What Analytics Is, and What It Is Not

A decision with a deadline

Alpenkorb AG is a fictional online grocery and household shop in Zurich. It delivers to eight Swiss regions, from Geneva to St. Gallen, and every order, visit and marketing franc from January 2025 to March 2026 is in the data you will use in this course. On 1 April 2026 Lena Frei starts as head of growth. Her first task is concrete: propose how to split the marketing budget for the second quarter across paid search, social ads, affiliate partners and email. In the first quarter those four channels cost CHF 4,308.96.

Over eight lessons you will do the work Lena has to do: sharpen the question, pick the numbers that matter, query the tables and write the one-page memo that ends the course. Every number in these lessons comes from the same data you query in the labs.

What analytics is

Analytics is the work of turning data into a better decision. The decision comes first. If no choice depends on the answer, the analysis is a hobby, however good the chart looks. For Lena the choice is money: which channel gets more, and which gets less.

• It is answering a specific question with data that fits the question.
• It is saying how sure you are, and what would change your mind.
• It is not collecting every possible number and hoping a pattern appears.
• It is not a dashboard that nobody opens before a decision.
• It is not proof of what someone already believed. A good analysis can end with we were wrong.
• It does not need big data. Alpenkorb has 3,427 orders, small enough for a phone and enough for this decision.

Look at the raw material

Data lives in tables. The orders table has one row per order: when it was placed, by which customer, from which region and marketing source, what it was worth, and whether it was delivered, cancelled or returned. SQL is the language for asking a table questions. The shortest useful question is: show me a few rows.

SELECT *
FROM orders
LIMIT 5;

SELECT * means all columns, FROM orders names the table and LIMIT 5 stops after five rows. Which five? Whatever order the table happens to be stored in. A table has no built-in sort order, so if you care which rows come first you must say so with ORDER BY.

SELECT order_id, order_date, source, net_amount
FROM orders
ORDER BY net_amount DESC
LIMIT 5;

This returns the five largest orders by net_amount, the value after discounts. DESC sorts from high to low; without it SQL sorts from low to high. Naming the columns instead of writing * keeps the result readable.

The trap: starting with the data

The most common mistake in analytics is opening the data before writing down the decision. You find something interesting, then something else, and after a week you have forty charts and no recommendation. Lena will be tempted too. The data holds 17,706 website and app sessions, and in the first quarter of 2026 social ads brought more sessions than any other source. That sounds like success. Whether it is depends on the question, and lesson 4 shows that this number alone says almost nothing about her decision.

Where this goes next

The Four Types of Analytics sorts questions by what they ask of the data. From Business Question to Data Question turns is social working? into something a table can answer. KPIs, Metrics and Vanity Numbers separates the numbers that decide from the ones that decorate. Your First Real Query and Grouping into a Chart-Ready Table build the SQL you need. The Funnel in Numbers follows visitors to first and repeat orders, and The One-Page Decision Memo turns all of it into a recommendation Lena can defend.