Choice Modeling

20 years ago   •   5 min read

By Marcia Kadanoff

Today’s business environment sometimes feels much like a reality TV show. The stakes are high, the contestants are mostly good looking, amazingly fit, and overqualified for the tasks thrown at them. Part of the allure of watching a reality TV show is the knowledge that the majority of the contestants will fail and fail miserably. This kind of thing has been fun to watch for centuries, and leaves the rest of us feeling much better about our prospects in the morning.

Watching a train wreck happen makes for good television but is hardly a business strategy. Yet many companies facing low or no growth in their core markets seem stuck in do nothing mode. A situation we find hard to fathom. It’s like watching The Apprentice except that everyone might get fired in the end.

At Firewhite, one of our favorite conceptual tools is (no surprise) a chart which depicts the risks and rewards of different strategies for growth.

Many clients are so risk-adverse that their preferred strategic mode is to do nothing. Doing nothing seems like it should minimize risk, but in today’s volatile and hypercompetitive markets while you’re sitting still you can be sure your competition is moving forward. So we typically assess “do nothing” with a risk score of +2. A strategy of “doing nothing” is likely to end badly, with your business getting smaller in the process. We indicate this here with a reward score of -1.

Optimization – meanwhile – requires a modicum of risk. When you optimize, you’re polishing: making minor, incremental alterations to deliver slightly better results. The risk is limited (+0) because you’re tinkering with something you thoroughly understand. The reward is likewise limited (+0). Bringing out a line extension in a familiar market is an example.

The middle of the spectrum is innovation: trying something new, perhaps to meet customer needs you’ve never really addressed before. This is a bigger degree of change than optimization, and involves proportionately more risk (+1). Here we’d assess the potential reward at +1. Case in point: Burberry, the maker of traditional plaid-lined raincoats. Last summer, Burberry innovated by moving into a new market, one where fashionistas rule. The company brought out a line of raincoats in popsicle colors: bright pink, neon green but still lined with its distinctive plaid. The results of this move are impressive: earnings +15%; stock price +50%. Proving that innovation can turn a staid performer into a growth stock.

In the upper-right-hand-corner right on our chart is change. These are the bet-the-company strategic decisions that alter the most basic elements of your business: what you’re selling, what it does, who you’re selling to, what needs it meets, and so on. Change is scary but also offers the biggest reward. Change often happens by moving into adjacent markets, where both the customers and the needs you are satisfying are new. Case in point: Apple Computer has added $100M to its bottom line through the introduction of iTunes, a downloadable music service.

On Risk

Decision makers in business have one thing in common with your average game-show contestants: they’re human and they don’t like risk. The devil you know is always preferable to the one you don’t, so human nature pushes decision makers towards the left side of the spectrum, toward a strategy of “do nothing” or “optimize” what we have today. Innovation happens rarely. Change almost never. Which means the really big strategic wins are taken off the table, because the cost of failure is simply too great.

At Firewhite, we work with our clients to put change and innovation “back on the table”. We don’t have a magic wand we can wave to make risk go away. What we do have is an arsenal of techniques in customer marketing and analytics, among them something called choice modeling.

Choice modeling combines market research with simulation. It was invented in the 1970s by Daniel L. McFadden (University of California, Berkeley) who won a Nobel Prize for this work. CM is used to determine the optimal price and/or feature set to maximize profits, contribution margin, and/or market share. In-depth analysis of the data collected can also be used to understand how you can build more sustainable advantage into your product or service offering.

Choice modeling isn’t cheap — it utilizes specialized practitioners, software and modeling techniques. That said, it can give your company an edge by reducing the risk of innovation and change, particularly when your strategy could greatly affect revenues, earnings, and market share.

On Choice Modeling

In the old days (up until a few years ago), the only way to remove the risk from a bet-the-business decision was by fielding an in-market test. In other words, go through the time, expense and complexity of developing your entire program for change and rolling it out in a single test market to see how it works. If it fails in a test-market scenario, don’t roll out your innovation in other markets, effectively killing the idea. The test market serves as the canary in the coal mine, so to speak. This is an expensive (and slow!) way to reduce risk.

Contrast this method with the type of virtual test market we can field using choice modeling. Here, you can take your next big idea and reduce it to its component parts. Add or remove ingredients. Change the price. Bundle it with other products. Change the packaging. Sell it through completely different channels. In other words, break your big idea into a set of variables that can be combined in a myriad of permutations.

Next, we expose the target customer to your big idea. To do this, we use an online survey and ask the target to choose between various products (or bundles of product) that vary in price or value. Responses are used to create a choice model depicting how demand for a given product varies with changes in elements of your value proposition.

By fielding virtual test markets, we can simulate how the market is likely to respond to your new offering. We don’t pretend the results are an accurate way to judge absolute demand for your product or service.0 Virtual test markets are best used to shed light on a relative basis, to tell you which combination of features and benefits is most likely to meet with success in market.

With a virtual test market, you can take the level of risk involved in change down to the level seen with innovation. Or, you can innovate, and reduce the risk to the level of optimization. Stop and ponder the implications of that for a second.

Virtual test-markets aren’t without limitations. You need to target a customer we can reach through internet marketing and/or direct mail. You also need a solid value proposition that can be articulated in specific terms. The results of choice modeling are simulations of how consumers will behave when they know all the alternatives available to them, which isn’t always how the real world operates. Also, choice modeling doesn’t shed much light at all on things like the size of your available market. Digging into some of these issues may require additional in-market testing.

However, what choice modeling can do is give you the information you need to make big, market-shifting bets, with less risk. Virtual AND in-market testing is the equivalent of getting up in the morning and packing your briefcase with a trail map, a census of the animal life, and a good pair of boots. After all, if you’re going to compete on Survivor we want to make sure you will win — and the best way to do that is by giving you an unfair advantage.

Note

  1. Is performance on The Apprentice an accurate predictor of on-the-job performance?

Originally published on Firewhite Consulting site, 2.04.

Spread the word

Keep reading