What is a conversion rate: the formula, and what you divide by what

What is a conversion rate? It is the share of a period's sessions that ended in the expected action: conversions divided by sessions, multiplied by 100. Three choices are enough to define it: the action counted, the base you divide by, the observation window.

This page is for the executive or marketing director who has to state the formula, name its calculation base and understand why two tools plugged into the same site rarely display the same value. It also settles the vocabulary: “conversion rate” and “close rate” name the same indicator.

What a conversion rate measures — and what it does not measure

A conversion is what you want the visitor to your website to do: the action linked to revenue. It changes with the business model:

• In e-commerce: most of the time it is a sale, and the conversion is easy to define.
• For a software vendor: the target action is a quote request.
• For a professional publisher: a sign-up serves as the conversion.

The conversion rate relates these actions to sessions, over one and the same period. The numerator counts actions, the denominator counts visits: you are not dividing two quantities of the same kind, which is why the choice of base changes the result. Each of the three elements is decided:

• The numerator: the action counted, for example an order or a qualified contact request.
• The denominator: sessions, i.e. the “session conversion rate” in GA4.
• The period: the observation window, shared by both terms of the ratio.

The canonical formula is therefore: conversion rate = conversions ÷ sessions × 100. If your site records 2,000 sessions and 100 sales during the month of March, your conversion rate is (100 ÷ 2,000) × 100, i.e. 5%.

The qualifier “over one and the same period” is not a turn of phrase: a rate published without its measurement window is a running total whose boundaries nobody knows.

However unavoidable an indicator of performance the conversion rate may be, it counts events without weighing them. Three measurements are therefore tracked alongside it:

• The average basket: an excellent rate on customers who buy almost nothing, or only the lowest-margin product, does not guarantee the durability of the business.
• The quality of what has been converted: a form filled in by a student looking for documentation and a form filled in by a mandated buyer each count as one conversion.
• The share of workable leads: where the visible conversion is a lead, a rate that improves while that share falls describes a deterioration that the figure announces as progress.

Knowing what the rate is a ratio of does not yet say whether it is satisfactory. A good conversion rate is recognised by three checks:

• It is profitable: it makes the acquisition cost compatible with the margin per conversion.
• It improves: it beats your own history, calculated on a constant definition.
• It rests on sufficient volume: a handful of conversions does not allow any conclusion.

Judging it then means setting out five criteria, from the denominator to the page type. The first of them is decided here.

The denominator: conversions divided by what?

Visitors, visits, unique people: these are not the same objects. It explains most disagreements between two dashboards plugged into the same site, in the same month. Most often the numerator is identical on both sides and it is the denominator that differs: nobody got it wrong, the two ratios simply do not divide by the same thing.

Sessions, users, clicks, leads: four bases, four rates

Four bases compete for that place, and each answers a different question. Confusing them makes any comparison impossible, including the most useful of all, from one month to the next:

These four choices produce four distinct measurements, and the gap between them is structural. Sessions are the base retained; a rate published on another base is quoted with its denominator and is not compared with yours.

Whichever base is used, three prerequisites determine how reliable the calculation is; without them, the choice of denominator remains a theoretical debate:

• A correctly tagged action: an event or a goal depending on the tool.
• A deduplication rule: the same business event must not be counted twice.
• A volume large enough to interpret: a “high” rate on 30 sessions remains fragile.

Set the base once, and hold to it

Document your definition and keep it stable so as to build reliable comparisons: it is the simplest rule on the subject, and the one most rarely written down. Three decisions are recorded together, and dated:

• The unit in the denominator: sessions.
• The actions in the numerator: those linked to revenue, and only those.
• The set of pages: the whole site, or a named family of pages.

A base that changes mid-year makes the history unreadable. A change of tracking or definition therefore opens a new series, and the same caution applies to values coming from elsewhere:

How much history to build up and how to read it are set out in the reference article on a good rate.

A base fixed once and for all is then used to place yourself: measuring how your rate moves over time and the impact of your actions. Breaking it down by channel or page type, and choosing how often to read it, belong to the conversion rate KPIs. The definition says what the rate is made of; the segmentation says where to look at it.

Conversion rate, close rate: one indicator, two words

The two expressions are interchangeable and name the same ratio between actions completed and sessions. Only usage tells them apart: “conversion” dominates in e-commerce and first names a target action, “close” comes up in emailing and with the sales director, for the moment when the prospect becomes a customer. Same indicator, two usages: the article on the site conversion rate details how to read it. When two teams diverge, settle the measurement rather than the word.

