Display Advertising: Understand Inventory, RPM and Publisher Revenue

Know what earns revenue before adding inventory

Display advertising can turn useful audience attention into publisher revenue. The operating question is whether that revenue, after relevant costs and adjustments, justifies the inventory and the effect on the reader experience. Installing more ad positions does not answer it.

This guide teaches the measurement chain and a practical publisher model. You will distinguish opportunities, requests, impressions, measurable impressions and viewable impressions; calculate RPM with the right denominator; and reconcile revenue with costs. Account application, approval and detailed placement implementation belong to separate lessons.

If you have not chosen advertising as a plausible model, begin with Monetization Readiness and Revenue Model Selection. A weak audience proposition is not repaired by adding ad technology.

Separate the publisher, advertiser and intermediary roles

A publisher provides a place where advertising may appear. An advertiser seeks a business result from reaching an audience. Networks, exchanges and ad-management services can connect demand with inventory and provide reporting, delivery or payment functions. The exact responsibilities depend on the arrangement.

Do not assume an advertiser’s bid is the amount the publisher receives. Ask which costs have already been deducted from the revenue shown in a report. A familiar platform name does not remove the need to understand the contract and the reporting basis.

Map the path from a page visit to a recorded ad event

An eligible page may contain several potential ad positions. A position is an opportunity, not a guaranteed impression. Some positions may never request an ad, some requests may return no ad and some responses may not become counted impressions. Keep these stages distinct when diagnosing a shortfall.

Google’s AdSense impression definition counts a request when at least one ad has begun downloading to the user’s device. It also notes that invalid traffic can affect counts. This is a provider-specific definition: use the definition attached to the metric in your own report.

Official source: AdSense impressions.

Our diagnostic approach is to locate the first stage that changed. If pageviews remain stable but requests fall, investigate which positions request ads. If requests remain stable but returned ads fall, investigate coverage and eligibility. Do not change every setting at once.

Coverage and fill need an explicit denominator

AdSense defines coverage as requests that returned at least one ad divided by total requests, expressed as a percentage. It answers whether requests obtained an ad, not whether every ad was viewed.

Official source: AdSense coverage.

In Google Ad Manager, the documented Ad Exchange delivery rate uses ad impressions divided by ad requests. That is a different named metric from AdSense coverage. A dashboard label such as fill should always be accompanied by its provider, numerator and denominator.

Official source: Ad Manager report metrics.

In an invented example, 100,000 requests produce 85,000 responses containing an ad and 80,000 counted impressions. Response coverage is 85%; impressions per request are 80%. Both can be correct. The five-percentage-point difference is a stage difference to investigate, not automatically an arithmetic error.

A high percentage does not establish good economics. Filling more opportunities with low-value demand may raise delivery while reducing average yield. Conversely, a lower fill level can sometimes coexist with higher revenue. Compare total retained revenue and the reader experience rather than treating 100% as a universal target.

Measure visibility without confusing it with attention

AdSense Active View Measurable is the share of total impressions whose viewability could be measured. Active View Viewable is the share of measurable impressions that were viewable. The second percentage does not use all impressions as its denominator.

Official source: Active View Measurable.

Official source: Active View Viewable.

Google’s Active View CPM explanation describes the standard threshold of at least half the ad on screen for at least one second. Use format-specific current documentation for other inventory; this lesson is not a universal specification for video or every display format.

Official source: Active View CPM.

Suppose 80,000 impressions include 72,000 measurable impressions, of which 43,200 are viewable. Measurability is 90%, viewability among measurable impressions is 60%, and viewable impressions as a share of all impressions are 54%. Label each result. Calling all three viewability creates misleading comparisons.

Meeting a visibility threshold does not prove that a person noticed, remembered or acted on the ad. Use the metric to understand placement quality, then consider the wider reading experience. A placement that obscures content may raise exposure while weakening the publication.

Calculate RPM with the same population as the revenue

Google defines Page RPM as estimated earnings divided by pageviews, multiplied by 1,000. Impression RPM divides estimated earnings by impressions instead. They normalize the same kind of money over different units.

Official source: Page RPM.

Official source: Impression RPM.

A session metric needs its own definition. Do not substitute an analytics session total for a provider’s ad-session metric without understanding how each is counted. Similarly, total analytics pageviews may not equal the pageviews in an advertising report.

