How to Design Good KPIs — That Actually Measure Performance

Most leaders and strategy and performance professionals say they want to define KPIs to measure and track performance. But in reality, many KPIs measure and track activity, busyness, or compliance — not whether the important outcomes are actually improving. That’s why so many dashboards look impressive but fail to answer the most important question:

— Are we getting better at what matters most?

If we want to define KPIs that really will support decision-making, learning, and improvement, we need to approach them very differently from the usual “brainstorm a list of metrics” exercise — we need to deliberately design KPIs.

This KPI Fundamentals Guide outlines the essential steps to design good KPIs, the most common mistakes to avoid along the way, and tips for best practices.

To wrap up, you’ll discover the basic test of a good KPI, and our top 3 recommended programs to get the professional skills you need to set KPIs that actually measure performance

Step 1: Understand what good KPIs really are (and aren’t).

Before we choose any KPIs, we need to get clear about what they actually represent. This is a question most people don’t realise they should ask themselves, and it allows a lot of misconception to sabotage their KPIs before they even get started.

Let’s explore what a KPI is and isn’t, with examples for a strategic theme of digital transformation…

A KPI is not:

• a target e.g. Increase digital adoption by 25%
• a project e.g. Migrate to cloud infrastructure
• a project milestone e.g. All users migrated to new digital tools by 2028
• an activity e.g. Number of digital training sessions delivered
• a collection of data e.g. Employee digital satisfaction surveys
• or a vague and undefined label e.g. Digital Adoption Rate

A clue is that if your “KPI” is no longer required once the project ends, the activity stops, or the target is met, then it was never a real KPI in the first place. It was driving completion, not improvement.

A KPI is:

“…a quantification that provides objective evidence of the degree to which a performance result is occurring over time.” – What Are KPIs and Performance Measures?

The following are good examples of meaningful measures for digital transformation:

• Percentage of customer transactions completed through digital channels, by week
• Percentage of customer requests resolved fully through self-service without human intervention, by month
• Average processing time for workflows undergoing automation, by month

And from these examples, we can see that a good KPI always includes:

1. a statistic, which shows the method of quantification
2. a performance attribute, which shows the result being quantified
3. a frequency, which shows how often the measure values are tracked
4. [optional] a scope or population, which shows if the measure excludes or includes any specific range of data

FUNDAMENTAL CONCEPT:
KPIs that actually measure performance follow PuMP’s “recipe” for quantitatively expressing a measure of performance.

Step 2: Start with measurable results — not data — to design good KPIs.

This is the step most organisations skip — and it’s the main reason so many KPIs seem useless and irrelevant.

KPIs that actually measure performance provide direct evidence of success. But many people jump straight into choosing numbers before they’ve defined what success means. Without clarity about what exactly they need evidence of, their KPIs are driven (and therefore limited) by data availability.

Better KPIs result when we first make sure our goals are measurable (able to be measured). This comes from how they are worded or articulated. In PuMP, we use five measurability tests to translate vague goals into measurable results (three of those tests are here).

The vague goal “Enhance customer experience” might become measurable results (worded as “future facts”) like these:

• Customers describe every experience as excellent.
• Customers recommend us to their peers.
• Customers give more of their business to us and less to competitors.
• Customers rarely need to escalate issues.
• Customers can access services when they need them.
• Services are delivered correctly the first time.
• Commitments made to customers are kept.
• Customers feel respected and listened to.
• Customers’ expectations are exceeded.
• [You can imagine there are many more possible, right?]

Of course, any vague goal can mean a huge variety of different things, depending on context. This is why we cannot design a meaningful KPI for a goal until it is expressed in a measurable way that is relevant to our own context.

FUNDAMENTAL CONCEPT:
If we can’t finish the sentence “We will know the result of this goal is successfully achieved when we can see, hear, touch, observe…” then we’re not ready to choose measures yet.

Step 3: Evaluate what you already measure (and cull ruthlessly).

Most organisations already track dozens — probably hundreds — of metrics.

