How to consolidate multiple KPIs
Can you combine different KPIs in a single measure? Would such measure make sense? I have seen this question so many times I decided to write a small post about it.
Consolidating KPIs seems at first glance contradictory. Weren't they chosen to be independent? Aren't they expected to be orthogonal, reflecting different part of business? How can you mix failures per hour with cost per click?
Strictly speaking, the average (or weighted average) of different KPIs do not make any sense. You can't mix different units the same way you can't mix a patient temperature with his blood pressure. On the other side obtaining an overall figure makes sense, like having an overall assessment of patient risk. A single number can assess the probability of patient survival or business success by combining multiple measurements.
So, how can you combine several KPIs?
First of all you shall associate a score to each KPI depending on its value ranges. You can define, for example, that conversion rate under 0.5% is terrible (score 0), poor between 0.5% an 2% (score 1), acceptable between (2 and 5%) score 2 and excellent over 5% (score 3). Now your KPIs map to score and score are homogeneous, so. you can combine them!

A convenient alternative to ranges is fuzzy logic. It is both more precise and will produce better results, but you need some math skill to implement it. It is not difficult but can't be described in this short post. Using fuzzy logic you assign an acceptable and unacceptable values to each KPI. Fuzzy math will take care in combining such values:

Finally you shall assign different weights to KPIs. The simpler way is to guess a number, say from 1 to 10 for each KPI. Frankly I prefer an exponential scale: 1 = little importance, 2 = under average, 4 = above average, 8 = extremely important. Each step doubles the score and, yes, there is no in-between to avoid human indecision!

Once you got scores and weights... Let Excel do the rest!
Last modified on 2011-05-23 by Administrator
