Using Weights and Scores Prioritising TechniqueScott Williams
When you are faced with many alternatives, it can often be very hard to choose between them. They will be strong on some things and weak on others; but hardly ever the same things. The sort of problem you face when choosing a car or a new staff member for example. A quick and effective way to work through such a maze is to use a Weights and Scores technique described below. Yellow Belt
First list all the criteria that you think are relevant and mark them ‘high’, ‘medium’ and ‘low’ in importance.
Let’s say we are choosing a car.
We might identify the following criteria:
- Resale value (high).
- Affordability (high).
- Environmentally responsible (low).
- Bluetooth connectivity (low).
- Safety (high).
It is also helpful to identify any other criteria that are essential – the ‘must haves’. We might say seating for 4 is required.
The process is as follows:
- Draw up a table (a spread sheet is ideal for this as you will see) like the one below. Put the alternatives across the top row (Car A, Car B etc.).
- List the selection criteria down the page ( 4 seats, resale value etc.)
- Convert the criteria importance to a number between 1 and 10 with 10 being the highest importance. We call these the “weight” (for importance) given to that criteria.
- Through experience I have learnt it best to be pretty tough on these weights. If they are all in a narrow range, say 7-10, they will not be much use later. I usually give highs a 9, mediums a 6 and lows a 3 value.
- It is helpful when doing this to choose the one in the column of criteria that you think most important and give it the highest weight (a 10). Then consider the next most important criteria in the column and weight it in comparison (maybe a seven against a ten), and so on. Usually you can separate out the options fairly easily this way.
- Write the weights into the second column.
- Now cross out any of the cars that fail the 4 seat ‘must have’ test as it is not worth looking at them any further. In this case, we have failed Car B.
- For each of the remaining criteria one at a time, see which of the choices best meets this criteria and give it a “score” of 10. Then score each of the remaining choices in comparison to the highest ranking. Once again, be pretty tough here and get a good spread of scores between choices or the results will be a bit muddy. Often many will end up having a similar score because they are not very different. Most cars today are fairly safe, for example, so it might be a bit had to score them too far apart. If this happens with several of your selection criteria, you might need to find even more selection criteria that the choices will score differently on so as to force them apart in the selection process.
- Now multiply each weight given to a choice by the score for the row. In the example table below, the score is in ( ). For example, Car A is a weigh to 6 times a score of 10 = 60 on the “resale” criterion.
- Add up the (weight X score) result for each choice column and write them in the total row. For example, Car A has a total of 241 whereas Car C has a total of 157. Work out the perfect score by multiplying each number in the Weight column by ten and adding them together. Because the best choice in each row gets a 10, a perfect score would be all tens. In the example, the perfect score is 310.
- Calculate the percentage each choice gets of a perfect result. The percentage is Car A (241) / Perfect Score (310) = 78%
0 (not there)
From the original seething mass of uncertainty, we can now see the best scoring option is Car D. Given your assessment of your preferences and how well each car scores against them, this is a logical outcome.
Doing the percentages at the bottom row helps to show how different they are. But it has another purpose as well. If all the options scored fairly low percentages (I would suggest under 60%), it might well mean that none of them really satisfy your needs and you should cast your net wider to get more choices. Also, if two percentages are close together, the selection is not clear so you should widen your range of choice and/or of selection criteria. As this will often happen if the scores for each criteria/choice option are very similar, you could go back and re-score tougher. This is the problem I alluded to above when the weights and scores are too close to each other.
This process can be improved sometimes by having several people contribute to the scores and weights. This can become problematic though if you have people with widely varying opinions about the right value and you average them out to get consensus. The result is that every choice will average out and all will be very similar. It is healthy to get various opinions but make sure you end up with a choice that is not a compromise!
Although the system of scoring mentioned above will generally be sufficient, there are a couple of other approaches you might use sometimes.
Instead of scoring from best (10) down, you could start with worst (1) and score up. Once again, use fairly large jumps in scores (e.g. 1, 3.6.9) otherwise all the results will ‘average’ out and you won’t get a clear choice. Note also that the percentage of the maximum possible (perfect) score is likely to be lower because you are starting with lower scores so the 60%+ test mentioned above might not apply.
A second alternative approach might be of use when some options are radically better than others and/or if you keep getting all the results about the same.
Instead of using 1-10, you can use a so called Fibonacci series. This works be setting scores like;
- and so on
Each new number is added to the last one. This forces the scores further and further apart and lets you really put emphasis on outstanding options.