Overview

Definitions

Revenue

The total revenue of a website is generally a good metric, but in many cases, it may not be accurate. For example, it can be influenced by traffic fluctuations. Let’s consider the case of a publisher who decides to add an ad unit to a web page but, after doing so, the site suffers a substantial drop in traffic, perhaps seasonal or anyway unrelated to the ad unit. This traffic variation could make the total revenue falls, even though the new ad unit could have potentially brought additional earnings. In such a case, looking at the total revenue is not the best way to measure the success of a strategy.

Auctions

An auction is an attempt to sell a placement and view an ad. During an auction, ad calls (bid requests) to Ad-Networks through various channels are issued and the highest bidder wins the auction and displays an ad in the placement, which is an Impression. If no Ad-Network is interested the auction results in an unfilled placement. An auction can bundle ad calls of multiple placements.

Page Views

As the name suggests, this metric counts the number of pages viewed by a user. Counting page views is different from counting sessions because a user could visit multiple pages within one session. And it’s also different from counting users because the same user could view many pages. In fact, if the same user opens two different pages or even refreshes the same page, two page views will be counted.

Unlike Google Analytics page views, page views are only counted if at least one ad is shown.

User Sessions

On its support page, Google defines a session as “a group of user interactions with your website that take place within a given time frame. For example, a single session can contain multiple page views, events, social interactions, and ecommerce transactions”. A session lasts until there are 30 minutes of inactivity.

Here’s how a session works: when a user enters a website, a session activates. If within 30 minutes the user doesn’t do anything, the session expires, otherwise every time the user interacts with some website element, session tracking adds 30 minutes to the expiration time of the current session.

The system will consider all the user interactions that occur within this time frame as a single session. On the other hand, if a user starts interacting with the website, stays inactive for more than 30 minutes, and then performs a new action, the system will count as two sessions. It’s also important to notice that this metric doesn’t measure the number of actions taken on the site, such as the number of page views; all the actions taken before the session expires are counted as just one session.

Unlike Google Analytics user sessions, user sessions are only counted if at least one ad is shown.

Impressions

Impressions are equivalent to Served Impressions. The standard for calculating an impression is measured when the advertisement or link is fetched and loaded on a web page. The metric associated with accounting for served impressions is easy to calculate, as the data is highly trackable by the server— did it load, or did it not.

The difficulty of accounting for impression value using this method is that it does not include the actual display of the ad, or define the impact of the impression on the viewer. The analysis does not incorporate whether the ad was in a viewable location or actually seen by the consumer, it merely registers that the ad was fetched and the load was initiated.

Served impressions may include a page that the viewer closed before the ad completed loading, or perhaps the viewer scrolled past the ad before it fully loaded. Other impressions that are included in the views, yet have no marketing value, could be a page with ad-blocking software enabled, a page opened by a robot (proxy server, crawler, spider, or fraudulent click bot), a website with mobile incompatibilities, or broken plug-ins that prevent the ad from being displayed.

Unfilled Impressions

Unfilled impressions or blank impressions happen when no demand partner is interested in the impressions and send a no-bid response, due to geo, size, viewability, floor prices, or similar. In that case, the placement is either shown blank or backfilled with house ads.

Placement Page Views

The number of page views when a placement was loaded. In contrast to page views, which count each page view where at least one placement is shown, this counts the page views where a placement is shown. This metric can be used to see how many page views an ad was used.

The Placement Page View counter is typically smaller than the Page View counter. An exception to this rule is when grouping placements by metadata, for example by position, in this case, the sum of the individual placements might be bigger than the Page View counter.

Page RPM (RPM)

RPM is an acronym for Revenue Per Mille (mille means “thousand” in Latin) which is basically “revenue per 1000 page views” of a web page. The metric provides a comprehensive view of how the ad setup of a site is performing.

Unlike CPM, which calculates the rate/price per ad unit, RPM is aggregating all ads on the page and calculates a rate. For example, If you have 4 ads on a page your RPM will be higher than if you just have 2. RPM doesn’t represent how much you have actually earned; rather, it’s calculated by dividing your estimated earnings by the number of page views, impressions, or queries you received, then multiplying by 1000.

RPM = (Revenue / Page Views) * 1000

Session RPM (RPS)

When a page loads in your browser, the system counts that as a page view, while a session is the period of time a reader is active on your site. RPS will always be higher than RPM, even though overall earnings are the same. In the ad tech industry, it’s pretty standard that RPM = page view RPM, but some ad providers use session RPM (in other words, RPS) so it’s important to know what each provider means when referencing RPM.

Publishers should pay attention to RPS because it provides a comprehensive view of how their site is performing. RPS looks into critical factors impacting site monetization like ad viewability, fill rate, refresh rate, ad density, page speed, etc.

Differently from the others, it shows exclusively the effects of the monetization and user experience strategy, limiting the influence of other external factors. Additionally, it offers a broad vision of the publisher earnings, beyond the economic value of the single ad units.

Measuring your website Session RPM is the right way to understand if there’s room for improvement in your monetization strategy, or if an action you have taken is bearing fruit or it needs to be adjusted. And it’s also important to measure Session RPM separately for the Desktop and Mobile versions of the site, as layouts and interaction patterns tend to be very different depending on the device used.

RPS = (Revenue / User Sessions) * 1000

CPM

CPM is an acronym for Cost Per Mille (mille means “thousand” in Latin) and is the ad unit rate per 1000 ad impressions. It is the price advertisers pay to have their ads shown 1,000 times. However, high CPMs do not always correspond to overall success for the publisher. For example, a brand that pays a high CPM but only fills 50% of the available inventory may result in fewer earnings than another brand that pays lower CPMs but has a higher fill rate.

CPM = (Revenue / Impressions) * 1000

Fill Rate

The ratio between Impressions and Unfilled Impressions. The ad fill rate is one of the simplest ways to tell whether or not an ad request has been a success. While aiming for a 100% ad fill rate seems like a goal there are reasons why this is not achievable and even not desirable. Every advertiser is particular about getting their ads across to their target audience which means that not all ad calls will receive placements.

By using floor prices to achieve a higher CPM and ensure high-quality ads a 100% fill rate is not the goal. On the contrary, a 100% fill rate may sometimes translate to low-quality ads.

Fill Rate = Impressions / (Impressions + Unfilled Impressions)