By structuring complex decisions using models, decision makers create a tangible structure for establishing their objectives, and priorities so that they can understand the drivers to value their alternative choices. Without a model, most decisions are made over table conversations and there is virtually nothing to review to understand the assumptions and priorities of the decision makers. It is very hard to improve on decision-making as a process without a framework that explicitly states the decision objectives, priorities, and performance scores for alternative options.
Here are a few tips and best practices:
- Use Brainstorming pane to list your potential decision criteria. Do not worry about where criteria should be located in the decision model until you have listed most of them. Drag and drop these criteria into the Treeview and build levels into the criteria framework.
- Break down higher level criteria into lower level criteria by putting the lower level criteria in the tree at lower levels. Higher level criteria should often be stated more as objectives with a verb describing the action e.g. Improve Customer Satisfaction, Reduce Costs, etc.
- Limit the number of criteria on each level of the tree to 9 where possible to minimize confusion when setting priorities. It is difficult for people when dealing with more than 9 factors when setting priorities.
- Try to develop very explicit definitions for criteria and don't be afraid to change those definitions again when weighting the criteria to ensure that they make sense.
- Clustering criteria into groups makes the measurement process easier and more accurate. In selecting a car, you would put Performance and Safety at the top level of the model, Power and Speed under Performance, and Braking System and Airbag under Safety.
- Make sure that criteria are homogeneous on every level of the model. It is very difficult to compare very small criteria directly to very large criteria, so it is better to try to compare criteria that are relatively commensurate in size or performance.
- Before you evaluate alternatives, it is good to have explicit definitions of rating scales that can be used for rating the choices against weighted criteria (see Understanding Rating Scales)
Cautions about Building Decision Models
When building Decision Models, it is important to remember that you can lose data if you change the model in certain ways.
- If you delete a decision criterion and you have already weighted it and rated alternatives against it, you will lose all data associated with that criterion permanently.
- If you move a decision criterion to another level of the model and you had already done comparisons of that criterion at its previous level, you will permanently lose the previous comparisons.
- You can reorder criteria on a level without losing data, but you will always permanently lose comparisons and ratings if you move the criteria to another level or to another set of sub-criteria on the same level.
- If you delete a rating in the rating scale and you have already rated alternatives with that rating, all ratings will be lost that used that word or number. (for example, if you delete Excellent from the scale, then you will lose all Excellent ratings for alternatives against the weighted criteria)
- When you are doing resource allocation decisions to allocate budgets, you should always SAVE AS and change the model name when you are actually going to change the structure of the model by adding or changing criteria and/or alternatives.