Business Analytics Terms Every Beginner Should Know 2026

Glossary · Business Analytics · 2026

Business Analytics Terms Every Beginner Should Know

NR Neha Rathore · Lead Trainer, Business Analytics | Updated 12 September 2026 | 14 min read Free · Glossary

You don't need to be a statistician to work in analytics — but you do need to speak the language. The terms on this page are the minimum vocabulary to be useful in any data conversation.

Quick summary — the analytics vocabulary you actually need

Business analytics has its own vocabulary, and beginners who don't learn it struggle to contribute — even when they understand the underlying concepts. The terms fall into three buckets: metrics (what you measure — KPI, MAU, churn, conversion rate), data (where it comes from — data warehouse, ETL, schema, dimension), and statistics (how you trust it — mean, median, distribution, correlation). Beyond those, there are business terms that show up in every analytics conversation: CAC, LTV, ARPU, funnel, cohort, A/B test. You do not need to memorise every term — you need to understand enough to follow a conversation, ask a good question, and know which term you are unsure about. This guide covers the terms you will hear in your first month on any analytics team, in plain English.

In this guide you will learn:

Metrics vs KPIs vs dimensions — the single distinction that unlocks analytics.
Data terms — warehouse, lake, ETL, schema, dimension, fact.
Statistics terms — mean, median, distribution, correlation vs causation.
Business KPIs — CAC, LTV, churn, ARPU, funnel, cohort.
The analytics maturity ladder — descriptive, diagnostic, predictive, prescriptive.

SECTION 01Metrics vs KPIs vs dimensions

This is the single most important distinction in analytics. A metric is a number you measure. A KPI is a metric you care about because it maps to a business goal. A dimension is the thing you slice the metric by.

Term	Definition	Example
Metric	A number that measures something	Number of sign-ups yesterday
KPI	A metric tied to a business goal	Weekly active users (target: 10,000)
Dimension	A way to slice a metric	Sign-ups by channel, by region, by device
Measure	A numeric column you aggregate	SUM(revenue), AVG(session_time)
Attribute	A non-numeric property	Customer name, product category

Key insight: Almost every analytics question boils down to "which metric, sliced by which dimension, compared to what?" If you can answer that sentence clearly, you can have a productive conversation with a data team.

SECTION 02Data terms every analyst should know

These are the terms you will hear from data engineers and analytics engineers. You do not need to build these systems — you need to know what they are so you can ask the right questions.

Term	What It Means	Why You Should Care
Database	Where app data lives (MySQL, Postgres, MongoDB)	Raw material for analytics
Data warehouse	Central analytics store (Snowflake, BigQuery, Redshift)	Where your dashboards read from
Data lake	Raw storage of unstructured/structured data	Where logs, events, and files live
ETL / ELT	Extract, Transform, Load — how data moves between systems	Why your dashboard "updates overnight"
Schema	The structure of tables (columns, types)	Determines what questions you can ask
Fact table	A table of measurable events (orders, sessions)	Where the numbers come from
Dimension table	A table of descriptive attributes (users, products)	How you slice the facts
Data model	The design of how tables relate	Why some queries are fast and others slow
Data quality	Completeness, accuracy, timeliness of data	Why "the dashboard is wrong" happens
Data governance	Who owns what data, and under what rules	Why you can't just query anything

Key insight: A beginner does not need to know how to build a warehouse. But knowing that your dashboard reads from a warehouse via an overnight ETL job — and therefore cannot show "right now" data — saves you from a dozen embarrassing conversations.

SECTION 03Statistics terms — without the textbook

You don't need a maths degree. You need to know what these terms mean when someone uses them in a meeting, and which ones to be careful with.

