Page RPM vs impression RPM: why they differ and which one to track
Key takeaways
- Page RPM is estimated earnings divided by page views, multiplied by 1,000; impression RPM is estimated earnings divided by impressions, multiplied by 1,000.
- Page RPM always equals impression RPM multiplied by the number of impressions per page view.
- AdSense counts an impression for each ad request where at least one ad has begun to download to the user's device.
- Adding ad units can lower impression RPM while raising page RPM, so neither metric should be read alone.
- Page RPM is the better measure of what your traffic earns; impression RPM is the better measure of how each ad slot is priced.

Page RPM tells you how much you earn per 1,000 page views; impression RPM tells you how much you earn per 1,000 ad impressions. Both use the same estimated earnings, only the denominator changes. Track page RPM as your headline metric, because it captures both ad density and ad pricing, and use impression RPM to diagnose why page RPM moved.
How does Google define each metric?
The two help center glossary entries give the formulas:
| Metric | Official formula | Source |
|---|---|---|
| Page RPM | (Estimated earnings / Number of page views) x 1000 | Page RPM |
| Impression RPM | (Estimated earnings / Impressions) x 1000 | Impression RPM |
| Ad RPM | (Estimated earnings / Ad impressions) x 1000 | Ad RPM |
Google’s help page on Impressions explains what the impression denominator counts: an impression is counted “for each ad request where at least one ad has begun to download to the user’s device.” It is the number of ad units (for content ads) or search queries (for search ads) that loaded ads. AdSense uses this “downloaded” definition for all impressions; whether the ad was actually seen is measured separately by Active View (see Active View viewability).
Served vs downloaded impressions
The same help page explains why the impression count is defined this way. Industry bodies, the Interactive Advertising Bureau (IAB) and the Media Rating Council (MRC), periodically update the standards for counting impressions, and the web standard has moved over time:
| Definition | When an impression is counted | Used by AdSense? |
|---|---|---|
| Served impression | When the ad server sends the ad; the creative may never download, render or be seen | No |
| Downloaded impression | Only after the ad has begun to download to the user’s device | Yes, for all impressions |
| Viewable impression | When the ad meets viewability criteria | Measured separately as Active View |
Google also notes that impression counts can fluctuate because of invalid traffic. For impression RPM, this means the denominator only contains ads that at least started loading. Requests that returned no ad, and slots that never loaded, are invisible to impression RPM but still reduce page RPM if the page view counted.
Ad RPM is a third, less common variant. It divides by individual ads rather than ad units, and Google notes it is only available under the “Advanced” metric family tab in reports.
Why do page RPM and impression RPM differ?
Because one page view can produce zero, one or many impressions. The link between them is a single identity:
Page RPM = Impression RPM * (Impressions / Page views)
The middle term, impressions per page view, is effectively your ad load as AdSense measured it. Anything that changes it moves page RPM without touching impression RPM, and vice versa.
| Factor | Effect on impressions per page view | Effect on impression RPM |
|---|---|---|
| Adding ad units or enabling more Auto ads formats | Up | Often down (extra slots tend to be less prominent) |
| Unfilled ad requests | Down (no ad began to download, no impression) | Unaffected directly |
| Lazy-loaded units below the fold that visitors never reach | Down | Can go up (the units that do load are the visible ones) |
| Short pages or bounce-heavy traffic | Down | Unaffected directly |
| Stronger advertiser demand (season, country mix) | Unchanged | Up |
| More restrictive blocking controls | Can go down if fewer ads are eligible | Can go down (less auction competition) |
A worked example: adding an ad unit
All figures are hypothetical, round numbers for illustration, not benchmarks.
A page template gets 100,000 page views a month. You test adding one more in-content unit.
| Before | After | |
|---|---|---|
| Page views | 100,000 | 100,000 |
| Impressions | 200,000 | 300,000 |
| Impressions per page view | 2.0 | 3.0 |
| Estimated earnings | $300 | $360 |
| Impression RPM | $1.50 | $1.20 |
| Page RPM | $3.00 | $3.60 |
Impression RPM fell by 20%, which looks like a loss if it is the only metric you watch. But page RPM rose by 20% and total earnings rose by $60. Each impression earned less on average, yet the page as a whole earned more.
The reverse case is just as common. If removing a poorly performing unit raises impression RPM from $1.50 to $1.80 but drops impressions per page view from 2.0 to 1.5, page RPM falls from $3.00 to $2.70. The cleaner page is worth that only if it brings other benefits (user experience, engagement, Core Web Vitals) that you value.
To test changes like these under controlled conditions, use AdSense experiments rather than before/after comparisons, which are distorted by seasonality.
