To calculate a website conversion rate, divide the number of users or sessions that completed a defined action by the corresponding total, then multiply by 100. The important choice is what you count: converting users, converting sessions, or individual events. These answer different questions.
A dashboard can show a higher percentage without the website generating a single additional enquiry. Changing the denominator is enough. Before using conversion rate to judge a campaign or redesign a page, write down exactly what the numerator and denominator represent.
Start with one outcome
Suppose you run a service business and want more demo requests. A useful conversion definition might be “a demo request successfully accepted by the website.” Clicking the submit button is an earlier action: it does not establish that the request succeeded.
Keep those two actions separate in your measurement plan. You may want to measure both to investigate the form, but naming both “demo conversion” would make the result difficult to interpret.
For the examples below, a conversion means a successfully recorded demo request. All figures are invented to explain the calculation; they are not industry benchmarks or Analyticse customer results.
Choose whether you are measuring users or sessions
A user-based rate answers: what share of the measured users completed this action during the reporting period?
User conversion rate = users with at least one qualifying conversion ÷ total measured users × 100.
A session-based rate answers: what share of the measured visits included this action?
Session conversion rate = sessions with at least one qualifying conversion ÷ total measured sessions × 100.
Google Analytics distinguishes user and session key event rates, including rates for a specific key event. This is a useful reminder to check whether a report measures your selected outcome or any configured key event. See the Google Analytics metric definitions.
Keep each numerator inside its denominator: converting users must belong to the user population being measured, and converting sessions must belong to the selected sessions. Use the same dates, time zone and eligibility rules for both sides.
One dataset, three different results
Imagine a reporting period with 100 measured users and 150 sessions. Ten users submit a demo request. Two of those users return and submit another request in a separate session. In this simplified example, there are 12 requests across 12 converting sessions, with no duplicate events.
| Measure | Calculation | Result |
|---|---|---|
| User conversion rate | 10 ÷ 100 × 100 | 10% |
| Session conversion rate | 12 ÷ 150 × 100 | 8% |
| Requests per user | 12 ÷ 100 | 0.12 |
Ten per cent describes the share of measured users who converted. Eight per cent describes the share of sessions that contained a conversion. The final value describes request frequency: 12 requests per 100 measured users. It does not mean that 12% of users converted.
Now move the two repeat requests into the users’ original converting sessions, leaving the total user, session and request counts unchanged. The user rate remains 10%, but only ten sessions now contain a conversion. The session rate becomes approximately 6.67%. Event totals alone cannot tell you how many sessions converted.
Check what your analytics tool means by “unique”
Tool definitions matter when comparing reports. Plausible, for example, defines goal conversion rate using unique conversions and unique visitors, subject to dashboard filters. It separately reports total conversions, including repeated completions. Its visitor definition also counts the same person separately across different days or devices. See Plausible’s metric definitions.
Consequently, do not rename a tool’s visitor rate “the percentage of people who eventually bought” without checking its measurement window and identity rules. A daily visitor count cannot, by itself, establish a person’s complete journey across several weeks.
When two tools disagree, first compare their definitions and selected populations. A percentage with the same label is not sufficient evidence that the underlying calculation is equivalent.
Keep the denominator aligned with the decision
For a site-wide demo goal, you might include every measured session. For a question about the demo form itself, you might instead include sessions that reached the form. Both can be useful, but they describe different stages.
Consider another hypothetical example: 20 of 1,000 site sessions contain a successful demo request, and 100 sessions reach the form. Assume those 20 successful sessions are all included in the 100 form sessions. The site-wide session rate is 2%; the form-session completion rate is 20%.
A report should name the eligible population explicitly. “Demo completion rate among sessions that reached the form” makes the second number understandable. “Conversion rate: 20%” hides the distinction.
For journeys that span several visits, define a separate follow-up window and cohort if your data supports them. Do not silently turn a same-session completion metric into an eventual-conversion claim.