Term	What It Means (Plain English)	When It Matters
Mean	The average — sum divided by count	Skewed by outliers
Median	The middle value when sorted	Better than mean for skewed data (income, spend)
Distribution	How values are spread out	Two groups can have the same average and different distributions
Correlation	Two things move together	Does NOT mean one causes the other
Causation	One thing causes another	Requires experiment or strong evidence
Statistical significance	The result is unlikely to be noise	Why small samples can be misleading
Confidence interval	A range the true value is likely in	Better than a single number
Sample	A subset used to estimate the whole	Why 100 responses can be worth more than 10,000
Bias	Systematic error in data or method	The most under-discussed problem in analytics
A/B test	Randomised experiment comparing two versions	The gold standard for causal claims

Key insight: The one word to internalise is correlation ≠ causation. Most bad analytics decisions come from confusing the two — from "sales went up when we launched the ad" to "the ad caused sales to go up". Those are different claims, and only one of them is provable.

SECTION 04Business KPIs — CAC, LTV, churn, ARPU

These are the terms you hear in business reviews and growth meetings. They are not complicated — but they are often misused, so it helps to know the exact formulas.

Term	Formula	What It Tells You
CAC — Customer Acquisition Cost	Marketing + sales spend ÷ new customers	How much it costs to win one customer
LTV — Lifetime Value	ARPU × average customer lifetime	How much a customer is worth over time
ARPU — Average Revenue Per User	Revenue ÷ active users	How much each user is worth per period
Churn	Customers lost ÷ customers at start	How many are leaving
Retention	Customers still active ÷ customers at start	The flip side of churn
Conversion rate	Conversions ÷ total visitors	How effectively you turn visits into actions
MAU / DAU	Monthly / Daily Active Users	Engagement level
Funnel	Series of stages users pass through	Where users drop off
Cohort	Group of users sharing a start period	How retention changes over time
NPS — Net Promoter Score	% promoters − % detractors	Customer satisfaction benchmark

Key insight: The LTV ÷ CAC ratio is one of the most important numbers in any consumer business. A ratio above 3 generally means the business can profitably spend on growth; below 1 means every new customer loses money. Most executives know this — so if you can compute it confidently, you can join a very different conversation.

SECTION 05The analytics maturity ladder

Analytics is not one thing. It is a ladder of increasingly ambitious questions — and each rung has a name you will hear in job descriptions.

-- Descriptive analytics — what happened?
-- The most common form of analytics. 80% of a beginner's work.

-- Revenue by month
SELECT
  DATE_TRUNC('month', order_date) AS month,
  SUM(amount) AS revenue,
  COUNT(*) AS orders
FROM orders
WHERE order_date >= '2026-01-01'
GROUP BY 1
ORDER BY 1;

-- Revenue by channel and region
SELECT
  channel,
  region,
  SUM(amount) AS revenue,
  COUNT(DISTINCT customer_id) AS customers
FROM orders
WHERE order_date >= '2026-01-01'
GROUP BY 1, 2
ORDER BY 3 DESC;

-- What this answers: "What happened last month?"
-- It does NOT answer: "Why?" or "What will happen next?"

-- Descriptive analytics is where every analyst starts.
-- It's also the foundation — you can't diagnose what you can't describe.

-- Diagnostic analytics — why did it happen?
-- The layer where the real value starts.

-- Compare this month to last month, by channel
WITH this_month AS (
  SELECT channel, SUM(amount) AS revenue
  FROM orders
  WHERE DATE_TRUNC('month', order_date) = DATE_TRUNC('month', CURRENT_DATE)
  GROUP BY 1
),
last_month AS (
  SELECT channel, SUM(amount) AS revenue
  FROM orders
  WHERE DATE_TRUNC('month', order_date) = DATE_TRUNC('month', CURRENT_DATE - INTERVAL '1 month')
  GROUP BY 1
)
SELECT
  t.channel,
  t.revenue AS this_month,
  l.revenue AS last_month,
  ROUND(
    (t.revenue - l.revenue) * 100.0 / NULLIF(l.revenue, 0), 1
  ) AS pct_change
FROM this_month t
LEFT JOIN last_month l USING (channel)
ORDER BY pct_change;

-- Then go deeper: is the drop due to fewer orders or lower order value?
SELECT
  channel,
  COUNT(*) AS orders,
  SUM(amount) / COUNT(*) AS avg_order_value
FROM orders
WHERE DATE_TRUNC('month', order_date) = DATE_TRUNC('month', CURRENT_DATE)
GROUP BY 1;

-- What this answers: "Why did revenue drop this month?"
-- It narrows the search: fewer orders? smaller orders? both?

