OSEMN Framework and KPI: The Basis of Data Analytics

In your data analytics career, you'll find countless projects with different sources, methods, and business objectives. But they all have common elements — and we need a framework to organize these elements. In this paper, we explain two concepts that form the foundation of data analytics: the OSEMN framework and the KPI (Key Performance Indicator).

This post what is data analytics a continuation of its guide.

What is the OSEMN framework?

The OSEMN framework was first described by Hilary Mason and Chris Wiggins to divide data projects into parts. The author pronounces it as «awesome», but it actually consists of the letters O-S-E-M-N and states five steps: acquire (top), scrub (clean), explore (explore), model (modeling), interpret (interpret). This framework helps you break down big data tasks into smaller, manageable sections.

Obtain — Collect

Depending on the purpose and the business question you want answered, the data may come from internal sources (e.g. sales dataset) or external sources (government databases, financial market dataset). Sometimes it's obvious where the dataset will come from; sometimes it takes research and creativity-you may even have to do a survey and create your own dataset.

Scrub — Clean

This step makes the dataset usable: adjusting formats, deciding how to deal with missing data points. Good cleaning prevents further problems. If you start with dirty data, the perfect model will do the wrong thing — the principle of «garbage entry, garbage outlet» (garbage in, garbage out).

Explore — Explore

Here, as a directive, you look for patterns, trends, and anomalies at the bottom of the dataset — by categorizing the dataset and visualizing it with simple graphs.

Model — Modeling

Better understand patterns and generate predictions with a statistical or mathematical approach. There are many easy-to-apply models for data analytics.

iNterpret — Commenting

You interpret results and build visualizations, stories, and presentations for others to understand.

OSEMN is only one of the popular frameworks; while there are differences with others, they all involve the same basic steps. OSEMN is a great choice because it's simple and flexible.

Why is OSEMN useful?

When you first approach a data project, it can seem big and complicated-where to start? That's where OSEMN comes in. It breaks a big task into five manageable steps, so you only focus on one step each time. This makes it easier and helps avoid skipping any important steps-for example, clearing the dataset or interpreting the output correctly. This framework is constantly useful for structuring real projects.

Notice that the steps of OSEMN are not too different from the previous definition of «what is data analytics?» — it just assembles those steps in a neat way. During the Explore phase, you look for patterns underneath the dataset as a directive; during the model phase, you turn those patterns into future predictions with a mathematical approach.

Start with: SMART GOAL

Before each data analysis, it is very helpful to formulate a clear goal. A good goal should be SMART. Example: “By the end of this year, I want to run the Honolulu marathon in less than four hours.» Why is this a SMART?

• Specific (specific) — which marathon is known.
• Measurable (measurable) — the marathon is 42 km and the goal is less than 4 hours.
• Achievable — possible if you exercise.
• Relevant (compliant) — in line with the overall fitness plan.
• Time-bound (closed) — need to prepare by Dec

What is a KPI?

KPI (Key Performance Indicator — key performance indicator) is a measurable value that tracks progress towards a goal. In the marathon example, the Strava app provides information like distance, time, pace, altitude, and calories. It's all interesting, but most important is the pace: if you miss 11 minutes and 49 seconds per mile and want to finish in less than 4 hours, you should land in about 9 minutes per mile. Temp here is a KPI — a metric indicating whether or not the goal has been reached.

A good KPI has three characteristics: it can be quantified, it is oriented (goes up or down), and it is directly related to the goal. KPIs help the analyst separate the really important metrics from the dataset that's just as interesting.

KPIs in different fields

The concept of KPI works in every field, just metrics change. Let's look at a few examples so you can better understand the principle:

• E-commerce: if the goal is to increase sales, the primary KPI may be the conversion rate (how many percent of visitors make a purchase), and the secondary KPI may be the add to cart ratio.
• Content/blog: if the goal is to build an audience, KPI is the number of unique monthly readers or an increase in email subscribers.
• SaaS/application: if the goal is to retain the user (retention), KPI is the monthly active user or churn (abandonment) rate.
• Fitness (marathon example): if the goal is to run less than 4 hours, the KPI is the pace in miles.

Notice that there are a lot of «interesting» metrics anyway, but only one directly measures the achievement of the goal. The job of the analyst is to select these key metrics and focus on them. Choosing the wrong KPI can take the whole analysis in the wrong direction-for example, if you just look at site views instead of measuring sales, you may see too much traffic and too late realize that there are few sales.

