Here are six guidelines to remove most of the subjectivity from any KPI or performance measure.

Subjectivity is one of the greatest enemies of performance improvement. It is a source of bias that reduces both accuracy and precision from the information we’re basing our decisions on. If information is subjective, it means that it is, at least in part, shaped by personal opinions, beliefs, feelings, attitudes, and perspectives.

Subjective information has a few identifying features.

Information’s subjectivity increases the more it possesses features like these:

• It’s qualitative, rather than quantitative.
• It’s captured serendipitously, without a deliberate and valid collection methodology.
• It’s held and communicated by a single person.
• It’s based on tiny samples.
• It’s shared in the form of an anecdote.
• It’s original source is unknown or unclear.
• It lacks the context of well-designed measure or construct.

For example, elected Councillors might lean too heavily on their recent conversations with constituents or community members who are more vocal. And then pressure their council’s leadership team to allocate resources based on that subjective information, rather than on more objective evidence from community research.

Subjective information has some time-wasting consequences.

Particularly in performance improvement and strategy execution conversations, subjective information causes some problems:

• It reduces group decision-making to a bunfight, where those with the most power win.
• It’s not representative of the populations it gets used to make decisions about, so leads to decisions that don’t have the intended impact.
• It provides no reliable baseline against which any impact can be compared to figure out if action worked or not.
• It isn’t defensible under scrutiny (like an audit).

Minimise subjectivity in your KPIs by following 7 guidelines.

To avoid introducing subjectivity into your KPIs or performance measures, we need to make sure their data is based on real-world evidence and not personal biases. Scientists do this all the time, by following guidelines like these:

1. Be very specific and unambiguous about the result you want to measure. Don’t leave it vague and broad.
2. Clearly define the population of people or things to which this result, and your measure, relates.
3. Define the pieces of data required to produce the measure, detailing their expected values and formats. For KPIs, most will be quantitative.
4. Design an instrument (like a form or survey) that asks for the required data in their required formats, without ambiguity or bias.
5. Design a process that can collect these pieces of data, which protects the data from biases of the data collectors.
6. Select a representative and random sample from the population of interest (or all if the population size is
   manageable), from which to collect the data.
7. Produce and use the information in teams, to reduce the impact of confirmation bias.

Of course, no data are ever perfect, and we can’t confuse them for facts. But just because data isn’t able to be perfect, doesn’t mean we can’t make it as objective as possible, to get closer to the truth. Numbers that are derived scientifically are the basic ingredient of any measure or KPI that provides objective information.

Numbers that are derived scientifically are the basic ingredient of any measure or KPI that provides objective information.
[tweet this]