USING THE WISDOM OF CROWDS IN BUSINESS

In a previous post, I described NASA’s standard decision-making process, which is generically the process that all of us use to make decisions of significant importance. This process is as follows:

1. Assess the situation
2. Gather facts and assess unknowns
3. Identify alternatives
4. Establish decision criteria
5. Weigh alternatives
6. Select ‘best’ alternative
7. Validate decision

So when we are buying a house, for example, we consider why we are making such a decision, collect data about the current market, identify some houses we like, identify the criteria (probably subconsciously) that affect our choice, use those criteria to weigh the alternatives, make a choice, and then probably second-guess ourselves: “Now, are you sure you like this one better than that one?”

In a business, the process is similar. If we need to decide on an action plan to deal with a revenue shortfall, for example, we’ll look at the revenue data versus budget, collect demographic and economic information and competitive intelligence that may account for the shortfall, come up with a series of alternative actions that could be taken, identify the criteria (again, probably subconsciously) that affect the choice, use those criteria to weigh the alternatives, make a choice, and then probably validate it by running it by other executives or advisors we trust: “Here’s our draft Sales Catch-Up Plan, take a look at it and tell me what you think.”

Surowiecki would argue, I think, that there are opportunities at several points in this process to gather the collective wisdom of large numbers of people (notably employees and current and prospective customers), and that doing so would almost inevitably produce a much better decision or solution. In our discussion last week, we identified four places in the NASA process where the Wisdom of Crowds might be tapped. In so doing, we split the first step in the process (Assess the situation) into two parts:

• Articulate the components of the issue (for example, our revenue shortfall may be localized in certain product lines or geographic areas, and different solutions may be needed in each different area)
• Identify the root causes behind each component, to ensure the decision affects the cause, not just the symptom, of the issue at hand (for example, the revenue shortfall in area X may be due to a declining local economy, a new competitor, big turnover in sales personnel, or setting an unrealistic budget in the first place)

We also concluded that the second step, Gathering facts and assessing unknowns, occurs throughout the process and not just at one particular stage. The revised decision-making process is shown in the figure above, and the four places in the process where collective intelligence of the organization and/or its customers might be used are shown in the green boxes. Continuing the sales shortfall example, we might use this collective intelligence to:

• Ensure we had accurately identified all the important product line, regional and other components of the shortfall
• Rank the components by their impact on the shortfall
• Ensure we had identified the real underlying causes of each component of the shortfall
• Qualify and rank the alternative solutions we had identified to address each underlying cause
• Critique the proposed implementation of the selected solution alternatives

Here’s another example, from my personal experience: We wanted to know why the use of our corporate intranet was declining. We began by identifying the evidence that led us to believe this was a problem (e.g. some of the newest tools had fewer users than some of the older, less robust tools). The first stage of an extensive user survey surprised us: We discovered that the data we were getting on usage was misleading, and while there was definitely a problem, we had misdiagnosed the components of the problem. We learned, for example, that one of the reasons for declining use was that users weren’t able to find what they were looking for, and speculated on the root causes of this problem component: The search tool was too complicated, or there was just too much stuff to wade through, or perhaps the users weren’t adequately trained, or perhaps they weren’t aware it existed, or possibly they couldn’t find what they were seeking because it wasn’t there at all. What they told us, in the next part of the survey, was that looking for information on the intranet, beyond very simple lookups like firm policies or government regulations, was simply not as efficient a way to get pertinent information as walking down the hall or picking up the phone and getting that information first-hand from a colleague, with more context than the intranet could provide. They told us to ask their assistants about the usability of the search tool, since the assistants were the people who actually used it, for relatively simple lookups when they were instructed to do so (this assertion was in fact consistent with the usage data we had collected). We decided as a result to redefine the problem from “People aren’t using the intranet” to “People aren’t effectively aided by our technology and information to do their jobs efficiently.” We then changed gears completely and refocused on what we could do to help the informal peer communities identify and connect with each other more easily, and share their ‘filing cabinets’. We proposed several ways we could do this, including introducing weblogs and creating expertise finders. When we surveyed the users again, they picked the simplest, least powerful alternative, and, when we showed them a proposal to implement that, they suggested ways to make it even simpler. Had we not repeatedly surveyed our users to garner the Wisdom of Crowds, we would have gone on doing refinements to the intranet search engine, training programs, awareness activities, and content rationalization, none of which would have had any effect on use, or on the value people were (not) getting from these costly resources.