-- This is where most analyst interviews start to probe.

-- Predictive and prescriptive analytics.
-- These require statistics or ML, but the SQL still starts the story.

-- PREDICTIVE: build a simple churn-risk feature set
WITH customer_activity AS (
  SELECT
    customer_id,
    MAX(order_date) AS last_order_date,
    COUNT(*) AS total_orders,
    AVG(amount) AS avg_order_value,
    CURRENT_DATE - MAX(order_date)::date AS days_since_last_order
  FROM orders
  GROUP BY 1
)
SELECT
  customer_id,
  days_since_last_order,
  total_orders,
  avg_order_value,
  -- Simple rule-based risk score (real models use ML)
  CASE
    WHEN days_since_last_order > 90 THEN 'high_risk'
    WHEN days_since_last_order > 45 THEN 'medium_risk'
    ELSE 'low_risk'
  END AS churn_risk
FROM customer_activity
ORDER BY days_since_last_order DESC;

-- PRESCRIPTIVE: which segment to target with a win-back campaign?
SELECT
  churn_risk,
  COUNT(*) AS customers,
  ROUND(AVG(avg_order_value), 2) AS avg_value
FROM (
  SELECT
    customer_id,
    avg_order_value,
    CASE
      WHEN CURRENT_DATE - MAX(order_date)::date > 90 THEN 'high_risk'
      WHEN CURRENT_DATE - MAX(order_date)::date > 45 THEN 'medium_risk'
      ELSE 'low_risk'
    END AS churn_risk
  FROM orders
  GROUP BY 1, 2
) sub
GROUP BY churn_risk;

-- What this answers: "Who is likely to churn, and who should we target?"
-- Predictive says what will happen; prescriptive says what to do.
-- Most entry-level analyst roles focus on descriptive and diagnostic.

analytics-maturity.sql

Key insight: The four levels of analytics — descriptive, diagnostic, predictive, prescriptive — map directly to job descriptions. A BI analyst mostly does descriptive; an analytics engineer or data scientist does more predictive; an MLOps or ML engineer operationalises prescriptive. Knowing the ladder tells you where you are and where you could go.

SECTION 06Terms that beginners confuse

These pairs are routinely mixed up — even by people who have been in analytics for a year. Knowing the difference puts you ahead of most candidates.

Pair	The Difference
Metric vs KPI	All KPIs are metrics, but not all metrics are KPIs. A KPI is tied to a business goal.
Data warehouse vs data lake	Warehouse is structured, query-ready; lake is raw, often unstructured.
ETL vs ELT	ETL transforms before loading; ELT loads first, transforms inside the warehouse.
Correlation vs causation	Correlation means two things move together; causation means one causes the other.
Mean vs median	Mean is the average; median is the middle value. Median is safer for skewed data.
Churn vs retention	Churn is who left; retention is who stayed. They are flip sides of the same coin.
MAU vs DAU	Monthly active users vs daily active users. The DAU/MAU ratio is a measure of stickiness.
Cohort vs segment	Cohort is defined by a start period; segment is defined by a shared characteristic.
Dimension vs measure	Dimension is what you slice by (region, channel); measure is what you aggregate (revenue).

Pro tip: If you can clearly explain three of these pairs in an interview, you already sound more credible than most candidates. Pick the ones you find most useful in real work — I recommend metric vs KPI, mean vs median, and correlation vs causation — and be ready to give an example of each.

SECTION 07Interview Q&A — business analytics terms

Q1What is the difference between a metric and a KPI?

