A KPI (key performance indicator) is a number that shows how well a team, product or company is doing against something it cares about. A good KPI is tied to a decision: if it moves, someone knows what to do.
TL;DR
- A KPI meaning in one line: a measurable value that tracks progress towards an objective you have already chosen.
- Pick a handful of KPIs per goal; a dashboard full of numbers nobody uses does not help decisions.
- Software teams combine customer and people KPIs with the five DORA delivery metrics and a few business KPIs.
- KPIs monitor what already runs; OKRs push a team to change something within a quarter.
- Start with three to five KPIs, review them on a fixed cadence and retire any that no longer drive a decision.
What is a KPI?
A KPI, or key performance indicator, is a measurable value that shows whether a team or company is moving towards a defined objective. It is used to monitor progress over time, from different points of view, and to decide where to act.
KPIs of the kind we describe here do not refer to turnover. They are mainly oriented towards the customer, the employees and the relationships between the two. In other words, they help you understand whether the enterprise is working and becoming a strong reality with a defined identity.
What are examples of customer and people KPIs?
The classic customer and people KPIs are Customer Satisfaction Score, Net Promoter Score, First Response Time, Customer Retention Rate, Servqual and Employee Engagement. Each one measures a different relationship: with the product, with the brand or with the team that builds it.
Customer Satisfaction Score (CSAT). It checks customer satisfaction when using a product by asking users to rate the product or service on a numeric or other scale. It depends heavily on the user's emotions and mood at the end of the action, so many factors can influence it.
Net Promoter Score (NPS). It measures how inclined customers are to recommend the company to others. Respondents who answer 9 or 10 are promoters, 7 or 8 are passives and 6 or lower are detractors, and the score is the percentage of promoters minus the percentage of detractors. Unlike CSAT, it is less influenced by emotion, because it focuses the respondent on an intention.
First Response Time. A study reported in the original version of this post found that more than a third of users appreciate a company that responds immediately to problems, even when the reply is not an actual solution. In practice this can mean a chatbot or an automatic email that reassures the customer that an expert is already working on the problem.
Customer Retention Rate. It is a company's ability to keep the attention and loyalty of a customer over a long period. It can be calculated weekly, monthly or annually.
Servqual. It asks users to rate their experience against their expectations, using five factors: reliability, the guarantee (the skills and kindness of employees), tangible assets (the look of premises, website and staff), empathy and reactivity.
Employee Engagement. According to a study cited in the original post, long-term employment relationships are among the factors that most help a company succeed, and frequent staff turnover costs much more. The tool is based on interviews that explore employees' relationship with work, satisfaction and sense of belonging to the team.
Which KPIs should a software team track?
A software team should track three groups of KPIs: customer and people indicators, the five DORA delivery metrics, and a couple of business indicators such as cost per feature and adoption. Together they show whether the team delivers quickly, delivers safely and delivers something people use.
The DORA research programme has identified five software delivery performance metrics. They split into throughput (change lead time, deployment frequency and failed deployment recovery time) and instability (change fail rate and deployment rework rate). DORA notes that sharing all five across development, operations and release teams fosters collaboration, while isolating teams with specific metrics can lead to finger-pointing.
| # | KPI | What it tells you | How to compute it |
|---|---|---|---|
| 1 | Customer Satisfaction Score | How happy users are after an interaction | Satisfied responses ÷ all responses |
| 2 | Net Promoter Score | Willingness to recommend | % promoters − % detractors |
| 3 | First Response Time | How fast customers get a first reply | Average time from request to first reply |
| 4 | Customer Retention Rate | Loyalty over time | Customers kept ÷ customers at start of period |
| 5 | Employee Engagement | Satisfaction and belonging in the team | Average score of a regular survey |
| 6 | Deployment frequency | How often you ship | Deployments per period |
| 7 | Change lead time | How fast a change reaches production | Time from committed change to production |
| 8 | Failed deployment recovery time | How fast you recover from a failed release | Time to restore service after a failed deployment |
| 9 | Change fail rate | Quality of releases | Deployments needing immediate intervention ÷ all deployments |
| 10 | Deployment rework rate | Unplanned work caused by incidents | Unplanned deployments due to incidents ÷ all deployments |
| 11 | Cost per feature | What each delivered feature costs | Cost of the period ÷ features delivered |
| 12 | Feature adoption rate | Whether users use what you shipped | Users of the feature ÷ active users |
We deliberately give no target ranges for the delivery metrics: the right range depends on your product, your team and your risk. Set your own baseline first and improve it. If you want to work on the delivery metrics with a team that does this daily, look at our DevOps services.
How do you choose the right KPIs?
Choose KPIs by starting from a decision, not from a metric: write down what you need to decide, then pick the one or two numbers that would change your mind. Fewer indicators, reviewed regularly, beat a wall of charts that nobody reads.
Check each candidate against four questions. Does it link to a goal you have already set? Can you measure it without heroic effort? Can the team influence it? Will someone act when it moves? If the answer to any is no, leave the KPI out. Cost and profitability data sit naturally next to these numbers, and our post on real-time financial visibility shows one way to see them as a project unfolds.
How do KPIs relate to OKRs?
KPIs measure the health of what already runs, while OKRs aim at a change a team wants to make in a quarter. A KPI can become the measure of a Key Result, for example reducing change fail rate, but once the goal is reached it goes back to being a KPI you simply keep an eye on.
Used together, the two answer different questions. KPIs tell you where you stand; OKRs tell you where you want to go next. Read what OKRs are and are not for the full comparison.
To choose and set up the right indicators for your team, write to us at business@wwg.it.
Sources
- DORA, Software delivery performance metrics — https://dora.dev/guides/dora-metrics/
- Wikipedia, Net promoter score — https://en.wikipedia.org/wiki/Net_promoter_score
- Wikipedia, SERVQUAL — https://en.wikipedia.org/wiki/SERVQUAL
- Wikipedia, Customer satisfaction — https://en.wikipedia.org/wiki/Customer_satisfaction
- Wikipedia, Customer retention — https://en.wikipedia.org/wiki/Customer_retention
- Wikipedia, Key performance indicator — https://en.wikipedia.org/wiki/Key_performance_indicator
FAQ
Frequently Asked Questions
A metric is any number you can measure, while a KPI is a metric tied to an objective that drives decisions. Page views are a metric; the share of visitors who ask for a call is a KPI once you set a goal for it. Every KPI is a metric, but few metrics deserve to be KPIs.
You need only as many KPIs as you can review and act on, and for most teams that means three to five. Each KPI should be linked to a specific goal and have a named owner. When a team tracks dozens of numbers, the important ones get lost, so retire any indicator that has stopped changing a decision.
CSAT measures how satisfied a customer is with a specific product or interaction, while NPS measures how likely the customer is to recommend the company. CSAT is shaped by the user's mood right after the action. NPS asks about an intention, so it is less influenced by emotion and better suited to tracking loyalty.
Review KPIs on a fixed cadence that matches how fast the number can change and how quickly you can act. Delivery metrics suit a weekly or per-sprint look, customer satisfaction a monthly one, and engagement or retention a quarterly one. Whatever the rhythm, put the review on the calendar so it happens.