There are two ways of gathering collective wisdom of a large number of employees or customers: (a) Asking mostly closed-ended questions in written or e-mail surveys, or (b) Interviewing these people one-on-one, to probe and qualify their answers. If the list of possible alternatives is small and known with near-certainty, surveys are sufficient, and cheaper. But if you’re not sure you have identified all the important issue components, root causes, or alternative solutions, a closed-ended survey will give you misinformation, and probably convince your employees and customers (when they note that you’ve missed the most important components, causes or solutions) that you’re out of touch with what’s really happening in the organization. In these situations, a more expensive personal interview approach is needed. Surowiecki warns that interviewers need to be carefully trained not to put their personal spin on the answers they receive during this more open-ended intelligence gathering. They might want to use Dave Snowden’s ‘cultural anthropologist‘ approach to ensure the results of these interviews are objective, accurate and complete.

What’s interesting to me about this process is that the decision-making in the chart above is all made by the ‘crowd’. The ‘Solution Team’ are really just facilitators — they gather information, do research and analysis, and brainstorm solutions and implement the decisions. But none of these activities requires executive or management skills, or ‘expertise’. Anyone in the company with decent creative, communication and analytical skills can do these things. Theoretically, a company with this process in place for key decisions could operate without management or outside ‘experts’ at all. And, theoretically, if Surowiecki is right, the calibre of the resultant decisions would be greatly improved. Think how much money could be saved on executive and expert salaries! And how much more collegial the organization would be without decision-making hierarchy. It’s nice to dream about, anyway.

This approach can even be used to enhance personal decision-making. In the example of deciding which house to buy, the house-hunters would be wise to talk with (a) a lot of peers looking for the same qualities in a home and neighbourhood that they are, (b) many of the residents in each new neighbourhood they’re considering, and (c) people they trust to assess whether the problem they’re trying to solve will really be solved by buying a new house at all. We are all prone to jump to conclusions about the real problems, their true root causes, and the best solutions. The Wisdom of Crowds (even small crowds) can help us ‘know better’.

Some other points that came up in our discussion last week that are worth pondering before trying to tap into the Wisdom of Crowds:

• We all want reassurance that what we believe is right, and we all relish simplicity. That tends to make us hear what we want to hear, over-emphasizing opinions that reinforce our own views, and discounting (or even not hearing) conflicting or complex views. If we fail to recognize these biases in ourselves, we won’t be open to others’ collective wisdom.
• Executives, who are rewarded for making quick decisions (sometimes even if they’re wrong) tend to operate on a ‘hypothesis’ basis: When the need for a decision arises, they will draw on their experience and formulate a hypothesis. At this point they become deaf to any alternatives. If you want to suggest an alternative, first you need to convince the executive that his hypothesis is wrong. Only then will he (or she, though in my experience women executives are more open-minded) entertain an alternative hypothesis. Even then, it’s advantageous to set up the executive so he thinks your hypothesis is his idea.
• This all has some far-reaching implications for Knowledge Management (as I think my intranet example above indicates). If business is in fact a complex, adaptive system, then it is impossible to know all of the relevant information before making a decision. In such systems the best possible answer tends to emerge, given sufficient time and collaboration, as when a flock of geese migrate, through collective wisdom, in perfect formation to exactly the location of their nesting area thousands of miles away. Why don’t optimal answers emerge more often in business? Is this because business is not so much complex as just complicated, and unduly so? Or is it just that civilized humans, who see selfish behaviour exhibited everywhere (and often rewarded), have lost the intuitive skills of collaboration? Can any static ‘knowledgebase’ be designed to facilitate, rather than impede, emergent solutions? What would such a knowledgebase look like?
• In today’s business world, where, as Drucker says, everyone knows more about their own specialized job than anyone else, including their boss, we are constantly making decisions, making them alone, and looking (often fruitlessly) for useful information that can make the decision-making process less precarious. Obviously we can’t employ the full process illustrated above every time we make a simple decision, but it’s worth thinking about whether finding means to help workers make decisions faster and more intelligently shouldn’t be Management Job One, and if so what those means might be.