A metric is any number you measure. A KPI (key performance indicator) is a metric tied to a specific business goal. Every KPI is a metric, but not every metric is a KPI. "Number of orders yesterday" is a metric. "Weekly active users target: 10,000" is a KPI — it has a target and connects to a business objective.

Q2What is a data warehouse, and why does it matter?

A data warehouse is a central, structured store of data designed for analytics queries — typically Snowflake, BigQuery, or Redshift. It is where dashboards and BI tools read from. It matters because most analytics questions need data joined across multiple source systems, and the warehouse is where that joining already happened via ETL.

Q3Why is median often better than mean?

Because the mean is pulled by outliers, while the median is not. In income, spend, or session-time data, a handful of extreme values can make the mean unrepresentative. The median gives you the "typical" value. When you see a large gap between mean and median, that is a strong signal that the distribution is skewed — which is often more informative than the number itself.

Q4What is the difference between correlation and causation?

Correlation means two things move together — when one goes up, the other tends to go up (or down). Causation means one thing actually causes the other. Correlation can exist without causation, often due to a third factor. Example: ice cream sales correlate with drowning deaths, but ice cream doesn't cause drowning — hot weather drives both. Only a randomised experiment or very strong evidence can establish causation.

Q5What are the four types of analytics?

Descriptive (what happened), diagnostic (why it happened), predictive (what will happen), and prescriptive (what should we do). Most analyst roles start with descriptive and diagnostic work. Predictive requires statistics or ML. Prescriptive requires decision science or optimisation. Knowing the ladder tells you where a given job sits.

SECTION 08Test yourself — analytics vocabulary

Five questions. No sign-up.

0 / 5

Pick an answer to see why it is right or wrong.

SECTION 09Frequently asked questions

What is the most important analytics term to learn first?

Metric vs KPI vs dimension. That single distinction unlocks almost every analytics conversation. Once you know that a metric is what you measure, a KPI is what matters to the business, and a dimension is how you slice it, you can follow most team discussions and ask meaningful questions.

Do I need to know statistics to work in analytics?

You need basic statistics — mean, median, distribution, correlation vs causation, sample size, and statistical significance. You do not need a full statistics degree. Most working analysts learn the essentials in their first month and pick up the rest on the job. What matters is knowing when a claim might be statistically weak.

What is the difference between a data analyst and a data scientist?

A data analyst usually focuses on descriptive and diagnostic analytics — dashboards, reports, "what happened and why". A data scientist usually works on predictive and prescriptive analytics — models, experiments, "what will happen and what should we do". In many companies the boundary is fuzzy, but that is the general shape.

What is a good LTV-to-CAC ratio?

A common benchmark is 3:1 — lifetime value is three times customer acquisition cost. Below 1 means every customer loses money; above 5 may mean you are under-investing in growth. The exact target varies by industry and business model, but 3:1 is a widely used starting point.

Do I need to know SQL to understand these terms?

Not to understand the terms — but you will be much more effective if you can query data yourself. Basic SQL (SELECT, WHERE, GROUP BY, JOIN) lets you verify the numbers you see in dashboards and answer follow-up questions without waiting for a data team. Most analytics roles assume at least basic SQL.

What I would do in the first 30 daysLayoff · 14 min Why sending 500 resumes won't workJob Search · 14 min Rebuild your resume for the AI job marketResume · 15 min Which skills to learn before applyingSkills · 14 min 90-day career recovery roadmapRecovery · 16 min How AI can help you prepare for interviewsAI · 11 min

Classroom & online · Noida

Learn business analytics — with Uncodemy

Our Business Analytics Training Course covers analytics vocabulary, SQL, Excel, dashboards, and the business metrics that matter — built for beginners who want to work with data teams and make better decisions.

₹32,000 · full programme ~~₹45,000~~

Explore the course Download the syllabus

Analytics vocabulary
SQL + Excel + dashboards
Business KPIs
Weekday & weekend batches