Optical Distortion Inc. (ODI) markets lenses that distort the vision of chickens. These lenses are used instead of debeaking. Lensed chickens are more likely to survive. They also eat more efficiently than debeaked chickens. The key issue facing ODI is "How to market these lenses?".

Markov Model.

ODI Managers believe direct sales to chicken farms are the best answer. But they need to determine which farms to call on and how large a salesforce to utilize. To figure this out, they have called in marketing-model experts. These experts have listened to their description of the problem and have suggested using a "Markov Chain" model. ODI has agreed and has hired them (disguised as Professors Robert Eng, Ashok Rao and Larry Isaacson) to build such a model.

A prototype has been completed, and your task is to use it -- by specifying the key marketing values in the model.  You can select what regions and what sizes of farms to sell to, how large a salesforce to hire, how much to spend on overheads and marketing activities, how to allocate sales calls, and what the probabilities of trial and adoption will be under your strategy and action plan.  A copy of each of several variations on the model is available for you to use or download from e-campus.  The three versions -- (Download"r")  (Download"ra") (Download"rr") -- are explained below.

Basis of the Markov Model.

This Markov model is based upon the concept of the diffusion of information – used by many marketers to describe how innovations become known and adopted in a marketplace. Markov models capture customers’ step-by-step decision-making processes and simulate them to estimate how an innovation diffuses through a market on a period-by-period basis. This often turns out to be a good approximation of how markets actually work. This approach can be much more helpful and accurate than simply projecting future sales in total, without identifying the underlying consumer decision processes. It often produces better forecasts of markets and/or market shares.

Model Variables.

The ODI model -- written as an .xls spreadsheet -- assumes that when a farmer is contacted by a sales rep, the farmer becomes more likely to "try or buy". Technically, contact increases the probability that a potential customer will become interested in a product, agree to try it, and eventually become a customer.

ODI managers see the market as consisting of three segments -- large, medium and small farms. When farmers decide to buy the lens, they generally begin by buying just enough for a test. Then, after one or more test periods, they may buy enough lenses for all of their chickens. The Markov Model permits you to replicate this "diffusion process" by assigning sales calls to different size farms, and to farms that are previously uncantacted, in trials, or have already adopted. You can set the number of sales reps and the number of sales calls each rep will make. You can then allocate these calls to different target segments.

The Markov model requires you to set the probabilities of success of sales calls. The model already contains some "guesstimates" as to how farmers will respond to sales calls.  But you can change these probabilities to reflect your best guesses about the impact of the sales program you elect to run. The model will then apply these probabilities over several time periods. When you use the model, you will enter your estimates of these probabilities in the "transition matrices" provided in the spreadsheet, just type over the numbers that are there, based on your assumptions, your pricing strategy and the other elements of the selling strategy you choose

Please note that this is not a "smart model". That is, it does not model strategic relationships and interactions. Rather, it accepts your estimates and uses them to produce multi-period results. So be sure to choose probabilities that reasonably reflect your policy choices. For example, if you charge higher prices, trial and adoption may be slowed down. You should reflect this in your choice of trial and adoption probabilities.

More About The Transition Matrices

The model assumes 4 possible "states" for each farm -- uncantacted, trial, adopt and reject. All farmers start off as "uncontacted".  As sales calls are made on them, they may move to "higher" states -- depending on the probabilities you assign to such outcomes.  Exhibit 1 shows transition matrices for small farms, medium farms, and large farms that have been visited by a salesperson and for farms that have not been visited. The probabilities in these sample matrices indicate that:

• If a farm has not been contacted, and is not contacted in this time period, the farm stays in the uncantacted state and has little or no likelihood of trying the lens.
• However, if the farm has received a prior visit, the chance of trying the lens is +/-30 percent and the chance of refusing to try it is +/-70 percent.
• Notice that a farmer who is in the trial or the adopt state is less likely to reject the idea (that is enter the reject state) if the salesperson has paid a recent visit.
• Also, the farmer is only likely to move from the trial to the adopt state if the salesperson has paid a visit.
• Once, a farmer enters the reject state, these matrices imply the farmer is less likely to leave that state and try the product. (Models "r" and "ra" are based on Zero probability of leaving the reject state.  Model "rr" is set up to assume some likelihood of a move out of the reject category.)

