Demand and conversion
The share of a defined population that completed a defined action — meaningless until both the population and the action are stated.
Also called CR · Конверсия
Formula
CR = Completions ÷ Eligible population
- Completions
- one named action, deduplicated by person rather than by event
- Eligible population
- everyone who could have completed it, over the same window
The denominator is where nearly all of the error lives. Sessions, users, visitors to the site, visitors to the page and people who reached the step are five different numbers, and the same business can report conversion rates that differ by a factor of four across them without anyone lying.
Conversion is a ratio between two decisions, and the second decision — what goes in the denominator — is the one that determines whether the number can be acted on. Rate per session answers a media question. Rate per user answers a demand question. Rate per person who reached the step answers a product question. Reporting one and discussing another is the most common analytical error in commercial work.
The window matters as much as the population. A user who visits on Monday and buys on Thursday counts as a conversion under a user-level definition with a thirty-day window, and as two non-converting sessions under a session-level one. Neither is wrong; mixing them within one report is.
- Sessions
- 48,000
- Unique users
- 31,000
- Users who viewed a product
- 12,400
- Users who began checkout
- 1,950
- Orders
- 620
Per session 1.3 %, per user 2.0 %, per product viewer 5.0 %, per checkout starter 31.8 %. Every one is correct. Only the last says anything about the checkout, and only the first two can be compared with last month if the traffic mix moved.
A conversion rate that falls after a successful awareness campaign has usually not got worse — the campaign added top-of-funnel traffic to the denominator. Segment by intent before concluding anything, and never compare a blended rate across periods where the traffic mix changed.
The definitions are the easy part. Whether the figure on your dashboard was computed this way is a different question, and usually the more expensive one.