Glossary

Statistics and business-metric terms in one place.

Statistics

Mean→ covered in a lesson: Sum of all values divided by their count. Sensitive to outliers.
Median→ covered in a lesson: The value exactly in the middle of the sorted list. Robust to outliers.
Mode→ covered in a lesson: The most frequent value. Useful for categories.
Variance→ covered in a lesson: Average of squared deviations from the mean; spread in "squared" units.
Standard deviation (σ)→ covered in a lesson: Square root of variance: the typical deviation from the mean, in the same units as the data.
Percentile / quartile→ covered in a lesson: The value below which a given share of the data lies (Q1 is 25%, the median 50%, Q3 75%).
Normal distribution→ covered in a lesson: The symmetric "bell" defined by μ and σ. The 68-95-99.7 rule.
Skewness→ covered in a lesson: Asymmetry of a distribution; a long tail on one side.
Probability→ covered in a lesson: The share of cases in which an event happens over a long run (from 0 to 1).
Law of large numbers→ covered in a lesson: As trials accumulate, the observed share converges to the true probability.
Conditional probability→ covered in a lesson: Probability of an event given that another is known. P(A|B) ≠ P(B|A).
Bayes’ theorem→ covered in a lesson: The rule for updating a probability with new data and the base rate.
CLT→ covered in a lesson: Central limit theorem: the distribution of sample means tends to normal.
Standard error→ covered in a lesson: Typical random spread of an estimate (e.g. of a mean); shrinks as 1/√n.
Confidence interval→ covered in a lesson: A range built so that, over repeated experiments, a set share of intervals (95%) covers the truth.
Bootstrap→ covered in a lesson: Estimating uncertainty by resampling the sample from itself with replacement.
Null hypothesis (H0)→ covered in a lesson: The "no effect" assumption; the one you try to reject.
p-value→ covered in a lesson: Probability of a result at least this extreme if there is no effect. Small p means chance is a poor explanation.
Type I error (α)→ covered in a lesson: False alarm: declaring an effect that does not exist.
Type II error (β)→ covered in a lesson: A miss: overlooking a real effect. Power = 1−β.
Power→ covered in a lesson: The chance of detecting an effect when it truly exists.
t-test→ covered in a lesson: A test of whether the means of two groups differ significantly.
Multiple comparisons→ covered in a lesson: Testing many hypotheses at once; false alarms pile up (Bonferroni correction).
Correlation (r)→ covered in a lesson: A number from −1 to +1: strength and direction of a linear relationship. Not causation.
Regression / OLS→ covered in a lesson: The line minimizing squared errors; predicts one quantity from another.
Simpson’s paradox→ covered in a lesson: An overall relationship flips sign when the data is split into groups.
Overfitting→ covered in a lesson: The model memorized the noise of the training data and fails on new data.
Precision / Recall→ covered in a lesson: Precision: what share of flagged items is correct; recall: what share of the target class you caught.
ROC / AUC→ covered in a lesson: Classifier quality curve across all thresholds; AUC is the area under it (1 is ideal, 0.5 is chance).
Survivorship bias→ covered in a lesson: Drawing conclusions from what "survived" into the data while ignoring what dropped out.
Goodhart’s law→ covered in a lesson: When a measure becomes a target, it gets gamed and stops being a good measure.

Business metrics

North Star Metric: The single metric that best reflects the core value of the product.
DAU / MAU: Daily / Monthly Active Users; their ratio is "stickiness".
Retention: The share of users who come back after 1/7/30 days.
Churn: The share of users or subscribers who left during a period.
ARPU / ARPPU: Average revenue per user / per paying user.
LTV: Lifetime value: total revenue from a customer over their whole life.
CAC: Customer acquisition cost. Healthy when LTV ≥ 3× CAC.
GMV: Gross Merchandise Value: total value of goods sold through the platform.
Take Rate: The platform’s effective commission: the share of GMV it keeps.
Conversion Rate (CR): The share of users who completed the target action.
AOV: Average Order Value.
NRR: Net Revenue Retention: revenue growth from existing customers (>100% means growth without new ones).
MRR: Monthly Recurring Revenue of a subscription business.
Fulfillment / Match Rate: The share of requests that ended in a successful deal or ride; matching quality.
Same-store sales: Sales of existing locations (excluding new openings): honest chain growth.
Guardrail metrics→ covered in a lesson: Quality limits watched alongside the target metric; they catch gaming (Goodhart’s law).
ACV: Annual Contract Value — the average yearly value of one customer’s contract in B2B/SaaS.
ARR: Annual Recurring Revenue of a subscription business; ARR = MRR × 12.
KYC: Know Your Customer — verifying a customer’s identity when opening an account; mandatory in banking and fintech.
Interchange: The fee a payment network withholds from every card transaction; the main revenue source for neobanks.
TPV: Total Payment Volume — the total value of payments processed by a platform; fintech’s equivalent of GMV.
SKU: Stock Keeping Unit — a unique product listing in a catalog; assortment width is measured in SKUs.
FBO: Fulfillment by Operator — storage and shipping from the platform’s own warehouse rather than the seller’s.
GBV: Gross Booking Value — the total value of services booked on a platform; travel/booking’s equivalent of GMV.
TAC: Traffic Acquisition Cost — what a search engine pays partners for being the default search provider.
ROAS: Return on Ad Spend — revenue per dollar of ad spend.
NPS: Net Promoter Score — a loyalty index: the share of promoters minus the share of detractors.
CPI: Cost Per Install — the cost of one app install; a core acquisition metric in gaming and mobile.
UA: User Acquisition — the team, budget and channels responsible for bringing in new players or customers.
F2P: Free-to-Play — a model where the product is free for everyone, and a minority of paying users drives revenue.
D1 / D7 / D30: Retention on day 1 / 7 / 30: the share of users who return exactly that many days after installing.
ETA: Estimated Time of Arrival — how soon a courier, driver or order will arrive.
LMS: Learning Management System — a platform for running and tracking online courses.
RPM: Revenue Per Mille — revenue per thousand views; the core monetization metric for video creators.
CPC: Cost Per Click — the price of one click on an ad in an auction.
SEM: Search Engine Marketing — paid promotion in search results, as opposed to free SEO.
Look-to-Book: The share of search sessions that end in a booking; a key conversion metric in travel.
Contribution margin: Revenue minus variable cost per unit (an order, a ride); shows whether a platform earns on each transaction before fixed costs.
Bookings: The total of all player payments over a period before app-store fees; the core revenue metric of F2P games.
Trial → Paid: Conversion from a trial period to a paying subscriber.
CTR: Click-Through Rate — the share of viewers who clicked an ad, thumbnail or link.
SEO: Search Engine Optimization — free promotion in search results through site relevance and quality, as opposed to paid SEM.
B2B / B2C: Business-to-business / business-to-consumer — selling to companies or directly to end consumers.

🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕🍕

Enjoying the materials?

This site is built by the DataSlice Telegram channel: data analytics in plain words — real cases, metrics, careers. The channel is in Russian.

Open the Telegram channel