Quick definition: Click-through rate (CTR) is the percentage of counted impressions, delivered messages, or eligible views that generate at least one defined click.
What is click-through rate?
CTR measures how frequently an audience responds to an opportunity to click. It is used for search and display advertising, email, push notifications, in-product banners, navigation, recommendation modules, and calls to action. The metric is simple only after its exposure rule is clear: a click can be counted only among impressions or recipients that had an opportunity to see the link.
The formula is CTR = clicks / counted impressions × 100. For email, teams may use clicks divided by delivered emails; for a search listing, clicks divided by served impressions; for a product module, unique clickers divided by users who rendered the module. All are reasonable measures when named accurately. They are not interchangeable, because a delivery, a render, an impression, and an active view are different denominators.
CTR is not conversion rate. A click is commonly an intermediate action, while conversion is a later outcome such as a purchase, verified signup, or successful task. It is also different from click rate, a label that is frequently used for email clicks per delivered recipient and can mean unique clickers rather than total clicks. Use an explicit metric name such as “unique CTA clickers per rendered eligible module” instead of relying on an abbreviation.
CTR formula and denominator choices
Choose one counting rule and persist it through a report or experiment:
| Metric | Numerator | Denominator |
|---|---|---|
| Ad CTR | Ad clicks | Served ad impressions |
| Unique module CTR | Unique users who click | Unique users with a rendered module |
| Email click-through rate | Unique recipients clicking a link | Delivered emails |
| Click-to-open rate | Unique email clickers | Reported opens |
Total clicks can exceed unique clickers when a person clicks repeatedly. The choice depends on the question: repeated clicks may signal engagement or confusion, while unique clickers describe reach. Do not mix the two across variants. Likewise, exclude impressions that were technically served but never had a realistic chance to render only if the rule is reliably measurable and applied symmetrically.
CTR can rise because the numerator improves, the denominator narrows, or both. An apparent improvement after moving a module lower on the page might arise because only more engaged users reach the location. Report impression counts, unique recipients, and visibility diagnostics with the rate.
CTR in A/B testing
CTR is a useful primary outcome when the direct objective is navigation, content discovery, or response to a clearly valuable prompt. It can also be a leading diagnostic metric for a broader conversion experiment. A new recommendation card may lift card CTR because it is more prominent, but the decision should account for downstream item views, purchases, satisfaction, and whether the placement displaces clicks from other valuable content.
Randomize before the opportunity to see the element and define the analysis population before launch. If treatment changes whether a module renders, analyzing only rendered modules can produce selection bias. An intent-to-treat measure among assigned eligible users answers the product decision; a rendered-impression CTR can be a secondary delivery diagnostic. Check sample allocation and exposure logging before interpreting the difference. Our SRM guide explains why unexpected allocation can invalidate a result.
CTR is prone to novelty and attention effects. A bright new button can win initial clicks but distract users or decay after familiarity. Pair it with guardrails such as task completion, conversion, dwell time interpreted carefully, error rate, complaint rate, and page performance. A click that sends customers to an irrelevant page is not customer value. For selecting such measures, see primary and guardrail metrics.
Worked CTR calculation
A media product tests a recommendation-module title. The primary metric is unique users clicking the module divided by unique users for whom the module rendered above the fold. In control, the module renders for 40,000 users and 2,800 click it. In treatment, it renders for 39,500 users and 3,160 click it.
control CTR = 2,800 / 40,000 = 7.0%treatment CTR = 3,160 / 39,500 = 8.0%absolute change = +1.0 percentage pointrelative lift = 1.0 / 7.0 = 14.3%
The team then finds that article starts increase only 0.2 points and subscriptions are unchanged. It checks whether treatment simply draws clicks away from the navigation menu, whether clicked articles load successfully, and whether the title overpromises. The evidence supports a stronger entry interaction, but not necessarily a large business outcome. A longer test may be needed to observe subscription behavior and avoid an early reading; see how to set test duration.
CTR data-quality caveats
An impression should be an actual measurable opportunity, not merely code inserted into a page. Ad blockers, lazy loading, hidden containers, scroll position, failed rendering, and client errors can make served, rendered, viewable, and seen exposures diverge. Record which definition applies. For client-side events, use a stable component and campaign identifier, and deduplicate click events caused by double taps, retries, or navigation callbacks.
Email CTR has special constraints. Tracking links can be altered by security scanners and automated privacy tools; some clicks are not human intent. Delivery failures, unsubscribes, and suppressions affect the denominator. Opens are particularly unreliable because privacy features can fetch pixels without a person reading the message, so click-to-open rate should never be treated as a clean engagement truth.
Segmenting after the result can mislead. A high CTR in a country, device, or placement might be noise among many explored cuts. Predefine important segments, inspect traffic composition, and use appropriate multiple-comparison discipline for claims about them. This article on multiple comparisons covers the risk.
Common CTR mistakes
- Calling every served request an impression: a hidden or failed module may not be an opportunity.
- Comparing total clicks with unique clickers: repeated behavior changes the numerator.
- Using CTR as proof of conversion: a click can be accidental, curious, or disappointed.
- Ignoring displaced behavior: one module’s gain may move clicks from another.
- Analyzing only people who rendered after treatment: rendering can itself be affected by the variant.
- Trusting email opens as ground truth: privacy tools distort them.
Frequently asked questions
What is a good CTR?
It depends on placement, intent, audience, channel, and counting method. Compare equivalent placements and focus on qualified downstream outcomes.
Is CTR the same as conversion rate?
No. CTR measures clicks after an exposure; conversion rate measures a specified success action among an eligible population.
Should CTR count multiple clicks from one person?
Use total clicks when repeated actions matter, or unique clickers when measuring the share who responded. State the rule and keep it constant.
Why did CTR increase but sales not change?
The variant may have attracted low-intent clicks, displaced other paths, or created expectations the destination did not meet.
Can CTR be above 100%?
Total-click CTR can exceed 100% if people can click more than once per counted exposure. Unique-clicker CTR cannot exceed 100% with consistent units.
Summary
CTR is clicks divided by a defined set of opportunities to click. Its usefulness depends on a truthful impression or delivery denominator, a clear distinction between unique and total clicks, and downstream measures of value. In experiments, use it to evaluate interaction while checking allocation, rendering, displacement, and later conversion.
Sources
- Google Ads Help: Clickthrough rate
- IAB: Measurement guidelines
- Nielsen Norman Group: Clickable elements