Multitracking

1. Multitracking

Chip and Dan Heath describe parallel exploration in Decisive. The company practices and study results below reflect the book’s account.

In Sausalito, California, there is a small firm called Lexicon that has coined the names for 15 billion-dollar brands, including BlackBerry, Dasani, Febreze, OnStar, Pentium, Scion, and Swiffer. These names don’t emerge from brainstorming sessions that yield sudden lightning-bolt insights—nobody gets struck by lightning 15 times. Rather, Lexicon’s magic is its creative process, which helps the team avoid getting stuck in a narrow frame.

Consider the firm’s 2006 work for Colgate, which was preparing to launch a disposable mini-toothbrush. The center of the brush held a dab of special toothpaste, which was designed to make rinsing unnecessary. So you could carry the toothbrush with you, use it in a cab or an airplane lavatory, and then toss it out.

When Lexicon founder and CEO David Placek first saw the toothbrush, he said, what stood out was its small size. So, if you were on the Lexicon team, with your mental spotlight pointed at the tiny toothbrush, you’d be tempted to start tossing out names that highlight its small size: Petite Brush, Mini-Brush, Brushlet, etc. Notice that, in brainstorming that way, you would have already locked yourself into a tight frame with two assumptions: (1) The name should connote smallness; and (2) “Brush” should be part of the name.

That early lock-in is something that the Lexicon team has learned to fight. Clients will often come to them with a narrow conception of what a good name is. Some at Intel, for instance, had wanted to call the Pentium “ProChip.” Some at P&G had wanted to call the Swiffer “EZMop.” Lexicon has learned that the best names emerge from what we’ll call “multitracking”—considering several options simultaneously.

To get familiar with the new toothbrush, Placek’s team at Lexicon began to use it in their daily lives, and what struck them was how odd it was, at first, not to spit out the toothpaste that it produced. (We always spit out the toothpaste.) Fortunately, unlike normal brushes, the new brush didn’t create a big mass of minty lather. The mouthfeel was lighter and more pleasant, more like a breath strip. It was this lack of foaminess that was the brush’s most distinctive trait. So it dawned on the team that the name of the brush should not signal smallness; it should signal lightness, cleanliness, softness.

Armed with that insight, Placek began to multitrack. He asked his network of linguists—70 of them in 50 countries—to brainstorm about metaphors, sounds, and word parts that connoted lightness. By working independently, they vastly increased the pool of considered names.

Meanwhile, he asked another two colleagues within Lexicon to help. But he kept these two in the dark about the client and the product. Instead, he gave this team—referred to as the “excursion team”—a fictional mission. He told them that the cosmetics brand Olay was interested in introducing a new line of oral-care products, and their job would be to help Olay brainstorm about product ideas.

Placek chose Olay because he believed that beauty was an implicit selling point for the new brush. “Good oral care means white teeth, and white teeth are better looking,” Placek said. After a period of exploration, the excursion team pitched some intriguing product ideas, including the “Olay Sparkling Rinse,” a mouthwash that would make your teeth gleam.

In the end, it was the insight about lightness, rather than beauty, that prevailed. The team of linguists produced a long list of possible words and phrases, and one word on the list jumped out at Placek’s team: “wisp.” It was the perfect association for the new brushing experience. It’s not something heavy and foamy; it’s barely there. It’s a wisp. Thus was born the Colgate Wisp.

Notice what’s missing from the Lexicon process: the part when everyone sits around a conference table, staring at the toothbrush and brainstorming names together. (“Hey, how about ToofBrutch—the URL is available!”)

Lexicon refuses to single-track the process. In fact, in most of its naming projects, Lexicon forms three teams of two, with each group pursuing a different angle. Usually there is an excursion team, blind to the client and the product, that spends its time chasing analogies from related domains. In naming Levi’s Curve ID jeans, which were engineered differently for different body types, the excursion team dug into references on surveying and architecture.

Lexicon’s multitracking often leads to “wasted” work. In the Wisp case, the excursion team found themselves at a dead end with the Olay assignment. But it’s precisely this willingness to work in parallel, and to endure inefficiency, that often leads to a break in the case. That’s what happened with one of Lexicon’s most famous projects: the BlackBerry, made by Research in Motion (RIM).

When RIM engaged Lexicon, Placek and his team knew that they were fighting negative associations with PDAs: They buzz, they vibrate, they irritate us and stress us out. He challenged the excursion team—again unfamiliar with the actual client—to catalog things in the world that bring us joy, that slow us down, that relax us. The goal was to discover names that might offset the negative PDA associations.

