Quick-Reference Glossary & Final Check

What you'll learn

Look up any term from the course in plain English
See a real workplace example for every term
Prove it with a final knowledge check

Keep this page bookmarked. Every term you met across the twelve lessons is gathered here in one sentence of plain English, each with a real workplace example — so when a word flies past in a meeting, you have somewhere to check. Skim it once now to see how the whole course fits together, then come back whenever you need a lookup. When you’re ready, the final knowledge check at the bottom spans the entire course; passing it completes the last module, and once every module is complete the site issues your certificate.

Twelve lessons, six movements, one habit at the center: ask what a number is really saying before it decides anything.

Text description of this diagram

At the center is a blue hub labeled “read it right.” Around it sit the six movements of the course: Sources & metrics (where a number is from, and whether it’s really a KPI), Averages (mean vs median), Percentages (up 300% — from what?), Charts (read the axes first), Cause & experiments (together isn’t because; the A/B test), and Significance & worth (real isn’t the same as important). Each is joined to the hub by a dashed connector with moving dots, because in real meetings the skills fire together — a chart hides a baseline trick, a survey feeds a causation claim, an A/B result gets summarized as an average. Six movements, one connected toolkit.

The glossary

Every term is defined at recognition level — enough to follow a conversation and ask the right question, not to run the analysis yourself.

Data literacy, sources & metrics

Data literacy: The ability to understand, question, and communicate with data — a reading skill, not a math skill. — Example: two teams read one dashboard differently, and the data-literate one asks how “customer” is defined.
Data source: The place a number was first created, like a CRM or a website. — Example: “which source is that from?” often explains why two reports disagree.
Data warehouse: A central store that copies data from many systems so it can be combined and analyzed. — Example: the company-wide dashboard reads from the warehouse, not the live CRM.
Dashboard: A screen gathering several metrics and charts in one place — convenient, and not automatically the truth. — Example: the sales dashboard is green, but it’s a day behind because one feed updates overnight.
Metric: A number tracked over time or against something, so it carries meaning. — Example: “visits up 12% this month” is a metric; a bare “1,200 visits” is just a measure.
Target: The value you’re aiming for. — Example: “reach 1,500 visits a week” turns a metric into a goal you can pass or miss.
Key performance indicator (KPI): One of the few metrics that genuinely show whether you’re hitting goals that matter. — Example: revenue growth is a KPI; “emails sent” usually isn’t.
Leading indicator: A metric that moves early and hints at what’s coming. — Example: demo bookings rise before the revenue they predict.
Lagging indicator: A metric that confirms a result after the fact. — Example: quarterly revenue tells you what already happened.
Vanity metric: A number that reliably rises and impresses but isn’t tied to a decision. — Example: “total sign-ups ever” only ever goes up and says nothing about active users.

Summarizing numbers

Average: The everyday word for the mean — add everything up and divide by the count. — Example: “average deal size $85k” can be pulled far above what a typical deal looks like.
Mean: The technical name for that same average; every value, including extremes, moves it. — Example: one huge salary lifts the mean above what most people earn.
Median: The middle value when everything is lined up in order; outliers barely touch it. — Example: median salary describes a real, typical employee.
Outlier: A value wildly unlike the rest. — Example: the one deal at $250k among deals near $45k.
Distribution: The overall shape of the data — where values cluster and how they spread. — Example: “average customer spend” hides two clusters, hobbyists and pros.

Percentages & comparisons

Baseline: The starting number a change is measured from. — Example: “up 300%” turned out to mean 4 sign-ups became 16.
Percentage: A number expressed as a share of 100 — meaningless without knowing “of what.” — Example: “80% approval” needs the count behind it.
Percentage change: The relative move from a baseline: (new − old) ÷ old. — Example: 2% to 3% conversion is a 50% increase.
Percentage point: The absolute gap between two percentages. — Example: 2% to 3% is a rise of one percentage point — and 50% relative.
Absolute change: The raw difference in real units. — Example: “+12 sign-ups” alongside “+300%” tells the honest story.
Relative change: The change expressed as a proportion of the starting value. — Example: the same 15-point rise is a 30% relative gain on a base of 50.

Charts, cause & effect

Correlation: Two things tend to move together — a pattern, not an explanation. — Example: app users spend more (but maybe big spenders just install the app).
Causation: Changing one thing actually changes the other. — Example: provable that the feature lowers churn only by a controlled test.
Bias: A systematic slant that pushes results away from the truth in a consistent direction. — Example: surveying only fans biases satisfaction upward.
Variation: The natural wobble in any measurement — two identical groups still differ a little by chance. — Example: the same button shown twice converts at slightly different rates.

