Campaign measurement

Campaign KPI dictionary for cleaner readouts

Organization byline: Measurement Press editorial desk. This page does not list an individual author or reviewer. Read the editorial method and corrections policy.

About 9 min read

On this page

A campaign KPI only helps when everyone knows what decision it can support. The same number can be useful for delivery QA, weak for business impact, and misleading when it is promoted into a causal claim.

Use this dictionary before a campaign starts and again when the report arrives. It is written for marketers, publishers, agencies, and analysts who need a shared language for delivery quality, exposure quality, traffic quality, lead quality, brand movement, and incrementality.

Read the KPI by job

campaign KPI dashboard cards sorted by measurement job — A KPI has a job before it has a verdict. Sort the number by whether it checks delivery, attention, traffic quality, lead quality, brand movement, or incrementality.

Most reporting disputes come from asking one metric to do several jobs. Start by naming the job, then decide whether the metric is strong enough for the decision.

KPI family	What it can show	What it cannot prove alone
Delivery	Whether the campaign reached the planned inventory, device mix, geography, timing, and placement group.	That the campaign changed behavior or created demand.
Viewability	Whether an impression had a measurable chance to be seen under the chosen standard.	That the person noticed, understood, or acted because of the ad.
Attention	Whether format, environment, creative, and exposure conditions produced stronger attention signals.	That attention became incremental sales, leads, or brand movement.
Traffic quality	Whether visits from the campaign behaved like qualified reader or buyer sessions.	That visits were incremental rather than selected from people already interested.
Lead quality	Whether forms, demo starts, or inquiries fit the business definition of qualified demand.	That the media caused the lead or that volume will become pipeline.
Brand movement	Whether survey outcomes moved among the measured respondents and window.	That a survey shift caused profitable sales movement.
Incrementality	Whether a designed comparison estimates what changed because of exposure or treatment.	That the same effect will hold across all future contexts, audiences, and creative.

Dictionary

Served impressions

Count the ad responses delivered by the ad system. Useful for pacing and trafficking checks. Weak as an exposure measure unless measurability, viewability, invalid-traffic filtering, and placement context are also visible.

Measurable impressions

Count impressions where the system could determine whether the ad met the viewability standard. A low measurable rate can make viewability hard to interpret, because the measured subset may not represent all delivery.

Viewable impressions

Show impressions that met the selected viewability definition. Treat them as an exposure-quality gate, not as proof that the campaign affected memory, demand, or revenue.

Reach and frequency

Show how many distinct users, browsers, devices, households, or modeled people were reached and how often. The definition matters: device-level reach, household reach, and modeled person reach are not interchangeable. Use the reach and frequency checklist when deduplication, caps, overlap, or high-frequency tails decide the readout.

Attention or engaged exposure

Summarizes signals such as in-view time, screen share, interaction, scroll behavior, audible completion, or modeled attention. Useful for comparing creative and placement quality. It needs a separate outcome design before it becomes an impact claim.

Click-through rate

Shows the share of impressions that produced clicks. It can diagnose creative relevance, placement mismatch, accidental clicks, or offer strength. It should not be treated as a universal measure of campaign quality.

Engaged session

Captures post-click behavior such as low bounce, time on site, content depth, return visits, or key page views. It is stronger than raw clicks for traffic quality, but it still selects for people willing to click.

Lead volume

Counts form fills, demo requests, calls, downloads, or signups. Volume becomes more useful when paired with qualification rate, disqualification reason, sales follow-up, and duplicate suppression.

Qualified lead rate

Shows the share of leads that meet a defined standard. Define the standard before the campaign starts so the readout does not reward easy volume over useful demand.

Brand lift

Reports survey movement in outcomes such as awareness, recall, favorability, consideration, or intent. The readout should show sample base, exposed-control balance, field dates, confidence, and the exact question wording.

Incremental conversions

Estimate conversions caused by the campaign compared with a holdout, matched market, or other defined counterfactual. The estimate is only as strong as the assignment rule, leakage control, outcome window, and uncertainty statement.