The list grew quickly: camping, riding a bicycle, having a martini on Friday night, taking a bubble bath, fly-fishing, cooking, having a martini on Thursday night, and on and on. Later someone added “picking strawberries” to the list. Someone else plucked out the word “strawberry.” But one of Lexicon’s linguists said, “No. ‘Strawberry’ sounds slow.” (Think of the similar vowels in “drawl,” “dawdle,” and “stall.”)

Soon it was crossed out and replaced with the word “blackberry” underneath. Someone else noticed that the keys on the PDA look like the seeds on a blackberry. Epiphany!

Actually, no. The RIM clients were not positive at first because of the frame they’d started with. They’d been leaning toward more descriptive names such as “EasyMail.” Placek said, “Most clients feel that they’re going to know the perfect name as soon as they see it, but it doesn’t happen that way.”

Eventually, the case for the name “BlackBerry” prevailed.

The client’s initial reluctance is instructive, though. Sometimes we’ll know the right option when we see it, and sometimes we won’t. Surfacing another option—even one we ultimately reject—can clarify the strengths and weaknesses of the preferred choice.

The distinction here is between considering options simultaneously and refining a single option before trying another.

A study of graphic designers demonstrates the value of multitracking. The designers, tasked with making a banner ad for a Web magazine, were randomly assigned to use one of two creative processes. Half of them were instructed to design one ad at a time, receiving feedback after each new design. Each designer started with a single ad and revised it five times based on rounds of feedback, yielding a total of six ads. The other half of the designers were instructed to use a “simultaneous” process, so that each one started with three ads and received feedback on all three. Then, in successive rounds, the set was whittled down with further feedback to two ads and then one final ad.

All of the designers ultimately created the same number of ads (six) and received the same quantity of feedback (five ad critiques). The only difference was the process: simultaneous versus one at a time.

As it turned out, process mattered a great deal: The simultaneous designers’ ads were judged superior by the magazine’s editors and by independent ad execs, and they earned higher click-through rates on a real-world test of the banners on the Web site. Why?

The study’s authors, trying to explain the better performance of the simultaneous designers, said, “Since [simultaneous] participants received feedback on multiple ideas simultaneously, they were more likely to read and analyze critique statements side-by-side. Direct comparison perhaps helped them better understand key design principles and led to more principled choices for subsequent prototypes.”

In other words, the simultaneous designers, by multitracking, were learning something useful about the shape of the problem. They were able to triangulate among the features of their three initial ads—combining their good elements and omitting the bad.

You may recall this is the same logic used by Steve Cole—the “think AND not OR” guy from the first chapter—in explaining why it’s helpful to hire multiple vendors for the same project. He said, “You get convergence on some issues, so you know they are right, and you also learn to appreciate what makes people different and special. None of this can you do if you’re just talking to one person.”

Multitracking has another advantage too, one that is more unexpected. It feels better. After the banner-ad study concluded, both sets of designers were interviewed. Asked to rate the usefulness of the feedback they received during the design process, over 80% of the simultaneous designers said the feedback was helpful. Only 35% of the one-at-a-time designers agreed, and in fact, over half of them believed the feedback they’d received was critical of them. (None of the simultaneous designers felt criticized.) The simultaneous designers also reported that, as a result of the experience, they felt more confident in their design abilities. The one-at-a-time designers didn’t agree.

Why was the experiment so frustrating for the one-at-a-time designers? The study’s authors speculated that people who work on a single track begin to take their work too personally, viewing criticism as a “rebuke of their only option.” Or as one of the authors, Scott Klemmer, said, “If I have only one design, then my ego is perfectly conflated with my design. But if I have multiple designs, I can separate the two.”

Multitracking may make criticism easier to use by separating attachment to a goal from attachment to one proposal. The design study supports that possibility; it does not mean that offering several projects will always make a manager receptive to criticism.

So, given the clear benefits of multitracking, what explains the failure of most organizations to embrace it? Many executives are worried that exploring multiple options will take too long. It’s a reasonable fear, though Kathleen Eisenhardt’s research found a different pattern in the teams she studied. In a study of top leadership teams in Silicon Valley, an environment that tends to place a premium on speed, she found that executives who weigh more options actually make faster decisions.

