Metrics and KPIs: BABOK 10.28 Guide With Frameworks and Formulas

Metrics and KPIs is BABOK technique 10.28. A metric is a quantified level of an indicator, and a key performance indicator (KPI) is an indicator that tracks progress toward a strategic goal [1, 2]. Business analysts are often the people asked to define KPIs [3], and they use the technique to set baselines and targets and to judge whether a solution delivers value. On the CBAP exam it supports Strategy Analysis and Solution Evaluation.

I learned to distrust raw numbers on real projects, not from the book. So I put together the map I wish I had then: the BABOK view, the types and roles of metrics, the main frameworks (North Star, One Metric That Matters, AARRR, HEART, Evidence-Based Management), typical formulas with worked numbers, and the ways metrics mislead. Each popular product metric will get its own breakdown later.

Metrics and KPIs at a glance

Metrics and KPIs is one of the 50 BABOK techniques, and it touches four of the six knowledge areas [4].

Last verified: October 2026.

What BABOK 10.28 says about metrics and KPIs

BABOK separates three words that people often mix: indicator, metric and KPI.

• Indicator. An indicator is a specific numerical measurement that shows the degree of progress toward a goal, objective, output, activity or further input [1]. Example: "share of orders paid online".
• Metric. A metric is a quantifiable level of an indicator that an organization uses to measure progress [1]. Example: "62% of orders paid online in September".
• KPI. A key performance indicator is an indicator that measures progress toward a strategic goal or objective [2]. A KPI is not a special kind of number. It is an ordinary indicator that leadership has linked to strategy.

The four elements

The technique has four elements. Together they turn "we want to improve checkout" into a monitoring system [2].

Two terms from the structure element matter on the exam:

• Baseline is the data from just before or at the start of the measurement period [2]. Without a baseline, a target is a guess. Practitioners advise finding the baseline before you set a target [16].
• Target can be a specific point, a threshold or a range, reached within a set period [2].

The six characteristics of a good indicator

BABOK lists six characteristics of a good indicator [2, 5]. The table shows what each one means in practice.

A secondary source on BABOK-based BI work adds two warnings that the exam likes: dirty data gives dirty results, and once you measure people, they start to change their behaviour to hit the number [5].

Where the technique is used in BABOK tasks

IIBA task cards list Metrics and KPIs among the frequently used techniques in three tasks [6, 7, 8]:

• 3.5 Identify Business Analysis Performance Improvements. The business analyst measures the analysis work itself [6].
• 8.1 Measure Solution Performance. The analyst defines performance measures and uses them to evaluate a solution; KPIs help to quantify issues [7].
• 8.2 Analyze Performance Measures. The analyst reads the measures to understand the value the solution brings [8].

Task 6.2 Define Future State sets business objectives and lists the technique as a reference [9]. A chapter study document maps 10.28 to Planning and Monitoring, Strategy Analysis, Requirements Analysis and Design Definition, and Solution Evaluation [4]. Study guides add the idea that performance measures should balance leading and lagging indicators [17]. For the full picture of where each technique sits, see the map of all 50 BABOK techniques by knowledge area.

Why metrics matter: data are facts, not opinions

Metrics turn an argument about opinions into an argument about facts. In The McKinsey Way, Ethan Rasiel describes problem solving at McKinsey as fact-based [18]. One summary of the book gives two reasons why the firm values facts: they make up for a lack of knowledge, and they close the credibility gap with clients [19].

I think this is the core of the whole technique. A number on its own decides nothing. A well-defined metric is a fact: the team looks at it together, agrees on it and judges whether a change worked. If two people define "active user" in two ways, they no longer share a fact, and the discussion slides back to opinions.

There is a catch, though. The map is not the territory. A metric is a map of the product, and every map leaves something out. That means the work is not only to collect facts, but to choose the facts that describe the territory you care about.