Quantity and qualitative dataset

KPIs are usually quantitative — that is, they are measured numerically. But data can also be qualitative: customer reviews, interviews, open survey responses. A qualitative dataset helps answer the question “why”, and a quantitative dataset helps answer the question “how much”. A good analyst combines the two: the number indicates a problem (for example, a conversion has fallen), and the quality dataset explains why (customers complain about the payment process).

Primary and secondary KPIs

Imra, owner of the flower business Calla & Ivy in Amsterdam, opened her website and set a SMART goal: «To get 10,000 site visits in May.» Its KPI is the number of monthly site visitors — measurable, direct, and directly connected to the goal. The dashboard of their site gives other metrics such as page number of views, average visit time, and number of transactions, but these are not keys to their specific purpose.

Metrics can also be separated as primary and secondary KPIs. Primary KPI measures whether a direct goal has been reached (monthly number of visits). The Secondary KPI correlates to primary, but does not confirm that the goal alone is reached-for example, Calla & Ivy's brand recognition. The likelihood of a visit increases as recognition increases, but recognition alone doesn't guarantee a visit. If you identify KPIs in advance, you're protected from being lost in a sea of data.

Frequently asked questions (FAQs)

Is the OSEMN framework mandatory?

No, but very helpful. Divides large projects into small, manageable phases and helps you keep no steps in mind.

What is the difference between KPI and regular metering?

Every KPI is a metric, but not every metric is a KPI. KPI is the primary metric chosen that directly measures goal attainment.

Why is SMART goal important?

Without a clear, measurable goal, you can't know what data to collect and how to measure success. The objective determines the direction of the entire analysis.

What to do if the data is too high

Define KPIs in advance. They show you which metrics are truly important and keep you from getting lost in a myriad of interesting but unimportant numbers.

Who created OSEMN?

Described by Hilary Mason and Chris Wiggins to structure data analytics projects.

What is an example of a primary and secondary KPI?

The primary KPI for site traffic is monthly visits, and the secondary KPI is brand recognition — recognition correlates with visit, but does not confirm the purpose alone.

What is a good KPI?

The three features are quantifiable, oriented (going up or down), and directly related to purpose.

What is the difference between a quantitative and qualitative dataset?

The quantitative dataset answers the question «how much» numerically; the qualitative dataset (reviews, interviews) helps the question «why». Together, the two give a more complete view.

What does each letter in OSEMN state?

O — Obtain (Gather), S — Scrub (Clear), E — Explore (Explore), M — Model (Modeling), N — iNterpret (Interpret).

Result: frame and size

The concept of the OSEMN framework and KPI are the two cornerstones of data analytics. OSEMN answers you the question «how» — which order of steps to take — The KPI answers the question “how will I measure success”. The SMART goal combines the two: aim clearly, measure the KPI, and show the OSEMN how to reach it. If you master these three concepts, you can confidently approach any data project — without mixing where to start or how to measure success.

As practical advice, start small: choose a SMART goal, assign it a KPI, and pass the five stages of OSEMN over a small dataset. This experience will help you “feel” the frame and build self-confidence for the next, bigger projects. Theory is important, but true learning comes in practice.

OSEMN in practice: a simple example

Let's see in an example how OSEMN works. Let's say there is a small online store and the goal is to “increase the number of repeat customers by 20% this quarter”; KPI's are the share of customers who make repeat purchases. Obtain: you extract the order dataset of the last year from the sales program. Scrub: you clear repeat orders, empty email boxes, and incorrect dates. Explore: You can explore the dataset to see that most repeat customers buy a second time within 30 days of their first purchase.

The model: based on this example, you build a simple model that predicts which customers will make a repeat purchase. iNterpret: this is how you interpret the result — 30 days after the first purchase is critical, so sending an email discount during that period can increase repeat purchases. As you can see, each step builds on the previous one, all together leading to a specific business decision. If it were not for OSEMN, it would be difficult to structure this process and not take any steps.

Why prioritize goals and KPIs?

There is a truth that will appear repeatedly in this guide: there is a huge amount of data available for most data tasks, and this abundance can quickly overwhelm people. If you determine your purpose and its associated KPIs before the analysis, you are protected from being lost in a sea of data. A clear goal tells you what data to collect, while a KPI tells you how to measure success. Without these two, the analysis remains inconsistent-you see a lot of numbers, but you don't know which ones are important.

Practical advice: start each project with a SMART goal and a KPI or two. Then go through the OSEMN steps. These two concepts are frameworks that hold the whole process together-what the goal is, the KPI measurement, and the OSEMN steps. When the goal and KPIs are ready, you're ready to start analysing.

Next step: Learn the first two steps of OSEMN — collecting and cleaning the dataset.

👉 Next Post data collection and cleaning