“. . . There are known knowns; there are things we know we know. We also know there are known unknowns; that is to say, we know there are some things we do not know. But there are also unknown unknowns—the ones we don’t know we don’t know.”—Donald Rumsfeld
Rummy sure has a way with words, concealing some powerful insight within bureaucratic gobbledygook. For most of us, uncertainty appears to be one large, amorphous mass, and Rummy has tackled that problem with a distillation, albeit one that’s a tad verbose. We should applaud him for even taking this on.
Let’s put Rummy’s idea to work. Suppose your company has decided to sell an established product into a new market. You have knowledge about the past and assumptions about the future. You understand that there are many possible outcomes, some of which are likelier than others. You know that one outcome will prevail, and even though you are fixated on your goal, you don’t know exactly how things will turn out. Question: how do you ensure the outcome you get is the outcome you envisioned? (Hint: the answer is probably not stay the course. The people who coined the term agile will get upset.)
This describes a classic uncertainty problem, and one that is especially common in revenue creation. How do vendors sort through the universe of data, artifacts, anecdotes, and information to develop sufficient knowledge to place bets intelligently? Rummy’s taxonomy can help.
Three distinguishing characteristics of an intelligent bet are 1) the odds of winning are understood, 2) the bettor can sustain a failed outcome, and 3) the best possible result should be one worth having. As I’ve learned, smart people can make dumb bets, and the converse is also true – it doesn’t require extraordinary brainpower to make wagers that are remarkably astute. Something to consider before forking over a hefty chunk of venture capital to a high-IQ adult. Want an extreme frinstance? Click here to see eight defunct dot.com’s that purchased expensive ads during SuperBowl XXXIV. “Oops. The money was nice while it lasted.” Dealing successfully with uncertainty involves having at least a shred of common sense.
Imagine that Rummy has a seat at the table as part of your strategic team. Here’s how he might whiteboard your planned market entrance:
The Known-Known’s. Pretty straightforward, but known-known’s are a small fraction of needed information: names of target organizations and their executives. Regulatory restrictions and pending legislation. Major competitors. Revenue and other financial information for each prospect. Specific Key Performance Indicators. Industry trends.
The Known-Unknown’s. Typical stuff that marketers and salespeople ask about: Size of the market. Trends. Forces. Competitive strengths and weaknesses. Average length of the selling cycle. Pain points. Influencers, movers, and shakers. Level of buyer knowledge and understanding. Decision criteria. Buying processes. Internal politics. Competing projects. Motivation. Money and budgets. Biases. Perceived opportunities. Perceived risks. The list stretches from here to forever.
The Unknown-unknown’s. Everything else. Things that nobody ever thought to ask about or discover. Events that happened before, but went under the radar. Events that never happened before, but might happen. Customer backlash over who-knows-what that might have a measurable impact on revenue. Mistakes that will be made that no one even knew could be made. The metaphorical blindside tackle. What author Nassim Nicholas Taleb calls Black Swans.
Rummy’s taxonomy guides a useful, and much needed conversation about revenue uncertainty. In the last twenty years, we’ve made great strides in adding to the corpus of known-known’s, and we’ve come a long, long way in learning how to discover the known-unknown’s. But we’re still left dangling, because we know that categorization only takes us so far. We still must answer, “now what?” And for that, we need mathematical rigor. Eighty years before Rummy, economist Frank Knight, author of Risk, Uncertainty, and Profit, examined uncertainty under that lens, outlining three types: a priori probability, statistical probability, and estimates. I’ll stick to the high level, so hang in there with me.
a Priori probability. You have a box with 12 blocks, and you know up front that six are green and six are red. Assuming you cannot see into the box, what is the probability of drawing a red block? The probability distribution has been determined by definition. This is an iconic example in which an individual can place a bet based on straightforward calculation.
Statistical probability. Imagine the same box, but now you don’t know how many blocks are in it, or how many different colors there are. This uncertainty problem is more complicated, and therefore more difficult to cope with. The probability distribution of the result is can be described by statistical analysis of empirically-collected data. Therefore, the way to manage the uncertainty in this scenario is to keep drawing and keep recording the result until you have sufficient information about the outcomes on which to base a future projection.
Estimates. Again, imagine the same box, but this time, you have no knowledge whatsoever about its contents. It could be holding anything. Any data that you might choose to collect don’t lend themselves to any statistical analysis.
Knight was keenly aware of the dangers of conflating “the problem of intuitive estimation” with “the logic of probability,” whether a priori or statistical.
Here’s what he wrote: The liability of opinion or estimate to error must be radically distinguished from probability or chance of either type, for there is no possibility of forming in any way groups of instances of sufficient homogeneity to make possible a quantitative determination of true probability. Business decisions, for example, deal with situations which are far too unique, generally speaking, for any sort of statistical tabulation to have any value for guidance. The conception of an objectively measurable probability or chance is simply inapplicable . . .
When I read his opinion, my immediate response was, “Wow. He must be turning over in his grave today.” I’d love to see his reaction watching sales executives discuss revenue forecasts, or listening to data wonks crow about the ‘predictive validity’ in their models for B2B decision-making. And I don’t see Knight endorsing any company’s policy for assigning increased purchase probability based on where a deal sits on a hypothetical sales process continuum. Yet, many companies abdicate probability to the “forecasting logic” embedded within their CRM applications, while their senior executives scratch their heads wondering why Sales can’t furnish a more accurate number. “If only our sales reps would populate the information we’re asking them for!” Hmmmm. Which unknown-unknown’s might you be referring to?
I’m not advocating that forecasts have no value, or that companies should discontinue preparing them. Only that we’re squandering opportunities to gain insight about what makes revenue uncertain, and we’re failing to use the insights that we do gain to reduce the volatility in revenue results.
We all want less uncertainty. I get that. But we expect people responsible for revenue generation to be prescient beyond their capacity – heck, beyond anyone’s capacity – and then kicking them in the rear when they are wrong. Happily, there’s a way out of this frustrating cycle. In Part II, Putting Uncertainty to Work at Your Company, I’ll cover how to create a repeatable process for identifying and assessing revenue uncertainties, and in Part III, How to Model Revenue Risk, I’ll show how probability distributions can be applied to specific uncertainties, and how to interpret and use the results.