Hi Vittorio
I feel not being and by far the more qualified to give you an answer, having
never used a model with a word of mouth hypothesis.
I have the same kind of problem then you.
I am working currently on a price setting problem, and one of the hypotheses
included in the layered diagram representing how the hypotheses are working
together is the famous word of mouth.
The word of mouth is important in our service activity (short term rent a
truck, van and car).
In the layered diagram it is driving the behaviour of many other hypotheses.
The problem is that the difficulty to calculate it is as high as its effect.
I have seen many models, mainly in Stermans Business dynamic book and
others and the way the word of mouth is calculated does not satisfy me. In
fact it is in my model the part that is the more ambiguous.
I have started converting the diagram in a model, and worked on the first
hypotheses and left aside the word of mouth that I will integrate later on.
This is the method I will try to follow to resolve the word of mouth
problematic, providing that I may change it depending on my further
experiences.
First let us criticize the usually found in text books formula that Michael
Bean has sent in his mail.
Word of mouth sales = Installed Base Customer x Number of Contacts Per Month
x
Fraction of Potential Customers x Probability of Adopting if Contacted
The number of contacts per month does not specify if the contacts have the
same effect.
For instance you can have revolving contacts, and occasional contacts.
A revolving contact is by example the contact with your wife or children or
colleagues at work etc.
The first contact with a revolving contact will have more effects then the
second or the third.
There are a lot of revolving contacts.
You can always say that it is an average and you have the probability of
adopting if contacted that will make the difference.
Imagine by example that the base of customers is relatively stable and the
only interesting contacts are the non revolving, because the revolving have
already been contacted.
In that case, logically there is very little word of mouth, depending only
on the revolving customers. Then the word of mouth is depending too on the
customer base being instable.
Another problem comes from the needs of the person being contacted. It the
person being contacted has the need only one year afterward, he may have
forgotten the contact.
All this to say that the number of contacts per period has an influence, but
it needs some other information.
I think that it is easier to consider the product number of contacts x
probability of adopting if contacted that makes sense.
The difficulty is to calculate the parameters: number of contacts and
efficiency.
But the worse problem comes from the fact that the probability of adopting
is influenced by the world of mouth of the competitors.
Your word of mouth is influenced by your own word of mouth plus the word of
mouth of competitors.
The result is that the efficiency can vary considerably with the time.
It can be negative too.
One can argue that a model can only consider endogenous factors. But the
objective is generally to find policies and is it worth considering an
endogenous factor that can vary suddenly and change the results of the
policy?
Other problems can be considered: for instance the fact that some people
contacted can be occasional or regular buyers. It will make a difference for
the efficiency. Other factors can be considered depending on the kind of
product.
Whatever the problem, word of mouth is worth considering it.
I will follow several steps.
1. Calculate the impact of word of mouth on the goal of the model.
What is its influence? Is it worth investing money and time and how much, to
add word of mouth to the model?
For instance, in my model, I am interested in price setting. Maybe whatever
the word of mouth effect, it may have no influence on the price setting or
the influence may be small.
If I was interested in a more general model whose goal is to maximize the
result on a long range period, then word of mouth should be considered.
2. Once I have an idea of the influence of the word of mouth, I will have a
close look at reality.
Going down to earth, talking to revolving or not revolving customers or
anybody linked to the subject, is a paramount way to understand how things
are working, get a feel of reality and is often better than tons of data.
3. After this close inspection of live data, I will decide if it is worth
estimating by some ways the effect of word of mouth, independently from any
other factors.
I want simply to know if there is an influence and how much.
To do this, I can make qualitative in depth questioning or quantitative
questioning.
I can for instance, ask every new customer in my agencies, why he decided to
choose our Society or choose some customers and talk with them a longer
time.
4. If the third step is positive, I will have proved that there is word of
mouth influence.
But I still do not know how. I personally think that for my case, it is
depending on the basis of
customers and potential customers multiplied by a factor. I can estimate the
factor if I know the customers the potential customers and the effect. One
of the factors that influence the word of mouth is the quality of the
product, relatively to the competitors.
I will have then to establish a close survey of the quality of our product
compared to the competition.
I have then the formula:
Word of mouth per period =
Installed Base Customer x
Fraction of Potential Customers x Quality of our product compared to the
competition x
a constant factor.
The constant factor can be evaluated by dividing the word of mouth per
period estimated by survey by the other calculated variables.
This is of course a lot of work and it is necessary to prove its usefulness.
After all I may simply leave the word of mouth effect for the next model.
Commercial policy.
Regards.
J.J. Laublé, Allocar, rent a car company.
From: =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <
JEAN-JACQUES.LAUBLE@WANADOO.FR>
Strasbourg France