but inferior to B. Having that third alternative allows people to say to themselves, “I can’t judge whether A or B is a better choice, but I know that A is better than this similar third alternative, so A must be a good deal.”

Ariely suggests that this is why the selection for print&electronic is high when print-only is offered for the same price—the print-only option gives people an alternative from they determine that print&electronics is a better deal, and that drives their choice.

I believe, however, that there’s another possibility for these results: people are more likely to select the print&electronic option in the presence of a print-only option because of the Free phenomenon. Let me explain.

Ariely’s explanation is partly correct: in the absence of knowing the cost of the print-only subscription, people cannot judge whether the combined subscription is a good deal or not. Maybe print-only was $66, and the cost of print&electronic was simply the sum of the costs of the individual subscriptions ($59+$66=$125). Or maybe print-only was $100 and getting the combined subscription would save $34. Or maybe print-only was also $50 and $125 for print&electronic was a rip-off. No one can tell whether both together is a good deal in the absence of a print-only option—in that, Ariely is correct.

The fact that the print-only and print&electronic options are priced the same, however, suggests another explanation for the large difference in behavior between whether or not the print-only option is offered. Ariely made the convincing case later in his book that getting something for free can have a hypnotizing allure on people’s decisions. Providing a print-only option at less than $125 probably would not be enough to drive such a large percentage of people to change their selection from print-only to the more expensive print&electronic option—the print-only option had to be priced at exactly same same price as print&electronics so that the electronics version was free if they chose both. It was the identification of something free that made such a large percentage of people select the print&electronic option. This is similar to the phenomenon observed by Amazon: 20–cent delivery didn’t change behavior but free delivery did. Given the convincing case that Ariely made for this free effect, one would probably have predicted that people would switch their decision from the electronic-only option to the print&electronic option when they discovered that the latter choice would get them the electronic version for free.

So, now you know why so many ads offer free cheap items if you purchase an expensive product A free month of HBO if I get the Lifetime Triple Gold package of cable—sign me up!

The Physics of Pricing

The Associated Press had a great article that I read today (in the SF Chronicle) on the use of advanced mathematics to help determine product pricing strategies. So-called price-optimization tools from companies such as Khimetrics (acquired by SAP) and Zilliant analyze massive amounts of historical sales data to find clues for better pricing that would be difficult to discern without their sophisticated models. (These software systems cost in the 7–figures—assuming that Khimetrics&Zilliant apply their own analysis to their own pricing in a kind of post-modern self-reflexive business analysis, their customers must be hugely self-conscious about the “optimization” of their bill).

This whole field of analysis in part began, according to the article, when Khimetrics’ founder, Ken Ouimet, was studying complex systems in a university Physics department and he epiphonied (which should be verb) that shoppers have no more sense than a hydrogen atom (or something like that). Next thing you know, mathematical models developed to simulate the motion of atoms in gas are being used to model the purchasing behavior of beer-and-Doritos-buying consumers.

The article and its sidebar have several interesting examples of the successes  of these systems. The systems determine, of course, which products have price elasticity and which don’t. That’s not necessarily difficult for experienced retailers to figure out. What is difficult is to fine-tune pricing on a huge inventory of widely differing categories of products with barely discernible differences within categories.

What’s even more interesting is their analysis of relationships between products and subsequent recommendations to exploit correlational behavior. Beer drinkers, for example, will pay careful attention to price when buying their brew (“10 cents cheaper? Mickey’s Big Mouth for me!”) , but will snatch up snacks to go along with the beer with little concern for cost. So, drop the margins on beer and crank them up on Cheetos—balance this adjustment correctly and you’ve just increased your bottom line.

And guess what? Those consumers care even less about price during big sporting event weekends—turn the pretzel-price up even higher during NCAA Tourney action! Increase the price of mint leaves during Kentucky Derby weekend and you’re golden. Who would have thought that sophisticated mathematics could improve the profitability of such unsophisticated businesses as Safeway’s and Albertson’s.

Personalization/Individualization is a consumer theme that has hit practically all consumer markets, with the extreme perhaps being the ability to completely customize your Nike shoes online so that no other shoe in the world is identical to the pair that you buy. This is normally viewed as empowering for consumers: a good thing.

Consider now the logical continuation of price-optimization. With store cards (e.g., Safeway cards) that people use to take advantage of item discounts, companies are amassing large amounts of personal information that can be correlated with buying habits. Cross-correlate this data with other databases that can be purchased from other sources, and the ability to personalize pricing becomes incredible once price-labeling become easily and quickly changeable (be afraid when grocery store prices are shown with LCD displays).

Herein lies the dark side of too much power from too much information.

Your price-optimization consultant tells you that students coming back from the bar after midnight don’t care too much about the price of frozen pizzas? Nudge those babies up by 15 cents every late-evening and watch your profits climb. Obese people less sensitive to the pricing of chocolate truffles? Put a weight sensor in the gourmet candy aisle. Mercedes drivers less discriminating towards wine prices? Do I really have to spell out for you what to do?

The AP article I mentioned at the beginning of this post predicts that store prices will become more like the mystical pricings of airplane tickets, and I doubt many consumers will relish that thought. Have you ever heard anyone say they wish other businesses priced their products like the airline industry? Where two people sitting beside each other who bought their tickets on the same website pay wildly different prices because their purchases were on different days? Imagine that you are in the register line at Virgin Records with the latest Bond DVD in hand, and the person in front of you buys the same DVD causing the price of your identical DVD to increase by $1. Welcome to the world of airline pricing strategy!