Use one revenue amount, currency, period and inventory scope. When combining segments, divide summed revenue by summed volume. Do not take an unweighted average of RPMs from segments with different traffic sizes. If the denominator is zero, report the ratio as unavailable rather than inventing a zero or infinite yield.

Distinguish advertiser pricing from publisher yield

CPM is a price expressed per thousand impressions under a defined arrangement. Effective CPM is a normalized result. CPC refers to cost per click; a revenue-share percentage refers to a portion of a defined monetary base. A label alone does not describe every fee, eligibility condition or payment rule.

AdSense’s current revenue-share documentation says Content publishers receive 80% after the advertiser platform’s fee. It gives approximately 68% of advertiser spend as an example when Google Ads buys the ads, and notes that a platform partner can introduce another revenue share. The page also describes publisher payment using effective CPM for advertiser bids.

Official source: AdSense revenue share.

The practical lesson is to identify the base. “80% after one fee” is not “80% of the original advertiser spend.” If your input already represents the publisher’s revenue after the applicable share, applying that share again double-counts a deduction. Record separately any additional fee that genuinely remains outside the report.

Do not compare two providers on percentage alone. Compare the retained amount for equivalent inventory, costs, payment feasibility and experience. A larger percentage of a smaller revenue base may produce less money.

Worked example: one coherent monthly inventory model

All numbers below are invented teaching assumptions, not AdSense averages, quoted rates or an earnings forecast. Assume 50,000 matching pageviews and 100,000 ad requests during one month. For this simplified example, each counted request produces no more than one counted display impression. There is no refresh inventory.

In this example the €4 impression RPM is already a publisher-revenue figure. Do not multiply €320 by coverage or viewability again: those stages are already reflected in the counted population and the observed yield used here. A separate model priced specifically per viewable impression would need a compatible rate and denominator.

The same €320 can therefore produce several different RPMs without contradiction. Reporting the largest number as the best performance would merely choose the smallest denominator. The business still has the same revenue before adjustments and costs.

Move from estimated revenue to retained contribution

Continue the invented example with €20 of revenue adjustments, leaving €300 of retained publisher revenue. Subtract €70 of incremental cash operating costs to obtain €230 before owner time. Value ten hours of content and ad-operations work at €15 per hour: contribution after owner time becomes €80, before tax and other shared overhead.

The retained page yield is €6 per thousand pageviews: €300 ÷ 50,000 × 1,000. At that constant yield, €220 of included cash and time costs would require about 36,667 pageviews to break even. This calculation holds costs and traffic mix fixed; it is not a forecast that costs remain fixed as the site grows.

At a lower retained yield of €4, the same 50,000 pageviews produce €200, leaving a €20 loss after the €220 cost base. If retained yield stays at €6 but traffic falls to 30,000 pageviews, revenue becomes €180 and the loss becomes €40. Write these downside cases before committing to recurring costs.

Reconcile earnings with actual payment records

Google explains that account activity reports are estimates rather than finalized earnings. Its earnings guidance allows adjustments for invalid activity and certain other causes, including advertiser nonpayment, and distinguishes reporting time zones from the billing time zone. A displayed balance is not necessarily the final payment amount.

Official source: Understanding earnings and adjustments.

For your worksheet, keep estimated revenue, finalized or adjusted revenue, unpaid balance and cash received in separate columns. Match the transaction period to the report period and explain timing differences. Do not describe a pending balance as money available to pay next month’s operating bills.

Confirm the applicable threshold, verification requirements, payment method, fees and timetable inside the chosen provider’s current terms. This lesson does not set those terms. Keep unresolved differences as reconciliation items and avoid allocating them to individual pages unless the evidence supports that allocation.

If the provider reports an invalid-activity adjustment, investigate the source and implementation. Do not treat a larger allowance for invalid activity as an acceptable substitute for correcting the problem.

Use weighted comparisons and comparable audience segments

An invented segment example: 10,000 pageviews at €12 Page RPM produce €120, while 40,000 pageviews at €3 produce another €120. Total revenue is €240 across 50,000 pageviews, so combined Page RPM is €4.80. The simple average of €12 and €3, or €7.50, is wrong for the combined population.