It’s a counterintuitive finding, but Eisenhardt offers three explanations. First, comparing alternatives helps executives to understand the “landscape”: what’s possible and what’s not, what variables are involved. That understanding provides the confidence needed to make a quick decision.

Second, considering multiple alternatives seems to undercut politics. With more options, people get less invested in any one of them, freeing them up to change positions as they learn. As with the banner-ad study, multitracking seems to help keep egos under control.

Third, when leaders weigh multiple options, they’ve given themselves a built-in fallback plan. As an example, one company studied by Eisenhardt was pursuing negotiations with several partners simultaneously. When the negotiations with the first-choice partner failed, the president simply cut a deal with the second-choice partner. If, instead, the firm had pursued only one option initially, those negotiations might well have dragged on as the president fought to salvage the deal. (And he would have been tempted to concede too much to make it work.)

With some decisions, finding more options is easy—you can just expand your search. You can interview three job candidates rather than one, or if you’re shopping for a house, you can visit ten rather than five. After all, you can’t move into the dream home you never saw.

There’s no “right number” of houses to see or job candidates to interview. One rule of thumb is to keep searching for options until you fall in love at least twice. If you’ve only identified one good candidate for a job, for instance, you’ll have the strong urge to talk yourself into hiring her, which is a recipe for the confirmation bias. You’ll start to make excuses for the flaws you see: She asked us not to call her old boss for a reference, but that’s probably okay, because the boss sounded like a real jerk …

The same search-expanding logic also applies to choosing a car or a college or a job, though there are certainly commonsense limits—i.e., you probably don’t need to fall in love with two hair dryers before picking one, and God help you if you apply this advice to marriage.

So far in this chapter, we’ve emphasized the benefits of multitracking your options. We’ve implied that more is better. However, if you’ve ever walked into an ice cream store and found yourself stymied by the array of choices, you know there may be a limit to the amount of “more” we can take. This leads us to an important concern about multitracking. Psychologists such as Barry Schwartz have written about the dangers of “choice overload,” our tendency to freeze in the face of too many options. Is multitracking likely to plunge people into choice overload?

There is research suggesting that extreme multitracking is detrimental. A classic study by Columbia’s Sheena Iyengar and Mark Lepper monitored the behavior of consumers in a grocery store. One day, the store set up a sampling table with 6 different kinds of jam, and customers loved it; another day, the store set up a table with 24 different kinds of jam, and it was even more popular than the first. The surprise came at the cash register: Customers who’d chosen among 6 jams were 10 times more likely to actually buy a jar of jam than customers who’d chosen among 24! It was fun to sample 24 flavors, it seems, but painful to pick among them. The choice was paralyzing.

Most decisions, though, don’t involve choice sets that force us to choose among 24 options. Remember what we saw in the last chapter: When most people and organizations make decisions, they are more likely to be choosing among, er, one kind of jam. (I’m deciding whether or not to buy this strawberry jam.)

Adding one more alternative can expose a tradeoff or assumption that a single proposal hides. That is a more limited task than evaluating a large set of barely distinguishable options.

For evidence that adding another alternative can lead to superior decisions, consider a study of every major decision made at a medium-sized private German technology firm. The researchers, professors at the University of Kiel in Germany, had discovered that the firm kept extremely detailed notes of its meetings, including deliberations on decisions. (The notes were sent to the firm’s major investor to keep him abreast of what was happening.)

Over one particular 18-month period, the archives revealed that the executive board had debated and resolved 83 major decisions. Decisions never involved more than three alternatives, and 95% were either a “whether or not” decision (40%) or a two-alternative choice (55%). (Thus these decision makers were noticeably savvier about escaping narrow framing than the typical firm in Nutt’s study on this page.)

The university researchers discovered the archive many years after the decisions had been made, so it was possible, with the help of the executive board, to assess the quality of the decisions in light of their subsequent success or failure. In an intense rating procedure that involved hours of discussion and debate, the board categorized each of its 83 decisions as having proved to be very good, satisfactory, or poor.

When the researchers analyzed the data, the evidence was striking: When the executive board considered more than one alternative, the proportion later rated “very good” was more than six times as high. (Specifically, 40% of the multi-option decisions were rated “very good,” compared with only 6% of the “whether or not” decisions.) The association is substantial, but the board rated outcomes retrospectively. The study does not isolate the effect of adding alternatives from other differences among the decisions.

The practical suggestion is modest: develop another viable option before committing. You do not need to imitate the 24-jam display; two or three alternatives may reveal the comparison you need.