Read the counts before reacting to a change
Always show the numerator and denominator alongside the percentage. Two conversions from 50 sessions and 200 conversions from 5,000 sessions both produce 4%, but the first result changes much more when one extra conversion arrives.
In the smaller example, one additional converting session among the same 50 sessions moves the rate from 4% to 6%. That is a rise of two percentage points, or 50% relative to the original rate. The arithmetic is correct; attributing that movement to a recent design change would require additional evidence.
Before proposing an intervention, check whether the outcome definition, tracking setup, reporting period or traffic composition changed. Use a rate movement to formulate a question you can investigate, rather than treating the percentage as an explanation of its own cause.
Write a short definition your team can reuse
For each conversion rate, record:
- Outcome: the successful action being counted.
- Unit: users or sessions, with repeat completions handled explicitly.
- Population: the eligible traffic and exclusions.
- Window: dates, time zone and any follow-up period.
- Source: the report and its documented metric definition.
- Decision: what a change would prompt you to investigate.
For our service-business example: “Weekly demo session conversion rate: sessions containing at least one successfully accepted demo request divided by all measured site sessions in the same week and time zone. Count each converting session once. Investigate changes alongside traffic source and form completion.”
Choose one important website outcome today and write this definition next to its dashboard figure. Then verify the numerator, denominator and repeat-counting rule against the underlying report. That gives your next conversion discussion a shared, checkable starting point.
A reusable way to choose the metric
The safest practical rule is to start from the decision, then choose the denominator. If you are deciding whether a page attracts the right visitors, a user-based view may be useful. If you are deciding whether a visit includes the right next step, a session-based view is usually clearer. If you are planning operational capacity, such as how many demo requests a team must handle, the event count may be the most useful number.
Formula: conversion rate = qualifying conversions ÷ eligible users or sessions × 100. The word “eligible” matters: the denominator must describe the population that had a real chance to complete the action.
This prevents a common reporting problem: a percentage looks precise, but no one can tell what action it supports. A conversion rate should always make the next discussion easier. If the number goes down, the team should know whether to inspect traffic quality, landing page clarity, form completion, tracking setup or repeat submissions.
Which denominator should you use?
Use the table below as a compact decision aid. It does not replace the tool documentation, but it helps your team avoid mixing different measurement questions.
| Question | Useful metric | What to show beside it |
|---|---|---|
| What share of measured people converted? | User conversion rate | Converted users, total users, date range |
| What share of visits included the action? | Session conversion rate | Converted sessions, total sessions, traffic segment |
| How many outcomes did the site generate? | Event or goal count | Total completions, duplicate rule, source |
| Where does the funnel lose people? | Step completion rate | Eligible step population and next step |
Common mistakes to avoid
| Mistake | Why it hurts the analysis | Better approach |
|---|---|---|
| Mixing users in the numerator with sessions in the denominator | The percentage no longer answers a clean question | Use the same unit on both sides |
| Calling every button click a conversion | Failed submissions can look like successful outcomes | Count the confirmed result and track clicks separately |
| Comparing two tools by label only | “Unique” and “session” can mean different things | Read the metric definitions before comparing |
| Hiding numerator and denominator | Small samples can look more stable than they are | Show counts beside the rate |
Before you report the number
Before a conversion rate goes into a weekly report, treat it like a small definition, not just a percentage. Name the action, the eligible traffic, the time window and the repeat rule. Then add one sentence that explains how the number should be used. This keeps the metric from becoming a loose score that different people interpret in different ways.
- Action: a successful demo request, signup, purchase, download or another confirmed outcome.
- Eligible traffic: all site sessions, only product-page sessions, only form sessions or another explicit group.
- Repeat rule: count each user once, each session once or every completed event.
- Comparison rule: compare equivalent periods and explain major traffic changes.
A useful report line might read: “Demo session conversion rate was 8% this week, based on 12 converting sessions out of 150 measured sessions. Organic traffic increased, paid traffic decreased and the form tracking rule did not change.” That sentence is longer than a dashboard label, but it prevents the most expensive misunderstanding: reacting to a rate without knowing what moved underneath it.