Business metrics and KPIs
Learn the core business metrics with their formulas, how to define a KPI precisely, how to avoid vanity metrics, and how to pair every target with a guardrail.
A metric is a number that measures something: orders per day, average bill, percentage of customers who come back. A KPI (key performance indicator) is a metric that is tied to a goal and watched closely. Brew & Bean might track a hundred metrics but only four or five KPIs.
Choosing and defining metrics is the heart of analytics. A dashboard full of the wrong numbers is worse than no dashboard, because people trust it.
Metrics vs KPIs
Think of a car. Speed, fuel, engine temperature and tyre pressure are all metrics. If your goal is to reach Mysuru before lunch, speed and fuel are the KPIs and you watch them constantly. The rest you check only when something looks wrong. A KPI always answers: "Are we getting closer to the goal?"
Common business metrics
Metrics every analyst should know
| Metric | Formula | Brew & Bean example (illustrative numbers) |
|---|---|---|
| Revenue | Sum of sales value | 420 orders in a week, total ₹84,000 |
| Average order value (AOV) | Revenue / number of orders | 84,000 / 420 = ₹200 |
| Orders per day | Orders / days | 420 / 7 = 60 |
| Repeat rate | Customers with 2 or more orders / all customers | 63 of 180 customers = 35% |
| Retention rate | Customers active in both period 1 and 2 / customers active in period 1 | 90 of 150 = 60% |
| Churn rate | 1 - retention rate | 40% |
| Cancellation rate | Cancelled orders / all orders | 21 of 420 = 5% |
| Gross margin | (Price - direct cost) / price | If a cappuccino sells at ₹180 and costs ₹55 to make, (180 - 55) / 180 = about 69% |
| Customer acquisition cost (CAC) | Marketing spend / new customers | ₹6,000 / 40 = ₹150 |
| Conversion rate | People who did the action / people who could | 3,000 website visitors, 120 placed orders = 4% |
The numbers in this table are made up to show the arithmetic; they are not real Brew & Bean data. Always check the maths yourself: 84,000 / 420 is 200, and 63 / 180 is 0.35.
Define each metric precisely
Two people can both say "revenue" and mean different numbers. Does it include cancelled orders? GST? Refunds? The day of the order or of the payment? A good metric definition answers these questions in writing.
Example definition card for one metric
| Field | Definition |
|---|---|
| Name | Daily revenue |
| Formula | Sum of order totals where status is picked_up |
| Time rule | Counted on the date the order was created, India time (IST) |
| Excludes | Cancelled orders, test orders, GST |
| Source | The orders table |
| Owner | The shop manager |
Same name, different numbers
When two reports show different revenue, the cause is usually a different definition, not an error. Write definitions down and put them next to the dashboard so everyone reads the same number the same way.
Good and bad metrics
Vanity metrics look good but don't change decisions: total app downloads (what matters is how many people use it), Instagram followers (how many order?), page views without conversions. Prefer actionable metrics that tell you what to do and that you can influence.
Good metrics are also rates and ratios rather than raw totals. "We got 500 more orders" says little; "orders per day rose from 55 to 60" shows the trend. Ratios also let you compare a big shop with a small one.
Vanity versus actionable
| Vanity metric | Better actionable version |
|---|---|
| Total registered customers | Customers who ordered in the last 30 days |
| Total orders ever | Orders per week and the trend |
| Social media followers | Orders that came from social media links |
| Average rating alone | Share of orders rated 4 or more, plus the common complaints |
Leading and lagging, and guardrails
A lagging metric tells you the result after it happens, for example monthly revenue. A leading metric moves earlier and predicts it, for example the number of new customers this week or repeat visits per customer. You need both: the leading one lets you act in time, the lagging one confirms results.
A guardrail metric protects you from improving one number by damaging another. If you push discounts to raise orders, watch gross margin. If you rush staff to cut preparation time, watch the cancellation rate and ratings. Any target you set will be gamed or over-optimised unless you also watch what could break.
Choosing KPIs for a goal
From goal to KPIs
- 1Write the goal in one sentence, for example "Grow repeat business at the Indiranagar shop this quarter."
- 2Pick one main KPI that measures it: repeat rate (customers with 2 or more orders in the last 30 days / customers in the last 30 days).
- 3Add one or two leading metrics you can influence: share of new customers who order again within 14 days.
- 4Add a guardrail: gross margin, so discounts for loyalty don't wipe out profit.
- 5Set a target and a date, based on the current baseline: from 35% to 40% by 31 December.
- 6Write the definition card and decide how often the number is reviewed.
Always show the comparison
A number alone means little. Is ₹200 average order value good? Compare with last week, the same week last month or last year, a target, or another shop. Seasonality matters: Diwali week and a rainy Monday are not comparable.
Common mistakes
Watch out for
- Averages hide the shape: an average order of ₹200 could be everyone paying ₹200 or a few huge orders and many small ones. Look at the median or a distribution too.
- Comparing unequal periods: a 31-day month with a 28-day month, or a week with a holiday.
- Percentages without the base: a 50% rise from 2 orders to 3 is not news.
- Correlation is not cause: ice-cream sales and cold-brew sales rise together because of the heat.
- Too many KPIs: if everything is a KPI, nothing is.
How this comes up in interviews. Common questions: "What metrics would you track for [a food delivery app / a UPI payments app / an online course platform]?" and "The average order value dropped. Why?" A strong answer states the goal first, names one main metric with its formula, a few supporting and guardrail metrics, and says how you would segment the data (by item, time of day, customer type, location).
For a product like a UPI app you could choose successful transactions per active user as the main metric, with the transaction failure rate as a guardrail. Always explain why you chose it.
I work on [a product or business, e.g. a small coffee shop with pickup-only orders]. Our goal is [goal, e.g. increase repeat customers]. Suggest one main KPI, 2 leading metrics and 2 guardrail metrics. For each give: the exact formula, what data I need, and one way it could mislead me. Keep it to a table.What to expect: Check that every formula is precise (what counts, which period) and that guardrails really protect something the goal could damage. Ignore any suggestion you cannot measure with the data you have.
Try it: Design KPIs for a tuition centre
A tuition centre in Pune wants more students to finish their course. Write: (1) the main KPI with a precise formula, (2) one leading metric, (3) one guardrail, (4) one vanity metric they might be tempted to report, and (5) a definition card for the main KPI. Then calculate it: 240 students enrolled in the batch, 168 attended 80% or more of classes, and 156 completed the final test.
Key takeaways
- A KPI is a metric tied to a goal; choose few and watch them closely.
- Know the standard formulas: AOV, repeat rate, retention, churn, cancellation rate, gross margin, CAC and conversion rate.
- Write down a precise definition for every metric: what counts, the period, the source and the owner.
- Prefer actionable rates over vanity totals, and always show a comparison.
- Pair every target with a guardrail so improving one number does not break another.
- Beware averages without distribution, unequal periods, tiny bases and confusing correlation with cause.
Check yourself
Finishing the quiz marks the lesson as done.
1. A shop has 420 orders and revenue of ₹84,000. What is the average order value?
2. Which is an actionable metric rather than a vanity metric?
3. You run discounts to increase orders. Which guardrail metric protects you?
4. Of 150 customers active last month, 90 were active this month too. What is the churn rate?
5. Two dashboards show different revenue for the same week. What is the most likely cause?
Saved on this device. Create a free account to keep it on every device.
Practise this with Sam
A real mock interview, out loud, with an honest scorecard. Free to start.