ROAS, CPA, and pipeline

Useful for business planning when the numerator, denominator, cost basis, attribution rule, and sales-stage definition are clear. Weak when attributed revenue is treated as incremental revenue without a comparison.

Decision ladder

A clean readout climbs from delivery to impact. Skipping the lower rungs makes a positive-looking result harder to trust.

Decision stage	Useful primary KPI	Comparison needed	Careful readout language
Did the flight run correctly?	Qualified delivery by placement, context, device, geography, and date.	Planned delivery and excluded inventory.	The campaign delivered the intended context.
Was the exposure credible?	Measurable rate, viewable rate, frequency, attention, and format mix.	Prior flights, placement norms, or planned quality thresholds.	The campaign met or missed exposure-quality expectations.
Did readers engage?	Qualified visits, engaged sessions, return visits, or content depth.	Other contexts, creative variants, or baseline referral quality.	The campaign produced stronger observed engagement.
Did the audience fit the business?	Qualified lead rate, disqualification reason, sales-accepted lead share, or follow-up status.	Lead quality from comparable channels or prior flights.	The campaign produced demand signals of a stated quality.
Did the campaign change outcomes?	Incremental conversions, brand lift, search lift, or market lift with uncertainty.	A protected holdout, balanced control, or credible matched baseline.	The campaign produced measured lift for this design and window.

Metric promotion boundary

Use this handoff only when the reporting question has been stated first. A KPI can move up the ladder only when the next column is present; otherwise keep the readout at the lower, descriptive level.

Observed metric	Required handoff evidence	Permitted conclusion	Stop condition
Clicks or CTR	Impression denominator, placement/creative breakdown, and a defined traffic-quality check.	The ad generated observed response at the reported rate.	Do not call clicks demand, lift, or incremental conversions.
Engaged sessions or lead volume	Pre-declared engagement or qualification rule, duplicate handling, and a comparable baseline.	The campaign produced observed sessions or leads of the stated quality.	Do not call the change incremental when selection or attribution is uncontrolled.
Brand Lift result	Study eligibility, exact question/metric, response window, exposed-control comparison, and uncertainty.	The measured respondent outcome moved under the stated study design.	Do not generalize a survey result to sales or all audiences without supporting evidence.
Incremental conversions	Protected treatment/control assignment, leakage review, outcome window, and uncertainty statement.	The design estimates incremental conversions for this population and window.	Do not forecast future performance when assignment, maturity, or outcome definition changes.

Reproduction note: copy the metric name, numerator, denominator, comparison group, observation window, and decision sentence into the readout before approving a conclusion. If any field is blank, downgrade the wording to a descriptive observation and record the missing evidence.

Source trail and a reproducible readout

The definitions in this page are intentionally bounded by the reporting job. Google Ads defines click-through rate as clicks divided by impressions, while its Brand Lift guidance makes study eligibility, question wording, response collection, and the measured lift type part of the interpretation. Google’s Experiment Center separates campaign-setting experiments from lift studies, and its Conversion Lift documentation describes treatment/control comparisons for incremental conversions. Read the official references before translating a dashboard label into a business claim: CTR definition, Brand Lift setup and data, Experiment Center, and Conversion Lift.

Source boundary: the Google documentation defines reporting objects and controlled study designs; it does not certify a publisher's implementation or turn a dashboard KPI into causal evidence. Treat served impressions, clicks, and attributed conversions as descriptive until assignment, comparison, outcome maturity, and uncertainty are documented.

Simulated Worked Example: A fictional campaign reports 100,000 served impressions, 1,200 clicks, 180 engaged sessions, and 48 leads. The calculations are 1.20% CTR (1,200 ÷ 100,000), 15.00% engaged-session rate (180 ÷ 1,200), and 4.00% lead-per-click rate (48 ÷ 1,200). If 12 of the 48 leads pass the stated qualification rule, qualified-lead rate is 25.00% (12 ÷ 48). These are descriptive funnel measures, not proof that the ads caused the sessions or leads.

