Ad Impressions vs Reach, Clicks, Views, and Conversions

An impression records an opportunity for an ad to appear. Reach estimates how many distinct people, devices, or households were exposed. Clicks record interactions, views apply platform-specific video rules, and conversions record a chosen business action. None of these numbers proves the others. A reliable campaign report keeps their definitions and denominators separate.

The short answer: what does each ad metric measure?#

The first reporting mistake is treating this sequence as a funnel in which every event automatically becomes the next. It does not. Each step is observed by a different system, at a different time, under a different counting rule.

What is an ad impression?#

An ad impression is a counted ad-delivery event. The exact trigger depends on the channel and measurement method. Google Ads, for example, says it counts an impression each time an ad is shown on Google or the Google Network. Video measurement may count closer to the point when playback begins. A publisher might also expose an earlier server-side delivery count.

This is why the word "impression" needs a qualifier. Ask whether the report means:

• an ad request;
• an ad response;
• a served impression;
• a render or playback-start event;
• a measurable impression;
• a viewable impression; or
• an impression remaining after invalid-traffic filtration.

Two reports can both contain a column called "impressions" while counting at different points. The labels match, but the events do not.

Google's impression definition is useful for its own platform. For an independent comparison, obtain the event specification used by every party in the chain.

Can an impression count if nobody saw the ad?#

Yes. A served impression can count even when the ad was below the fold, covered by another element, rendered in a background tab, or abandoned before a person had a realistic chance to notice it. An impression usually measures delivery or display, not human attention.

A viewable impression applies an additional rule about the portion of the creative in view and the time it remained there. The Media Rating Council's viewable impression guidelines define the relevant measurement framework. Viewability is still a technical qualification. It does not prove that the person looked at the ad, understood it, or remembered it.

For campaign reporting, show at least three separate counts when the channel supports them:

1. total measured impressions;
2. impressions eligible for viewability measurement; and
3. impressions that met the viewability rule.

Do not divide viewable impressions by all impressions unless the methodology explicitly calls for that denominator. A viewability rate built on measurable impressions answers a different question from a rate built on total served impressions.

What is the difference between impressions and reach?#

Impressions count exposures. Reach counts distinct identities exposed at least once during a defined period.

If one household receives 12 impressions, the campaign has 12 impressions and household reach of one. If those impressions appear on a television, two phones, and a laptop, device-level reach might be four while household reach remains one. Person-level reach could be one, two, or more depending on who actually used or watched those devices.

That example shows why every reach figure needs four disclosures:

• Identity unit: person, signed-in account, device, browser, or household.
• Deduplication boundary: within one platform, across publishers, or across all measured channels.
• Observation window: day, week, campaign lifetime, or another period.
• Method: deterministic identifiers, probabilistic matching, modeled estimates, or a blend.

Meta describes reach as the number of people who saw content and notes that the metric is estimated. The MRC's Audience Reach Measurement Guidelines also emphasize definitions, filtration, and disclosure around unique measurement.

How do you calculate frequency?

Average frequency is:

frequency = impressions / reach

Suppose a campaign records 1,000,000 impressions and an estimated reach of 250,000 households. Average household frequency is four.

That average can hide a poor distribution. Half the households might have received one impression while a small group received many more. A useful report adds frequency buckets such as 1, 2, 3, 4 to 6, 7 to 10, and more than 10 exposures. The distribution tells a media buyer whether more budget is adding new reach or repeatedly hitting the same identities.

For a deeper treatment, see cross-device frequency capping and household identity in streaming.

What is the difference between impressions and clicks?#

An impression records exposure under the platform's counting rule. A click records interaction with a clickable ad element. A click can be recorded even when the destination page fails to load. Google's click definition says this explicitly.

Click-through rate connects the two:

CTR = clicks / impressions x 100

If an ad receives 800 clicks from 100,000 impressions, CTR is 0.8 percent. That says the ad generated eight recorded clicks for every 1,000 counted impressions. It does not tell you whether those clicks became valid website sessions or valuable customers.

The denominator also matters. A CTR calculated from served impressions is not interchangeable with one calculated from viewable impressions. Label the rate if your report offers both.

Why can clicks be higher than sessions?

Clicks and sessions occur in different systems. A person may click twice during one analytics session. They may close the page before analytics loads. A redirect may remove the campaign identifier. Consent or browser settings may prevent analytics collection. Invalid-click filtering can also be applied on a different timetable.

If that is the problem you are investigating, use the step-by-step guide to Google Ads clicks versus GA4 sessions.

What is the difference between an impression and a video view?#

A video impression normally occurs earlier than a video view. The impression indicates that the video ad entered a qualifying delivery or playback state. A view applies an additional platform-specific rule, such as watched duration or interaction.