Sales Force Allocation

ODI management believes that each salesperson can cover 80 farms per period, according to the case. Based on the number of salespeople you have chosen to hire, you will therefore be able to cover a multiple of 80 farms per salesperson per period. In the model you will then have to allocate these visits to specific groups of farms. If you believe that salespeople may do a better job if they cover fewer farms, you can reduce this number.  This will increase the cost of covering farms, as you will need more salespeople, but it may pay-off as you will increase the probabilities of trial and adoption to reflect their increased effort.

Sales Forecast.

The sales forecast for each period is based on 5 elements:

• The number of farms in each market segment in the regions you elect to sell into.
• Their state - uncantacted, trial, adopt or reject - at the beginning of each period.
• The number of farms in each of the above categories that you have assigned the salesforce to visit.
• The number of farms in each of the above categories that will not be visited that period.
• The transition matrix you defined earlier.

Marketing Variables.

In this model, you can control expenditures for Headquarters Expenses, Regional Expenses, Advertising and Trade Shows.  Just enter these variables by typing over the data in the various "blue" cells for these data.

Regions.

The first "page" of the model shows how many farms -- and how large they are -- by region and for the US as a whole.  It also shows "updated" data for the entire country.

For each period you will first decide which regions -- or the whole US -- to cover.  You will enter data on the "first page" of the model to define the farms you plan to call on.  To do so, just "copy" the data for the first "blue" cell from the region you plan to serve -- or the whole US -- and then copy across to the next 5 cells.  If you wish to serve 2 or more regions, just enter the "sum" of these regions.  If you want to work with "updated" data, copy it from below the "blue" cells.

Salesforce Allocations.

You will then decide -- on "page 3" of the Model, how to allocate sales calls, and the model will compute your results.  Before you make these choices, you should decide on a priority of calls. For example you may decide that:

• Large farms that have not yet been visited should have the highest priority.
• Large farms that are in the trial state are your next priority level, and so on.

The basic model -- the "r" version (Download"r")  -- requires you to set the priorities for how you will assign sales calls to farms.  You will have to go through all 8 periods and make these assignments "by hand".  It takes quite a bit of time -- but please do it a couple of times as it will help you better understand the Markov Model.  Please note that the "green" counter helps tell you when you have assigned all the calls you have available.  If you need more sales calls, add a salesperson and do the assignments again -- until you use up all the available sales calls.

To save a lot of time and effort, models versions "ra"  (Download"ra")  and "rr" (Download"rr")  have set certain sales call priorities.  You may agree with these priorities or not.  But they are "embedded" in these models and do all the sales call assignments for you -- as soon as you set the number of salespeople.   The "ra" version assumes that once a farmer has rejected the lenses s/he will never elect to try or adopt them.  Thus the Markov Matrix has "zero" probabilities on these events and the sales call matrix priorities do not assign any sales calls to rejecters.

By contrast, the "rr" version assumes that even though a farmer has rejected the lenses s/he may reconsider and become a "tryer" and then move on to become an adopter.  So the transition matrix has positive probabilities for these events -- which you can modify -- and the sales call matrix priorities include assigning sales calls to the larger rejecters after all the other farms have been covered.  Obviously, it makes sense to select the version that best conforms to your view of the market and to then alter the probabilities to reflect your marketing strategy.

Marketing and Financial Reports

Once the data has been entered, the Model produces a marketing profit and loss report.  This report shows the share of the potential market achieved over the 8 quarters. It also reports the number of farms that have been visited in each quarter.

This report is based on the number of salespersons you elected to hire at the beginning of each run. It cannot be modified from period to period. The number of technical reps is computed assuming one per 5 salespeople.

The report also uses your inputs to determine the prices at which you will sell and the costs you will incur.  You will enter these costs on the "second page" of the Markov Model.  It projects an income statement for each of the next 8 time periods.

As you work with the model, you may choose to change the selling price, or some of the costs. As you do, you should keep track of what you have changed and consider the impact. For example, if you decide to raise the price of the lens the farmer is less likely to purchase it. So, the probabilities in the transition matrices should be adjusted to reflect the fact that the economics for farmers will change as they take into account the higher lens price relative to the cost of debeaking.

Operating the Spreadsheet Model

The model -- as it is set up -- will provide results – but they may not be meaningful if you have not chosen your inputs wisely. To use the model well you will need to iterate through the following steps:

• Check all your assumptions and costs. (The figures you are permitted to change are highlighted in blue. Figures you should not change are shown in black or green.)

• You are now ready to use the model to produce meaningful results.