The problem isn’t usually a lack of data. It’s a lack of useful data, or a lack of approach to match useful data to the right KPIs that evidence success. But this doesn’t mean that every existing KPI is useless.

Before throwing out every KPI and starting from scratch, it’s worth evaluating what you already have.

You’ll make one of four choices for each KPI:

1. Keep it with the goal it’s currently good evidence of
2. Move it to a different goal that it’s better evidence of
3. Test it along with other new potential measures for a goal it might be evidence of
4. Trash it if you can’t find a goal it provides evidence of, and it’s not informing any decision process

To help make these decisions about each KPI, set up an evaluation framework (or use this KPI Evaluation template), that contains the PuMP criteria for excellent KPIs.

FUNDAMENTAL CONCEPT:
If no-one uses a measure to make better decisions, it’s not a KPI. It’s just numeric noise. In contrast, a KPI that actually measures performance will give you signals you can’t afford to miss.

Step 4: Design measures — don’t brain-dump them.

Have you noticed that most KPI conversations start with someone asking a question like this:

So, what could we measure?

Wrong question.

It triggers a brain dump of KPIs that leads us down a rabbit-hole as we rush to quick-fix KPIs. We end up creating the wrong KPIs for the goals we have. Or we end up with very trivial measures that don’t help us make better decisions.

Professional KPI design is a structured process. The key steps in PuMP’s Measure Design technique provide that structured process:

1. Start with a measurable result.
2. Identify what sensory evidence would really prove that result is improving.
3. Build a list of potential quantitative measures from that evidence.
4. Evaluate the potential measures to choose the one that is the best evidence of the result.
5. Double-check for any unintended consequences of the chosen measure (go back to Step 2, 3 or 4 if required).

FUNDAMENTAL CONCEPT:
KPIs that actually measure performance are direct, quantitative evidence of your goal, that are also (importantly) feasible to implement and report.

Step 5: Build buy-in from the people who use the measures.

Even the best-designed KPI is useless if:

• Leaders don’t trust it
• Teams don’t understand it
• No-one feels ownership for it (let alone accountable for it!)

These are the consequences of not involving the right people in the selection of KPIs they are expected to use.

Involving people doesn’t mean everyone has to be in the room while all the above four Steps are being followed. But, they do need the opportunity to see the thinking thread that led to the KPIs, and the opportunity to have their say.

Some proven approaches to building KPI buy-in from people include:

• Co-designing measures with the people affected, by inviting them to join in on Steps 2 and 4 (or part of your Measures Team)
• Explaining how each KPI links to strategic goals, such as by taking them on a tour through a PuMP Results Map (if you have one)
• Capturing your entire measure design process (like we do with all the PuMP templates) and openly discussing risks, assumptions, and limitations (like we do with a PuMP Measure Gallery)

Regardless of the approach we use, it needs to feel like an invitation, not an order.

FUNDAMENTAL CONCEPT:
KPIs that actually measure performance are understood, valued, owned and used by the people who will improve that performance through their decisions, learning and actions.

Step 6: Implement, interpret, and use the KPIs.

Choosing KPIs that actually measure performance is only the beginning. They won’t improve performance until they are put into use. Putting KPIs into use has a few significant stages, including:

1. Collating or collecting the required data and calculating measure values
2. Reporting the KPI values in an appropriate way to see signals
3. Interpreting signals from KPIs validly, to avoid wasting resources and time acting on the wrong conclusions
4. Diagnosing causes of signals, to learn where the leverage is to improve performance
5. Taking deliberate actions to improve performance and close performance gaps

If a KPI does not lead to learning or improvement, it is not a performance measure — it’s just data.

FUNDAMENTAL CONCEPT:
KPIs that actually measure performance need to be implemented and put to use to prioritise and close performance gaps.

The basic test of a KPI that actually measures performance?

Ask yourself — and your leaders or colleagues — this question:

If this measure showed us a true signal of change next month, would we know, with confidence, what to do differently?

If the answer is no — it’s not a meaningful KPI yet.

Because the purpose of KPIs isn’t reporting. It’s improving performance.