Compare similar device, country, page-type and traffic-source groups across a suitable period. An aggregate RPM can rise because traffic shifted toward a higher-yield segment even when no placement improved. Keep the counts beside the percentages so a small segment does not appear more certain than it is.

Avoid explaining every movement as a site change. Demand, audience composition and the comparison period can change too. Use a before-and-after comparison as a clue; a controlled test with a clearly defined population offers stronger evidence when feasible.

Maintain a metric dictionary naming each report, field, currency, time zone and denominator. This small record prevents recurring confusion when a second provider or analytics tool is added.

Diagnose the stage that changed before increasing ad density

Keep a dated change log for ad units, templates, consent configuration, traffic campaigns and provider changes. Investigate the most plausible cause with the smallest relevant check. A full rebuild rarely makes the first diagnosis easier.

If reader complaints, accidental interactions or layout problems appear, pause the affected placement and inspect it. A monetization system should preserve a useful route through the content and a clear distinction between content, navigation and advertising.

Run one measured inventory experiment

Begin with a baseline for one representative page type. Record the inventory, traffic population, revenue basis and relevant costs. State which metric you expect to improve and which reader-experience conditions must remain acceptable. A higher impression count is an intermediate result, not the final objective.

1. Define the hypothesis.For example, a less disruptive placement may preserve retained page revenue while reducing reading interruptions.
2. Set the population and limit.Choose the pages, devices, observation period and maximum acceptable downside.
3. Check the implementation.Confirm permitted configuration, correct loading and consistent event definitions without interacting with your own live ads.
4. Compare complete outcomes.Use retained revenue, costs and experience alongside requests, impressions and visibility.
5. Keep, revise or remove.Decide after sufficient comparable observations and the relevant adjustment cycle; document remaining uncertainty.

Do not assume automated refreshing, extra overlays or more requests are allowed or beneficial. Verify the chosen provider’s rules before changing delivery behavior. Avoid testing several major changes simultaneously unless the current setup must be stopped for an identified problem.

Build a monthly publisher worksheet

Check three identities before using the worksheet: estimated revenue agrees with impression RPM and impressions; the same revenue agrees with Page RPM and matching pageviews; retained revenue minus included costs agrees with contribution. Differences should have an explanation, such as rounding or a different scope, rather than being silently ignored.

If a field is unavailable, label it unavailable. Do not estimate a missing stage from an unrelated provider’s average and then present the completed chain as measured performance.

Use these exercises to test your understanding

Exercise 1 — denominator correction. A report shows €90 from 15,000 pageviews and 45,000 impressions. Page RPM is €6 and impression RPM is €2. Explain why both are correct and why adding them has no useful meaning.

Exercise 2 — visibility correction. Out of 10,000 impressions, 8,000 are measurable and 4,000 viewable. Measurability is 80%, Active View-style viewability is 50%, and the viewable share of all impressions is 40%. Name the denominator before each percentage.

Exercise 3 — cost decision. A placement change adds €35 of retained monthly revenue but creates €10 in cash costs and two additional hours of work valued at €15 per hour. Incremental contribution is −€5. Explain what further benefit would have to be evidenced before keeping the change.

Before moving to implementation, write your own metric dictionary and one downside model. If the required revenue depends on an unspecified RPM or uncounted work, return to the assumptions rather than treating the calculation as finished.

COMPLETE THE WORK

An advertising contribution worksheet

Use the guide’s example to separate traffic, measured advertising performance and retained revenue. Keep the observation window and denominator attached to each rate.

Fields to include

Period; page scope; traffic definition; reported earnings state; included costs; retained contribution.

Review before proceeding

Can the calculation be repeated from the recorded totals? Are estimated and finalized earnings visibly different?

Sources, scope and independence

Public-source review: September 11, 2026. Google documentation supports the named metric and AdSense payment definitions. Provider rules and reporting interfaces can change. The operating framework and all numerical scenarios are AffiliateBest’s educational analysis, not measured results, benchmarks or promised income.

• AdSense coverage
• AdSense impressions
• Page RPM
• Impression RPM
• Active View Viewable
• Active View Measurable
• Active View CPM
• AdSense revenue share
• Understanding earnings and adjustments
• Ad Manager report metrics

This lesson installs no advertising code and makes no claim of account acceptance or tested payouts. External links are official information sources, without personal referral parameters.