Evidence row	Calculation	Bounded decision
Delivery and click response	1,200 ÷ 100,000 = 1.20% CTR	Inspect creative, placement, and traffic quality; do not call the result incremental.
Post-click engagement	180 ÷ 1,200 = 15.00% engaged-session rate	Compare with a defined baseline or comparable placement mix before ranking traffic quality.
Lead quality	12 ÷ 48 = 25.00% qualified-lead rate	Use the qualification rule and disqualification reasons before promoting volume into pipeline language.
Controlled outcome example	Treatment 520 ÷ 10,000 = 5.20%; control 480 ÷ 10,000 = 4.80%; absolute lift = 0.40 percentage points; relative lift = 8.33%; incremental conversions = 40.	This supports a measured lift for the stated design and window only; report uncertainty, leakage, maturity, and whether assignment was protected.

Assumptions and limits: every count is fictional, the funnel denominators are shown explicitly, and the treatment/control rows assume equal-sized groups with a valid assignment rule. No confidence interval, leakage check, seasonality adjustment, or persistence estimate is supplied here, so the example cannot establish a real campaign effect or forecast future performance. Change the conclusion if the denominator, outcome window, qualification rule, assignment, or comparison changes.

Readout wording

dashboard metric cards translated into bounded readout language — A clean readout turns metrics into bounded sentences: what changed, what the metric can support, and what still needs a better comparison.

If the KPI shows	Stronger wording	Weaker wording
High viewability and no outcome comparison.	The campaign achieved strong exposure quality in the planned placements.	The campaign drove business impact.
High click-through rate on one creative.	This creative generated more observed clicks and should be tested for visit quality.	This creative is the highest-performing message.
More engaged sessions than a prior flight.	The campaign produced stronger observed traffic quality than the chosen baseline.	The campaign created incremental demand.
Lead volume rose but qualification fell.	The campaign increased lead volume while lowering average lead quality.	The campaign improved pipeline.
A lift test has leakage and wide uncertainty.	The result is directionally useful but needs cautious interpretation because the design was noisy.	The test conclusively proved the channel works.

Common failure modes

campaign KPI warning cards beside a cleaner decision card — Most KPI mistakes are promotion errors. A diagnostic number becomes a business-impact claim before the evidence design has caught up.

Reporting served impressions when the decision depends on viewable or attentive exposure.
Using clicks as the main signal when the campaign was bought for awareness or context.
Pooling unlike contexts, such as homepage discovery, source pages, and long-form case studies.
Counting every attributed lead without checking duplication, quality, and sales follow-up.
Calling a before-and-after improvement lift when no credible comparison was preserved.
Ranking placements without enough delivery, uncertainty, or creative rotation balance to compare them.

Pair with

Use this dictionary with the private marketplace campaign measurement checklist to plan the readout, the campaign reporting terms glossary to define delivery, traffic, lead, matchback, attribution, and incrementality terms, the outcome quality scorecard to check whether reported visits, leads, pipeline, or matched conversions are clean enough for the claim, the campaign data-layer spec to preserve fields before launch, the reach and frequency checklist to inspect delivery scale, caps, and overlap, the measurement method selector to choose the right evidence design, the incrementality test plan template when the decision needs a causal estimate, and the media kit for Measurement Press's placement IDs and campaign context map.

Keep reading

Choose the next guide

After naming the metric, move into report structure, outcome scoring, or method choice so the KPI stays tied to the decision it can support.

Report samplePlace KPIs in a readoutShow delivery, context, creative, traffic, lead quality, and allowed claim language in one report structure. Outcome laneScore the outcome qualitySeparate visits, leads, pipeline, matched outcomes, and lift before choosing renewal wording. Method choiceChoose the evidence designMatch descriptive reporting, baselines, lift tests, and model reads to the decision being made.