There is no safe universal definition of "view." Google's YouTube ad metrics documentation distinguishes impressions, TrueView views, viewability, engagements, and quartile playback metrics. Other publishers use their own rules.

When comparing video partners, request these fields separately:

• playback start;
• 25 percent complete;
• 50 percent complete;
• 75 percent complete;
• 100 percent complete;
• mute or unmute state, where supported;
• measurable and viewable status;
• skip or interaction events; and
• the rule used to count a billed view.

A high completion rate can be genuine, but it can also reflect autoplay, short creative, non-skippable inventory, server-side beacon behavior, or invalid activity. Interpret it with source, device, duration, and fraud signals.

What is a conversion?#

A conversion is a business action selected for measurement, such as a purchase, lead form, app install, qualified call, or subscription. The event becomes an advertising conversion only after an attribution system assigns some credit to an exposure or interaction.

That distinction matters:

• Observed outcome: the purchase or lead occurred.
• Matched outcome: the system linked it to an eligible ad event.
• Attributed outcome: the system applied a rule or model and assigned credit.
• Incremental outcome: an experiment estimates that the ad caused an outcome that would not otherwise have happened.

Platform-attributed conversions are not the same as causal lift. Attribution windows, view-through rules, identity matching, consent, and cross-device models can all change the number. Use attribution for operational optimization and a suitable experiment when the decision requires a causal claim.

Which metric should you use for each campaign goal?#

The primary metric should match the business decision. A campaign bought for reach should not be declared successful because it had a high CTR. A direct-response campaign should not be defended with impressions when qualified outcomes are missing.

Why do the same metrics differ across ad platforms?#

Different systems observe different parts of the journey. Common sources of divergence include:

1. Event timing. One system counts at the ad server, another at render, and another after a landing page loads.
2. Identity. One report deduplicates signed-in accounts, another devices, and another modeled households.
3. Time zones and cutoffs. A report labeled August 1 may cover different hours.
4. Late data. Fraud filtration, conversion matching, and modeled results can revise prior days.
5. Attribution. Click-through and view-through windows allocate outcomes differently.
6. Consent and blocking. Browser or app controls can prevent one observer from receiving the event.
7. Invalid traffic. Systems use different detection inputs, thresholds, and adjustment schedules.
8. Rounding and privacy thresholds. Small segments can be suppressed, modeled, or rounded.

Do not force the totals to match by changing definitions after the campaign starts. Create a measurement contract before launch, then reconcile each transition with evidence.

A measurement contract you can use before launch#

For every reported metric, record:

• the event name and plain-language definition;
• where the event fires;
• whether it is client-side, server-side, or both;
• the identity unit and deduplication rules;
• the time zone and reporting delay;
• the invalid-traffic filter applied;
• the numerator and denominator for every rate;
• the attribution window and credit rule;
• the source of record for billing; and
• the evidence retained for disputes.

This one-page contract prevents a large share of post-campaign arguments. It also makes discrepancies useful. A gap can then be located between two defined events instead of dismissed as "the platforms count differently."

Where independent measurement fits#

Platform reporting is best for decisions inside that platform because it has the richest auction and optimization context. Site analytics is best for observed website behavior. CRM or commerce systems own accepted leads, orders, cancellations, and revenue. Independent measurement adds a separate observation of delivery quality across publishers and channels.

AdProof is designed for that verification layer. It collects supported delivery signals across web, mobile, video, and VAST environments, normalizes them into a consistent event model, and adds geo, device, source, fraud, reach, and frequency analysis. It does not make every platform definition identical. It gives the measurement team a consistent second record and the event-level evidence needed to explain where numbers diverge.

See AdProof's measurement and verification capabilities or read why platform and independent measurement differ.

Frequently asked questions#

Are impressions more important than clicks?

Neither is inherently more important. Impressions measure counted exposure; clicks measure interaction. Use impressions and valid reach for awareness decisions, and use valid sessions or qualified outcomes for response campaigns. Pair the metric with the campaign objective.

Can reach be higher than impressions?

Not under a consistent definition and time window, because every reached identity needs at least one qualifying impression. If reach exceeds impressions, check whether the reports use different filters, dates, channels, modeled estimates, or identity scopes.

Are views the same as viewable impressions?

No. A video view follows a platform-specific playback or engagement rule. A viewable impression follows a visibility and duration measurement rule. A single event may qualify for both, one, or neither.

Does a conversion prove the ad worked?

A conversion report proves that an observed outcome was attributed under the system's rules. It does not, by itself, prove that advertising caused the outcome. Causal questions require an appropriate incrementality or lift design.

What is the best single advertising metric?

There is no universal best metric. Choose the metric closest to the decision you need to make, then keep the upstream quality checks that make that metric trustworthy. For example, cost per qualified lead still needs valid delivery, functioning landing-page measurement, and a stable definition of a qualified lead.

← Back to all field notes