Calibrating Unknown Intangibles

["Rod Brown"]

Timothy Fong writes:

|Question: has anyone done this sort of thing [calibrating "soft"
variables in the absence of performance history] .. for real and
|developed a case-study where the "guestimates" are used to
|build a compelling model? And what was the reaction of the
|client?

Timothy,

My guesstimate is that this has been done many, many times; my
colleagues and I have certainly trained management teams to address
exactly this sort of launch issue. It is common in any genuinely new
venture, not just marketing of a new product.

Modelling can certainly help but at least two critical points should be
remembered:

1) It all comes down to managerial judgement and nobodys judgement
is perfect. In fact, anticipating the time path of - in this case -
future sales in the absence of history is inherently tricky and wrong
more often than not.

2) Forecasting accurate sales growth is not the point. The point is
understanding the system whereby sales are created, making a plan with
clear assumptions as to how a particular path of sales is to be achieved
and then monitoring all the critical rates and accumulations to see
whether the plan is coming to fruition as intended ..or not.

Understanding the business engine of growth, as far as management
judgement allows, means that you can and ought to set policies for
future action: if we are 50% down on the assumed rate of customers
becoming aware, we will double awareness advertising and halve trial
expenditure ..or such like "rules". Commonly, once a model is built
and the forecasts made or "learning" extracted it is not referred to
gain - job done. But keeping some living means of monitoring success is
very important if launch or any other strategy is to adapt intelligently
under management control. (I dont mean necessarily a slavish
recalibration of the model each month.)

More practically: there are very few cases where a product launch is
entirely in the dark. Analogues can usually be taken from the same or
similar industry at some point in the past. Some interpretation will
necessary, of course.

You also have to be careful about the meaning of soft variables. In the
case of awareness, it depends how you want to handle it. My preference
is not to treat it softly: people are either aware or they are not
aware; people is the unit and thats a hard number or number/time.
Advertising agencies, good ones anyway, are pretty smart at estimating
numbers of people made aware per buck spent in a certain way - a shame
they wont extend that to the sales figure but then I guess thats where
the management team - ably assisted by us dynamicists - earn their pay!

Kind regards,

Rod.

Roderick Brown
Centre for Strategy Dynamics
rod@strategydynamics.com

[Timothy Fong]

I met with a company that is rolling out a new product and the
problem they want to address is how and how much marketing
dollars to spend to achieve certain sales targets.

Obviously, there are lots of "soft" variables like awareness
and word-of-mouth, but they dont have any hard data or
historical information, not even at an aggregate level.

However, I believe there is still significant decision-making
value in creating the model.

Question: has anyone done this sort of thing for real and
developed a case-study where the "guestimates" are used to
build a compelling model? And what was the reaction of the
client?

Thanks.
---------------------------------------------------------
Timothy Fong
timfong@ureach.com
---------------------------------------------------------

["Jim Hines"]

Timothy Fong asks an interesting question about modeling when it you
dont have enough information to answer a clients specific question --
e.g. how much marketing dollars to spend.

The first question -- how much to spend -- probably cant be answered as
precisely as the client wants, because you dont know the precisely
enough the strength of the relationship between marketing spending and
buying nor do you know the time lags.

OTOH, Timothy can probably shed considerable light on the question of
**how** to spend marketing dollars when the company is relatively
ignorant of certain delay-times and relationship-strengths. The policy
would determine how much to spend each week (or month), and would use
information from buying decisions (etc) as the information becomes
available.

My guess is that the information Timothy has **now** (e.g. how long it
takes to build capacity, the channels through which customers learn
about the product, where significant delays are likely to be...)is
probably adequate for him to help the managers create a spending policy
thats better than the policy theyd use without his help.

Jim Hines
jhines@mit.edu