What counts as a conversion, and what that changes in the calculation

The numerator counts a single target action per rate, the one that genuinely commits the customer: adding up orders, quote requests and account creations gives a number that matches no decision. What can count as a conversion, the place of micro conversions, tracked separately and outside the rate, and the choice of the action itself are covered in the conversion definition.

Reading your rate when part of the journey no longer goes through a click

Part of the journey no longer goes through a click on your site: a prospect can get their answer in an AI interface, then come back later directly. The ratio you publish sees only part of it, and nothing in standard measurement links that direct visit back to what preceded it.

A share of searches now ends without a click, the answer being delivered in the engine's interface. An observed citation is not, for all that, a measured exposure or an attributed conversion: it can be noted, it does not enter your rate.

Direct conversions, assisted conversions, the contribution of generative engines

These three terms often come up together, and that is precisely what must not be done with them: adding them up. The same conversion appears both as direct and as assisted, and the contribution of generative engines is an estimate, not a count. Each measures something different:

Hence a publication rule: three columns, three definitions, never a total. A report that presents only direct conversions underestimates the pages high in the journey; a report that adds them together manufactures a figure.

What the disappearance of the click does to the denominator

If your denominator is the number of sessions, a share of searches that no longer reaches the site removes sessions from the calculation. With the number of conversions unchanged, the rate rises without any page having changed and without a single extra sale having been recorded.

That kind of rise is read, not celebrated. The check comes down to three readings taken together:

• the number of conversions, in absolute terms: it tells you whether activity has really grown;
• the number of sessions: it tells you whether the traffic reaching the site has contracted;
• the rate itself, which can only be interpreted with the other two.

Read together, these three values distinguish the situations that the rate alone confuses:

Continue reading

With the definition set, what comes next depends on your situation:

• You steer a complete site: the headline indicator and what to track alongside it are detailed in the site conversion rate.
• You sell online: with the order as the unit, the calculation base and the breaking points are discussed differently in the e-commerce website conversion rate.
• Your base is fixed and your rate does not suit you: choosing which page to start with and measuring the effect of a fix belong to optimizing conversion rate.
• You have to say in a steering committee what the organic channel brings in: attributing a conversion to SEO and to generative engines is covered in SEO conversion.

To present direct conversions, assisted conversions and the contribution of generative engines side by side once the calculation base is fixed, the performance reporting module renders them in a single dashboard, each with its own definition.

Frequently asked questions (FAQ)

What is a conversion rate?

The conversion rate is the share of a period's sessions that ended in the target action, such as a purchase, a qualified quote request or a sign-up. It is calculated by dividing conversions by sessions, then multiplying by 100. It can only be read with the action counted and the period named.

How is the conversion rate calculated?

Divide the number of conversions by the number of sessions for the period, then multiply by 100: 120 sales for 5,000 sessions give 2.4%. Count only the action linked to revenue, an order or a qualified contact request; micro conversions are tracked separately. Any change to this definition opens a new series.

Why does the conversion rate matter?

The conversion rate is the only indicator that links traffic to the result. Moving from 1% to 2% doubles the number of conversions, and revenue with it at constant traffic and average basket; if either of the two falls, the gain shrinks accordingly. It only informs usefully provided it is tracked on a stable base.

What is the difference between the conversion rate and the close rate?

There is no difference in measurement: the two expressions name the same indicator, calculated in the same way. Only usage changes: “conversion” dominates in e-commerce, “close” comes up in emailing and in sales vocabulary. When two teams diverge, fix the numerator, the denominator and the window rather than the word.

What is a good conversion rate?

A good conversion rate is a profitable rate that improves: it makes the acquisition cost compatible with the margin per conversion, and it beats your history calculated on a constant definition. Published benchmarks place it, on an identical denominator, without setting a threshold. To judge it, set out five criteria: denominator, channel, sales cycle, offer and page type.

Which factors make the conversion rate vary?

On the content side, the rate depends on the fit between the page's promise and the search intent, the quality of the traffic brought in, the clarity of the offer and the legibility of the action expected. Speed, usability and the ordering journey belong to the product team, UX and development. A change of calculation base also makes the rate vary without any of these factors having moved.

How is the conversion rate measured in GA4?

In GA4, a conversion corresponds to an event marked as a key event, for example generate_lead, purchase or sign_up. Use the session conversion rate, which divides conversions by sessions, and check this base before taking the value elsewhere. Make sure as well that the tagging is reliable and deduplicated: otherwise, the numerator is wrong before the division even happens.