What is Conversion rate?

A campaign delivers 12,400 clicks and 434 conversions, so the click-based conversion rate is 434 ÷ 12,400 = 3.5%. The site's analytics records 9,300 sessions from the same campaign, giving a session-based rate of 434 ÷ 9,300 = 4.7%. Both describe the same week. The gap between them is the traffic that clicked and never arrived, which is a page-speed and redirect problem rather than a conversion problem.

Conversion rate divides conversions by the traffic that produced them. The numerator is settled by the account's conversion definition. The denominator is where the arguments happen.

Conversion rate = conversions ÷ clicks is what an ad platform reports, using its own click count.

Conversions ÷ sessions is what analytics reports, using its own session count.

Conversions ÷ unique visitors is a third figure again, higher than both, because a person who visits three times and buys once counts as one visitor.

These are different numbers from the same week and all three are correct. A report quoting one without saying which is not information, and comparisons between accounts, agencies or case studies almost always compare denominators rather than performance.

The practical discipline is to choose one for the account, write it into the report definition, and stop renegotiating it.

The click-based and session-based rates differ because traffic is lost between the tap and the page.

People abandon slow pages. Redirects time out or drop parameters. Privacy settings and content blockers stop analytics scripts running. Impatient people tap twice. Bots are filtered differently by the two systems.

A shortfall of a fifth to a third is ordinary. What matters is whether it is stable: a gap that suddenly widens points at the destination — a deployment, a new redirect, a consent banner change, a page-speed regression — rather than at the campaign, and looking for it in the ad account wastes a day.

Cost per acquisition decomposes into exactly two terms.

Cost per acquisition = cost per click ÷ conversion rate

Halving the click price and doubling the conversion rate produce the same result. They are arithmetically equivalent and they are not equally available.

The click price is set in an auction with other bidders in it, most of whom are also trying to lower it. The conversion rate is decided by the page, the offer, the form, the checkout and the trust the business has built — all of which are entirely the advertiser's, none of which anybody is bidding against.

Most accounts have more headroom on the side they control and spend most of their attention on the side they do not. That imbalance is the single most reliable observation in performance advertising.

Message match. Somebody arriving from an ad should see, immediately and without scrolling, the thing the ad promised. A specific ad pointing at a general page loses the people the ad worked hardest on.

Speed. A slow page loses visitors before it has said anything, and it loses them disproportionately on mobile connections where most of the traffic is.

Fewer decisions. Every field, step, option and link is a chance to stop. The shortest path that still qualifies is almost always better than a longer path with better copy.

Qualification, deliberately placed. Removing friction raises the rate and lowers the quality. Sometimes that is right and sometimes the opposite is: a longer form that asks for a budget will lower the rate and raise the value of what arrives.

Obvious trust. Delivery terms, returns, contact details, evidence. Most of what stops a purchase is an unanswered question rather than an unmade argument.

The offer. The largest lever and the least tested, because testing it requires a decision from outside the marketing team.

A blended conversion rate is an average across populations that convert very differently.

Branded and non-branded search differ by an order of magnitude. Retargeting and prospecting differ. Mobile and desktop differ, sometimes because of the device and sometimes because of what people do on each. New and returning visitors differ. Geographies differ where delivery, language or availability differ.

Because the mix shifts on its own — a campaign broadens, delivery moves to a cheaper placement, a seasonal peak brings different traffic — the blended figure moves without anything changing. Checking the rate by segment before investigating the page is what separates a mix change from a regression, and the mix change is the more common of the two.

It says nothing about value. A page optimised toward the rate can discount its way to a better number and a worse business.

It rewards removing qualification. Fewer fields, more leads, lower quality, and the advertising report improves while the sales team's does not.

It is unstable at low volume. A page with forty visits and one conversion has a rate that moves on the next visit.

It inherits every fault in the conversion definition. A rate computed on the wrong event is a precise measurement of the wrong thing.

Page testing is where most conversion-rate work happens and where most of it goes wrong, for reasons that are statistical rather than creative.

Volume. A difference of a few conversions is noise. Tests on pages with modest traffic need to run far longer than teams expect, and stopping when the numbers look good is how an account accumulates a series of changes that did nothing.

One change at a time. A redesign that alters the headline, the form, the images and the layout produces a result that cannot be attributed to any of them, which means nothing was learned even if the number improved.

Whole sessions, not pages. A change that improves a landing page and worsens the checkout has not helped. The measure that matters is conversions per session for the whole path.

Seasonality and mix. Running a test across a promotional period, or across a week when a campaign broadened its targeting, measures the period rather than the change.

Conversion rate is the second term in the funnel and it is most useful read against its neighbours.

Click-through rate above it says whether the ad earned interest. Conversion rate says whether the destination repaid it. Average order value and margin below it say whether the conversion was worth having.

Strong click-through with weak conversion is a mismatch between what the ad promised and what the page delivers — the most common and most fixable pattern in advertising. Weak click-through with strong conversion usually means an under-delivering ad against an audience that would respond to better creative. Both strong with poor economics means the offer is priced wrong rather than marketed wrong.

• State the denominator before comparing anything. A click-based rate and a session-based rate from the same campaign differ by everything that was lost between the tap and the page load.
• Work the page before the bidding. Cost per acquisition is the click price divided by this rate, so the two halves are worth the same and only one of them is entirely yours.
• Match the page to the ad. A page that does not immediately show what the ad promised loses the people the ad worked on.
• Remove fields, steps and decisions rather than adding persuasion. The shortest path that still qualifies the lead almost always converts better than a longer one with better copy.
• Segment before concluding. A blended rate across devices, channels, new and returning visitors and branded and non-branded traffic describes an average that no individual segment produced.