New gTLD Applicants Must
Reduce Cognitive Biases
Alex Tajirian
July 6, 2011
To improve the viability of your application for the
recently
approved launch of new generic top-level domains (gTLDs),
you must reduce
decision biases
in estimating a new gTLD’s expected revenues. To do so, you must either
delegate the responsibility to an independent multi-disciplinary team within
your company, or seek input from a consultant.
Cognitive researchers have identified two modes of
thinking: intuitive and reflective. The first is referred to as System One,
while the second is referred to as System Two. System One is a mode of thinking
that gives us a stable representation of the world around us, which allows us
to perform certain actions without much thought. An example of this would be
walking and contemplating something else at the same time. System Two, on the
other hand, is a mode of thinking that is reflective and analytical. Professors
Thaler and Sunstein popularized this distinction in their book “Nudge”.
Cognitive biases arise due to System One thinking.
The essay deals with reducing such biases by looking at the decision process
through a Systems Two lens. Experts need System Two thinking to complement
their intuitive thinking, as they are unable to fix errors that they do not
recognize. The essay does not deal with errors in strategic and marketing
analyses. (I have outlined such
biases
when appraising domain names.)
You are obviously better off with reduced bias than with
irrational exuberance when it comes to your profit forecast. After all, the
total cost of an application can easily top $400,000. You will not want to
throw away that kind of money.
You must work to assist the group in charge of
estimating profits to reduce cognitive biases. You and/or your consultant must
play the delicate role of not being seen as a quality control boss. You do not
want to appear to question a witness’s expertise or integrity. Instead, this complementary
role you and/or your consultant will undertake is intended to stimulate
discussion and debate.
Eliminating biases becomes an especially curtailed
process when you do not rely on
prediction
markets to estimate demand for your proposed gTLDs.
Nevertheless, even when using prediction markets you can reduce operating cost
estimation biases that are estimated more accurately using internal methods.
Below are some questions that your System Two team
needs to ask the team estimating your new gTLD profits:
-
Have you rejected the potential value of new gTLDs based on the performance
of existing TLDs? Such an outright rejection might be based on
comparing apples and oranges, i.e., using the wrong
analogy and cognitive biases by not focusing on
niche vs. scale gTLD markets.
-
Have you fallen in love with the idea of jumping on the new gTLDs bandwagon?
(This bias is called reference class forecasting. See, for example,
“Delusions of Successes: How Optimism Undermines Executives’ Decisions,” by
Dan Lovallo and Daniel Kahneman, Harvard Business Review July 2003.)
-
Are your analyses taking into account the cost and scope advantages of
incumbent registries?
-
As an incumbent registry, are you unconsciously motivated to “build an
empire” in the gTLD space? Or does your unbiased analysis suggest value
creation?
-
Were there any dissenting opinions within the profit estimating team? What
was the team’s motivation to converge on the presented recommendation? These
questions are intended to reduce the possibility of the group thinking
and/or converging on a position simply because it was gaining momentum.
-
Are the demand numbers of domain name registration under your proposed gTLD
very optimistic/pessimistic? Or are they based on prediction markets?
Pessimism can arise, for example, when you incorrectly rely on data from
current TLDs.
-
Is the worst-case scenario overly optimistic? Conversely, to assess whether the
recommending team is overly cautious, ask yourself if your company tolerates
unpredictable failures or if such failures result in unjustifiably firing the
deciding team?