DM6: Analysing behaviour: metrics, funnels, cohorts and attribution
Metrics that matter
🎯You'll be able to
Define conversion rate, average order value, revenue per visitor, customer acquisition cost and lifetime value, and tell useful metrics from vanity metrics.
🏠The everyday picture
A tea stall owner who counts only the people passing by learns little. Counting how many stop, how many buy, and how many come back tomorrow tells them what to change.
📘What it really is
-
Conversion rate: the share of visitors (or sessions) who complete a goal. Say which: per visitor and per session give different numbers.
-
Average order value (AOV): revenue ÷ orders.
- Revenue per visitor (RPV): revenue ÷ visitors. It combines conversion and order value in one number.
- Customer acquisition cost (CAC): marketing spend for a channel ÷ new customers it brought. Include agency fees and discounts.
- Lifetime value (LTV): the profit (not just revenue) a customer brings over time. Early estimates are guesses; label them.
-
A guide metric (sometimes called a north star) sums up the value customers get, such as "orders delivered on time per week". Pick one and support it with a few others.
-
Vanity metrics go up without telling you what to do: page views, followers, app downloads without activity.
- Counts and their denominators: "694 buyers" means little without "of 22,600 visitors".
📝A worked example
Input: MittiMart's cleaned test data (experiment/): 22,600 visitors, 694 who ordered, 735 orders, and ₹8,53,071 of orders.
Steps:
- Conversion rate per visitor: 694 ÷ 22,600 = 3.07%.
- AOV: ₹8,53,071 ÷ 735 = ₹1,160.64.
- RPV: ₹8,53,071 ÷ 22,600 = ₹37.75.
Expected output: three metrics, each with its definition. Note that 41 visitors ordered twice, which is why orders (735) are more than buyers (694). (All practice-kit data.)
🌍Where it's used
- Shop managers watch conversion, AOV and RPV weekly.
- Performance marketers judge channels by CAC against expected LTV.
- Founders report a guide metric to investors with its supporting metrics.
📝Case scenario
A meal-kit start-up celebrated 1 lakh app downloads. Only 4,000 people ordered, and most didn't order twice. Switching its main metric to "customers with a second order within 30 days" refocused the team on delivery quality and menus.
🖱️Try it
Using the figures above, work out what RPV would be if the number of orders stayed at 735 and AOV rose by ₹100. Success: about ₹41.00 (735 × ₹1,260.64 ÷ 22,600), with the steps shown.
⚠️Common mistakes
- Counting orders as converters. Sign: 735 "conversions" from 694 people. Fix: say whether you count people or orders.
- CAC without all costs. Sign: ad spend only. Fix: include fees, discounts and free shipping.
- Reporting LTV as fact. Sign: "Each customer is worth ₹12,000." Fix: show the assumptions and label it an estimate.
🛠️For your project
Project 2 counts conversion per visitor (test 5: "visitors with at least one order, not orders").
✅Quick check
What's revenue per visitor?
Show the answer
Total revenue divided by visitors; it combines conversion rate and order value.
Why are app downloads often a vanity metric?
Show the answer
They can rise without anyone using the app or paying.