What is Click?

A click is one counted interaction with an ad. The trouble is that platforms count different sets of interactions under the word, and frequently report more than one set in the same table.

The widest definitions include reactions, comments, shares, saves, taps on the advertiser's page or profile name, expansions of an image or carousel, and interactions with elements of the ad that do not lead anywhere. The narrowest definitions count only interactions that sent somebody to a destination, and some platforms narrow further to outbound clicks that left the platform entirely.

Every metric derived from clicks inherits the ambiguity. Click-through rate, cost per click and the conversion rate computed from clicks all change substantially depending on which count was used, and nothing in the metric name says which.

The working rule is short: name the definition beside the number, and use the narrowest one available when comparing across platforms.

Between a counted click and a recorded visit on the destination there are several places where the two can diverge, and they normally do.

Abandonment. A person taps, waits, and leaves before the page renders. On slow pages and poor connections this is a substantial share.

Redirects. Each hop between the ad and the landing page is a chance to lose tracking parameters, time out, or drop the referrer.

Blocking and prevention. Privacy settings, content blockers and browser tracking prevention stop some analytics scripts from running at all.

Double clicks. One impatient person tapping twice produces two clicks and one session.

Bots. Filtered differently, and at different points, by the two systems.

A gap of a fifth to a third between platform clicks and analytics sessions is ordinary. The number itself is less informative than its stability: a gap that suddenly widens is a signal about the destination — a deploy, a redirect change, a page-speed regression — rather than about the campaign.

Not every click was intended. A mis-tap during a fast scroll, a tap on an ad positioned next to an interface control, a tap on a small target on a crowded screen: all produce counted clicks.

They are not evenly distributed. Some placements generate them at many times the rate of others, and the signature is consistent: a healthy click count paired with very short time on the destination, almost no scroll depth, and a bounce rate close to total.

Because the pattern is visible in destination analytics rather than in the ad platform, it is most often found by someone looking at the site rather than at the campaign. The remedy is a placement exclusion, and the effect on the campaign's headline click-through rate is to lower it — which looks like a regression and is an improvement.

Measuring creative interest. Holding the placement and the audience constant, a difference in clicks between two creatives is a real difference in how compelling they are. This is the metric's strongest use.

Filling retargeting pools. A click is evidence of enough interest to be worth addressing again.

Fast iteration. Clicks accumulate far faster than conversions, which makes them the practical signal for early creative testing in accounts where conversions are too sparse to compare quickly.

Diagnosing a funnel. Clicks with no conversions locate the problem after the click; impressions with no clicks locate it before.

Clicks say nothing about intent, quality or value. Optimising toward them instructs the bidder to find the people most likely to click, and the people most likely to click on an advertisement are not reliably the people most likely to buy anything. Campaign objectives set to traffic in accounts that actually want sales are one of the more expensive common mistakes, because they are not obviously wrong — traffic goes up, everything looks busy, and the conversion rate quietly collapses.

They are also easy to buy in quantity. Broad audiences, cheap placements, provocative headlines and misleading images all raise the click count, and each of them lowers what the clicks are worth.

Start by naming the definition. If a report says "clicks", find out which count produced it before doing anything else with the figure.

Then read it against what came next. Clicks beside sessions catches destination problems. Sessions beside engaged sessions or scroll depth catches accidental and low-intent traffic. Conversions beside clicks gives the conversion rate, which is where most of the diagnostic value is.

Then split by placement and by creative rather than reading the blended figure. Click behaviour varies more by placement than by almost anything else in an account, and the blended rate describes an average that no individual placement produced.

And treat a rising click count with flat outcomes as a warning rather than a result. It usually means the campaign found a cheaper and less interested audience, which is exactly what a click-optimised bidder is built to do.

Clicks are the middle term of the delivery chain, and each neighbour answers a question they cannot.

Above them sit impressions, which say how much delivery was bought. Clicks divided by impressions give the click-through rate, which is the cleanest available read on whether the creative earned attention from the people it reached — uncontaminated by the landing page, the offer or the checkout.

Below them sit conversions. Conversions divided by clicks give the conversion rate, which is a read on everything after the ad: the destination, the offer, the form, the price, the trust. A campaign with strong clicks and weak conversions has a problem the media cannot fix, and a campaign with weak clicks and strong conversions is usually under-delivering against an audience that would respond to better creative.

Beside them sit costs. Cost per click is the impression price divided by the click-through rate, which means a click price that moved was caused by one of those two and the pair says which.

Read as a chain rather than as a row of independent numbers, these figures localise a problem to a specific step in about a minute. Read individually, they support a great deal of confident activity aimed at the wrong stage.

Every major platform filters some proportion of clicks as invalid before reporting, and the filtering is neither perfect nor visible. Clicks removed after the fact may appear as a credit rather than as a correction to the original figure, which makes historical comparisons slightly unstable in ways that are rarely worth chasing.

The practical stance is to treat small discrepancies as noise, to reconcile against the destination rather than against another platform's count, and to escalate only when a pattern is large, sudden and localised to one placement or one geography — which is what genuine invalid traffic usually looks like.

• Name the click definition in every report. An all-clicks figure and an outbound figure from the same campaign differ by a third or more and both are called clicks.
• Reconcile clicks against sessions on the destination. The gap is real and its size tells you about page speed, redirects and tracking rather than about the campaign.
• Use the narrower definition when comparing across platforms, and say which one it is.
• Split by placement before concluding anything about creative. Placements differ enormously in accidental-interaction rate.
• Read clicks beside what happened after them. Cheap clicks that bounce are the easiest thing in advertising to buy and the least worth having.