Surveys & samples

Population: Everyone you want to know about. — Example: all customers, not just the ones at the superfan event.
Sample: The subset you actually heard from. — Example: the 40 people who answered the survey.
Sample size: How many are in the sample; bigger samples wobble less. — Example: “83% prefer it” from six people is not evidence.
Representative sample: A sample that mirrors the population, achieved by random selection. — Example: 600 randomly chosen customers beat 40,000 forum volunteers.
Selection bias: The people studied weren’t typical to begin with. — Example: surveying only customers whose issue was resolved.
Response bias: Those who bother to answer differ from those who don’t. — Example: NPS inboxes fill with the delighted and the furious.
Survey: A method of asking a sample of people questions to estimate what a whole population thinks. — Example: the engagement survey with a 28% response rate.

Experiments & significance

A/B test: Show version A to a random half and version B to the other half, then compare an outcome. — Example: “Create Account” vs “Start Free” on the sign-up button.
Experiment: Any controlled test that changes one thing on purpose and compares the result. — Example: an A/B test is the everyday business kind.
Control group: The half that sees the current version — the baseline. — Example: the visitors who kept seeing “Create Account.”
Treatment group: The half that sees the change. — Example: the visitors shown “Start Free.”
Random assignment: Deciding who sees which version by chance, so the groups are alike except for the change. — Example: a coin flip per visitor is what makes the test fair.
Lift: The improvement of one result over another. — Example: 4.0% to 4.6% is a 0.6-point (15% relative) lift.
Statistical significance: The difference is bigger than ordinary random variation would produce — a luck-check, not an importance-check. — Example: “p was 0.03” means luck rarely produces this gap.
Confidence: How sure we are a result isn’t just luck; it rises with sample size and effect size. — Example: 40,000 users per side gives more confidence than 400.
Practical significance: Whether the effect is large enough to make a real-world difference. — Example: a 0.02% lift can be significant yet practically meaningless.

Match the term to its movement

Drag each item into the part of the course it belongs to — or tap an item, then tap a category.

Summarizing numberswhat's typical?

Percentagesfrom what, to what?

Surveys & sampleswho was asked?

Experimentsis it real?

Tip: drag with a mouse, or tap an item then tap a category on touch screens. Get one wrong and the answer key appears.

Final knowledge check

This one spans the whole course. Pass it and the last module is complete.

Final check

1. A report shows "average deal size $85k," but the median deal is $46k. The gap most likely means…

2. "Conversion improved 50%!" turns out to be a move from 2% to 3%. The most complete description is…

3. An A/B test on 5 million users per side finds a 0.03% lift, statistically significant. The right takeaway is…

Answers explained

C is correct — a mean well above the median is the signature of a right-skew: a few large deals pull the average up past what a typical deal looks like. (If you picked A: the gap is a normal property of skewed data, not an error. If you picked B: if most deals were $85k, the median would be near it too.)
B is correct — 2% to 3% is a one-point absolute rise and a 50% relative rise; naming both is the honest, complete description. (If you picked A: that would be 2% to 52%. If you picked C: it’s very meaningful once you separate points from percent.)
A is correct — significance confirms the lift is real, but acting on it depends on business significance: how big, how many it touches, and what it costs. (If you picked B or C: significance is a luck-check — at 5 million users even a trivial effect goes significant.)

You did it

You started this course nodding along to “the average is $85k,” “up 300%,” “is that lift significant?”, and “satisfaction rose 30%.” Now you can open every one of those boxes — find the source, check the baseline, read the axes, separate correlation from cause, ask who was surveyed, and tell “real” from “worth it.” You never needed to be a statistician. You needed to be curious out loud, and now you are. Your certificate appears once all modules are marked complete. Well earned.

Course summary

Data literacy is the ability to understand, question, and communicate with data. A number is only meaningful once you know its source, definition, time period, and baseline. Averages can hide important differences, and medians often better describe a typical result. Percentages need their original values and counts. Charts can mislead through scales, labels, and selective time periods. Correlation doesn’t automatically prove causation. Surveys depend on who was asked, who responded, and how the questions were worded. A/B tests compare alternatives under controlled conditions. Statistical significance describes evidence, not business importance. Good decisions weigh impact, cost, risk, uncertainty, and scale together. Data-literate employees don’t accept or reject numbers automatically — they ask thoughtful questions. And you don’t need to be a statistician to become genuinely confident with data.