The IIBA view of data analytics says the same thing in its own words: decision-makers choose the measures together, and these measures can mix strategic and operational KPIs [20].

A good metric is a ratio, not a raw number

A raw count rarely tells you whether things are good or bad. A ratio or a rate does.

I learned this on a simulator exercise in a product analytics course. The task was to judge whether a messenger was useful. I looked at the data and answered that DAU kept growing, other numbers grew too, so the product seemed useful. I was wrong: the growth hid the real picture, and the course showed it a few chapters later [21]. Since then I treat a growing count as a question, not as an answer.

Lean Analytics by Alistair Croll and Benjamin Yoskovitz gives four tests of a good metric: it is comparative, understandable, a ratio or a rate, and it changes the way you behave [22, 23]. Metrics that fail these tests are vanity metrics: they make you feel good but give no clear guidance on what to do [24]. The last test is the hardest. If you do not know what you would do differently when the number moves, it is a bad metric [23].

Numbers that only make sense over time

Some numbers carry meaning only as a trend. Revenue, monthly active users (MAU), daily active users (DAU), the number of orders or messages: each of these says little on its own. "We have 45,000 MAU" is neither good nor bad. You need the same number a month ago, a year ago, or for a comparable product.

Even a trend can mislead. A growing count can hide a shrinking product. One product analytics course makes this point with a simple model [21, 25]:

Growth metric = number of new users × product metric

DAU equals new users multiplied by retention. Gross profit equals new users multiplied by lifetime value [25]. If marketing brings more new users every month, DAU and revenue can grow while the product gets worse.

The course tells a story about a product manager who shipped a release every month, and each release made the product a little worse. Marketing kept increasing the flow of new users, so MAU and revenue kept rising, the company tracked only those two numbers, and the manager received a bonus. The first signals of the damage appeared only after about ten months [21]. A metric without a denominator describes the business, not the product.

A public guide from the same school adds that growth metrics are easy to inflate and can easily misrepresent reality [26].

Numbers that make sense on their own

Other metrics carry meaning even as a single value, because a denominator is built in:

• Retention rate: the share of a cohort that comes back.
• Churn rate: the share of customers lost in a period.
• Conversion rate: the share of visitors or users who complete an action.
• Average order value (AOV), ARPU, LTV:CAC: money per unit.

You can compare these across periods, segments and even companies, with care. A Day 30 retention of 10% is weak for a social network and strong for an e-commerce app, so benchmarks still depend on the category [21, 27].

The same school calls these product metrics: they describe how the product turns new users into active users, payers and profit, and they do not depend on how many new users arrive [21, 25]. Retention and lifetime value are the two it names as most useful for almost any product with a regular use case [21].

Three traps inside ratios

1. The denominator is the active audience. Metrics such as "share of DAU who bought today" or revenue per daily active user look like product metrics, but they move when the mix of new and old users changes. A spike of new users can raise the share of payers without any product change [21].
2. The same name hides different formulas. "Conversion to purchase" can mean the share of new users who ever buy, the share of DAU who bought today, or the share of sessions with an order. Each is a different metric. Always ask for the exact definition [21, 28].
3. Users versus events. A user who sent ten messages is one unique user and ten events. Mixing the two counts produces wrong ratios. "Active user" should rest on one precise event, not on "any event", because some events are logged for a user who did nothing [21].

Lean Analytics also describes a case where recalculating the main metric changed how the product looked [22]. The lesson is the one the book repeats: the right metric for your situation can show a working product, and a wrong or wrongly calculated metric can make a working product look dead.

Types of metrics

Metrics can be grouped by the question they answer. The groups below overlap; the point is to know what each metric is for.

Note that the product analytics school splits metrics into product metrics and growth metrics, not product and marketing, and calls the split informal [21, 25]. Marketing metrics sit next to both [26].

Roles a metric can play

The same metric can play different roles in different decisions. Retention can be the North Star of one team and a guardrail for another.