The days of setting prices based on a fixed margin, on prices from competitors, or simply on an incremental increase over last year’s prices is becoming a fading, quaint tradition. Welcome to the Machine.

I sound cynical, yes. But still…

I have to appreciate the evolution of process sophistication. Of modern thinking challenging and overcoming the wisdom of experience (goodbye Willie Loman). This is the essential nature of science, of business, of innovation: new ideas obsoleting the currently accepted lore. I particularly like seeing advanced mathematical theory being applied to such innocuous business processes as the pricing of ketchup. Luddites take note!

Truth be told, I’ve always had a warm spot towards the practical application of esoteric mathematical theory. In fact, since my college days I’ve had a warm spot towards the mathematical discipline of Information Theory (welcome “information theory” googlers!), probably because I tried a long time ago to apply it to neural signals and failed completely (but others eventually succeeded).

Information theory tells us, well, how much information is in something and how much information can be transferred by a transmission channel. How informative is the weatherman in San Diego when all he predicts in every forecast is that the weather the next day will be sunny and in the 80s? Even if is accurate 95% of the time? Information theory would tell you that his information content is low because there’s not much information in declarations of a near sure thing (Listen and be astounded: I am here to tell you that you will take a breath in the next 60 seconds! See?! How amazing is that?).

That being said, my prediction is that the next New Thing in business/marketing will be an information theoretic approach to marketing. Some marketing channels will be proven to have much more information capacity to target consumers than others, and the capacity of each channel will be calculated to the nearest bit, allowing companies to charge millions of dollars for advice on which channel will provide the highest information ROI for your marketing message. Advertising a NASCAR race on American Idol, or promoting hearing aids in Mad Magazine? That’s less than 1 bit of advertising information. Advertising for either product in Golf Magazine, however—that’s called maximizing your channel capacity.

Informationtheory.com is already taken. Marketinginformationtheory.com is not. What’s your guess on how long until all of the latter’s related domains are taken?

Simplicity Creates User Satisfaction

An article in the latest Harvard Business Review titled Defeating Feature Fatigue reminds me of a post at Creating Passionate Users on breakthrough ideas. The Harvard article goes into analytical detail on how companies continue to add features to their products but in the process severely hurt the usability of their products. Customers may be more motivated to buy the product that offers the most features, but will be less happy with the product once they realize how unusable the product is. Post-sale satisfaction is maximized with the simplest features that provides the best usability (sound familiar, IDEO?) So, there must be a happy medium that optimally trades off sellability with post-sale user satisfaction. The figure to the right demonstrates the HBR authors' theoretical analysis of this trade-off, indicating that the happy medium is--surprise, surprise--not too many features, not too few. Sounds like the Goldilocks Strategy: the number of features is just right.

Last year, the Creating Passionate Users blog posted a very similar looking curve, which is shown here on the left. Look familiar? The point that CPU made was the same as the HBR authors. Usability has a big impact on user satisfaction, and often simplicity provides the best solution for product design. Having the most features might get customers to buy the product when they are considering different items at a store, but users prefer simplicity and ease-of-use after they actually own a product and therefore simplicity provides the most long-term user value. Users don't want complexity and don't want to have to read the manual whenever they want to use one of the product's features. Great scoop on HBR, CPU!

Select Choices for Satisfied Consumers

Graham Gordon's post Innovate by giving less choice at Broken Bulbs: Innovation inspired me to write my own thoughts on selection size for consumers.

I've thought for a long time that too much consumer choice can be a bad thing. Whenever I read about being able to download any movie or TV show at any time, I envision myself sitting in front of the TV with an hour to spare, and being frozen by indecision because of the overwhelming selection from which I have to choose. Give me 30 channels of actual variety to choose from and I'll find something to pass an hour; give me every show that was ever made to choose from and I'll give up and go do something else.

Gourmet chefs have known this for a long time. In the main dining room at Chez Panisse, the culinary mecca in Berkeley that started the California cuisine movement, diners have no choice at all for dinner. There's one set menu for the night, whatever wonderful menu Alice Waters has decided for the evening. At many restaurants, the most expensive dinner you can have is the chef's tasting menu, where not only do diners not get a choice but they don't even know what the courses are until until they arrive at the table.

My favorite bookstores are not the ones with the biggest selection, but the ones with a few creatively selected books expertly chosen by the staff and prominently displayed to guide me to new discoveries. Give me a small selection of high quality and unique items over a massive selection anyday.

Obviously there are many exceptions to this rule--searching for the best price on something that you know you want is the most obvious place where complete selection is beneficial. When I'm searching for the best price on a flight where I know the destination and date, I want a website that offers me every possible flight available so that I can find the best price. When I'm searching for a vacation idea, however, I don't want a website that gives me every possible vacation choice--I want a site that has selected a few quality options to highlight and whose selections fit my own tastes.

There's obviously value in the Amazon model, but for me there's also considerable value in the narrowed-by-experts small selection model. We are seeing a current craze for simplicity in design, and the standard innovation development strategy of Subtraction has been around a long time--selectively restricting consumer choice is consistent with both of these approaches.