My daughter needed a car to drive to college. It needed to be safe, cool, and inexpensive, among other things. Too bad there’s no really cool, really inexpensive car out there, so, once again, we needed to make trade-offs. We started by listing the major goals and considerations. This turned out, as usual, to be a mix of objective and subjective factors. We then identified some cars that seemed to be good choices. We filled in a matrix evaluating each car against each factor. It looked something like this (using made-up cars for the example).
This alone is a useful exercise and often gets you to an answer. You’ve made sure you’ve considered all the important factors. Sometimes a clear winner pops out, but not this time. Each of the choices tops in at least one consideration. How can you compare these? Some of the factors are very specific and others are subjective. They involve big numbers, small numbers, or words. Sometimes more is better and other times less is better. To make things even more confusing, the factors are clearly not all equally important.
I used my all-time favorite tool for making choices. I’d been using it for years before I found out it had a cutesy acronym, SMART, for Simple Multi-Attribute Rating Technique. The first step is to make everything numbers. For example, yes is 1 and no is 0. Appearance is on a scale from yuck to cool, with yuck being 1 and cool being 5. Here’s what I got.
A simple weighted average (see earlier post) won’t work here because the numbers are so varied, but SMART is designed for problems like this. You can do the calculations manually—they’re simple but tedious—or you can use this spreadsheet, choose-a-car. You can change the factors and alternatives for whatever decision you’re considering. If you have fewer of either, just delete the extra rows or columns. Here’s how to do it.
Step 1. For each factor, enter the worst that you would consider and the best that you would reasonably expect. For example, I can’t get a new car for less than $9000, so that’s the best price. I’m not about to spend more than $20,000 for a car for this kid, so that’s the worst price. The consumer rating is on a scale of 1 to 10, and I don’t have enough faith in it to throw out a choice based on it, so the best is 10 and the worst is 1. Here’s what I got.
Notice that the worst and best are chosen without looking at the values for the choices at all. This helps you put the differences in perspective. Here, the consumer ratings are practically the same. Even the prices are very similar. Putting this all in the context of best and worst keeps me from worrying about minor differences.
Step 2. Next, for each factor, rate its importance using any scale you wish. Here’s where SMART really shines. Most decision techniques ask you to rate importance without reference to anything—what’s more important, price or appearance? It’s a meaningless question. How much price? How much appearance? SMART anchors it to something. The key is to compare the value to you of moving from worst to best. For example, ask, “Would you rather have the price go from $20,000 to $9000 or have the warranty go from 3 years to 6 years?” I rate the importance of the warranty only 10 and rate price 50, since I would much rather save the money. I give each of the other factors a preference rating similarly. It doesn’t matter what numbers you use, just so the ratios represent your preferences. For example, price is five times as important than warranty. Here’s what I got.
Step 3. Finally, we’re ready to do the calculations. The easiest thing is to use the spreadsheet. Here’s what I got.
The scores for each choice are at the bottom. The Kumquat comes out on top with a score of 0.649, or 64.9% of perfect. If you didn’t get the winner you expected, check your preferences. If you change them, the answer could change. By the way, I had my daughter put in her own preferences, which included a lot for appearance and absolutely nothing for price, and got the same answer. Whew!
Where did these scores come from? What you see listed under each choice are the normalized scores. This tells you where the option lies between worst and best. For example, the prices are all about in the middle, the consumer ratings are all really good, and the appearance ratings are all over the map. This helps you see which of these factors is a big deal. If you’re interested, the formula to do this is (score – worst)/(best – worst). The score is the sum of these values normalized by preference. By the way, if you want to do this manually, or are just interested in the underlying formulas, here’s how it works. Get the weightings by dividing each preference score by the sum of the preference scores. This gives you weightings that add to 1. Then to get the final score for a choice, multiply each factor score in the column by the corresponding weighting and add them all up.
I’d be interested in hearing from you if you use this great technique to make a complex choice of your own. Any questions? I’d love to hear from you.