Which one should you track?
| Situation | Track | Why |
|---|---|---|
| Monthly health check of the site | Page RPM and total earnings | Shows what your audience is worth, whatever the ad setup |
| Comparing two sites or two sections | Page RPM | Normalises for traffic size and ad density |
| Comparing ad units or formats | Impression RPM | Isolates pricing per slot |
| Revenue dropped but traffic did not | Both, plus impressions per page view | Tells you whether pricing or ad load changed |
| Comparing countries or devices | Impression RPM first, then page RPM | Demand differs by market; ad load can differ by device layout |
Diagnosing a drop in three checks
- Did page views change? If page views fell and page RPM is stable, the drop is a traffic problem, not a monetisation problem.
- Did impressions per page view change? Divide impressions by page views for both periods. A fall points to fewer slots loading: a code change, a template change, a consent or ad blocking effect, or more unfilled requests. The ad requests and coverage metrics help here; the API defines coverage as the ratio of requested ad units or queries to the number returned to the site.
- Did impression RPM change? If ad load is stable but impression RPM fell, look at pricing factors: season, country mix, device mix, blocking changes. The guides on seasonality and RPM by country and device cover these mechanisms.
Other RPM variants you may see
The AdSense Management API metric reference lists more denominators:
- Ad request RPM: earnings per 1,000 ad requests. An ad request may result in zero, one or multiple individual ad impressions.
- Matched ad request RPM: earnings per 1,000 requests that returned at least one ad.
- Individual ad impression RPM: earnings per 1,000 individual ads shown.
- Funnel RPM: earnings per 1,000 funnel impressions (non-ad units such as related search that can trigger further ad requests).
Ad request RPM is useful when fill is the issue: it falls when requests go unfilled, while impression RPM ignores unfilled requests completely.
Putting the whole chain together gives a useful decomposition, all from report metrics:
Page RPM = (Ad requests / Page views)
* (Impressions / Ad requests)
* Impression RPM
The first term is how many slots each page asks to fill, the second is how often a request actually loads an ad, and the third is price. When page RPM moves, one of the three moved. If you want to see what an unfilled slot looks like on the page, see unfilled ad slots.
Common mistakes
- Optimising impression RPM in isolation. It rewards removing ads, which can lower total revenue.
- Optimising page RPM in isolation. It rewards adding ads, which can hurt user experience and, with intrusive placements, run into ad placement policies.
- Comparing today with last month. Recent estimated earnings can change, and season alone moves impression RPM.
- Forgetting that page views and impressions are counted differently. A page view with no ad that began to download produces no impression, so pages without ads lower impressions per page view.
- Quoting RPM without the denominator. Always say “page RPM” or “impression RPM”.
For the general definitions of RPM, CPM and eCPM, see RPM vs CPM vs eCPM explained.
Frequently asked questions
Which is more important, page RPM or impression RPM?
For judging what your traffic earns, page RPM is more useful because it combines how many ads each page shows with how much each ad earns. Impression RPM is better for diagnosing pricing: whether a specific ad unit, format, country or device is getting weaker bids. Most publishers track page RPM as the headline number and use impression RPM to explain its movements.
Can page RPM be lower than impression RPM?
Yes, when impressions are fewer than page views. That happens when many page views produce no impression at all, for example pages without ad code, ad units that went unfilled, or lazy-loaded units the visitor never scrolled to. Since page RPM equals impression RPM times impressions per page view, a ratio below 1 makes page RPM the smaller number.
Is impression RPM the same as ad RPM?
They use different denominators. Impression RPM divides estimated earnings by impressions (ad units or search queries that showed ads). Ad RPM divides by individual ad impressions, which can be larger because one ad unit can show more than one ad. Google notes that ad RPM is only available under the Advanced metric family tab in reports.
Why did my impression RPM drop after I enabled more ads?
More ad slots usually means each additional slot is less visible or less valuable, so the average earnings per impression can fall. What matters is whether total earnings and page RPM rose. If page RPM went up while impression RPM went down, the extra ads added revenue; if both fell, the change probably hurt.
Sources
- Page RPM · Google AdSense Help
- Impression RPM · Google AdSense Help
- Impressions · Google AdSense Help
- Ad RPM · Google AdSense Help
- Metric (AdSense Management API reference) · Google for Developers
Related guides
- RPM vs CPM vs eCPM: what each AdSense metric means and how to calculate it
RPM, CPM and eCPM explained with the official AdSense formulas, worked examples and the revenue share link between what advertisers pay and what you earn.
- Active View viewability in AdSense: definition, measurement and the effect of placement
What Active View viewable and measurable mean in AdSense, the 50% for one second rule, how the percentages are calculated and how placement changes them.
- AdSense reports: dimensions, metrics, saved reports, CSV exports and finding what changed
How AdSense reports work: breakdowns and metrics with official definitions, saved reports, CSV exports via the Management API, and a method to find what changed.
- AdSense experiments: how they run, what you can test, and how to read the results
How AdSense experiments compare an original and a variation, what you can test, how Auto optimize works, and how to read results without fooling yourself.