All of the major ad networks are using a mechanism called ‘quality score’ or ‘ad relevance score’.
Google was first by introducing the ‘AdWords Quality Score’. Other ad networks have followed using similar names, for example Bing Quality Score and Facebook Ad Relevance Score.
Ad networks often emphasize that quality scores are about user experience, i.e. ensuring that ads are relevant to their users.
But what are these metrics really about?
In this article we will explore why ad networks have economic reasons for using quality scores and how they are used to optimize click pricing and profitability.
How Ad Networks Make Money Selling Impressions and Clicks
In order to understand why quality scores exist we need to take a look at the monetization strategy of ad networks.
The overall ad inventory of a network consists of the total number of available ad impressions. Impressions (not clicks) are the most basic unit of ad inventory.
In the early days of online advertising the most common bidding model was to charge advertisers for ad impressions (CPM, cost per thousand impressions). Today, the most popular model is to charge advertisers per click (CPC, cost per click).
Moving from CPM pricing to CPC pricing has significant implications for ad networks.
Under a CPM model an ad networks revenue is predictable. Networks deliver impressions and charge for impressions. Advertisers bear the risk of paying for impressions without receiving any clicks.
Under the CPC model, the situation is reversed. Networks deliver impressions and only get paid for clicks on an ad. They face the risk of delivering a thousand impressions without generating any clicks (i.e. revenues). This means ad networks suddenly need to be concerned with the click through rate of an ad (i.e. how many clicks it generates for every one thousand impressions).
Here’s an example of a Google Adwords search ad:
Let’s assume there are 1,000 searches (potential ad impressions) a day for the keyword “car repair new york”. For simplicity reasons we’ll assume that there’s only one available ad slot:
- Under the CPM model, Google would sell these 1,000 ad impressions for a fixed price to the advertiser. Let’s assume the highest bidder pays $20 for a thousand impressions.
- Under the CPC model, Google still monetizes the same 1,000 ad impressions. However, advertiser bids are now based on how much they are willing to pay for each click. Advertiser A might bid $1 per click, advertiser B $2 per click and advertiser C $3 per click. The highest bidder isn’t necessarily the most profitable option for Google anymore. Here’s why…
Under the CPC model, Google’s earnings for the same one thousand impressions depend on cost per click bid and click through rate:
- If advertiser A’s ad gets a click through rate of 4% (= 40 clicks on a thousand impressions), Google would earn $40 (40 clicks x $1 cost per click).
- If advertiser B’s ad generates a click through rate of 2% ( = 20 clicks per thousand impressions), Google would also earn $40 (20 clicks x $2 cost per click).
- If advertiser C’s ad gets a click through rate of 1% (= 10 clicks on a thousand impressions), Google would earn $30 (10 clicks x $3 cost per click).
Even though Advertiser A and B are willing to pay less per click they could easily be a more profitable choice for Google if the click through rate on their ads is high enough.
So how do ad networks decide which ads to run and how to they manage their risk of running low-performing advertisements?
The Economic Argument for Quality Scores
As discussed above, the CPC bidding model introduces an element of risk into an ad network’s monetization strategy. The risk arises from the fact that networks deliver ad impressions but they can only charge advertisers for clicks.
The CPC model requires ad networks to manage two main risks:
- The risk of running ads that don’t generate any clicks. Why waste impressions on ads that don’t generate revenues?
- The risk of running ads with a (comparatively) low click through rate. Why run a low performing ad if there’s another one that could generate more revenue (clicks) per thousand impressions?
That’s where quality scores come in.
In short, quality scores are a mechanism that optimizes click pricing by taking into account the value of impressions.
Two advertisers might each take up one thousand ad impressions. Advertiser A might generate 10 clicks, advertiser B 100 clicks. Both advertisers have taken up the same amount of ad inventory, one thousand impressions. In an ideal scenario, the ad network would be able to charge both advertisers the same fees. In other words, advertiser A’s cost per click would be ten times higher than advertiser B’s.
Here’s how it works in simple terms:
- Quality scores first quantify the ‘relevance’ of an ad, typically with a score from 1 to 10 (the higher the more ‘relevant’)
- Based on the score they adjust click prices. Low quality scores increase the cost per click, high quality score decrease the cost per click.
Once a new ad has received some exposure, the network assigns a quality score to the ad or also keyword (on AdWords). The click through rate is typically the most important determinant in calculating quality scores across all networks.
In other words, the click through rate of an ad is one of the most important factors in determining ‘relevance’.
Ads with a low click through rate automatically pay a higher cost per click. Ads with a high click through rate are rewarded with a lower cost per click.
There are other factors that influence quality scores on AdWords, Facebook and other networks. But it shouldn’t be a big surprise that the click through rate, which is directly linked to the bottom line of each ad network, has more weight than other factors.
From an economic perspective, ad relevance scores are a great instrument for ad networks to manage their risk and optimize profitability. In the absence of quality scores, an advertiser could receive the same one thousand impressions for less money just because his ad generates a low click through rate.
Ultimately, ad networks need to maximize the earnings per thousand impressions. If two advertisers receive the same one thousand impressions. Advertiser A’s ad only generates 10 clicks while advertiser B’s ad generated 100 clicks. Advertiser A will have to pay a much higher cost per click to account for the fact that his ad is using up the same amount of ad inventory but generating much fewer clicks.
If your ad doesn’t generate enough revenues (i.e. clicks), your quality score goes down and your cost per click increases. This means your ad will stop running unless your cost per click bid meets the minimum required level.
A low performing ad is typically an indication that factors like ad copy and targeting settings haven’t been setup effectively. In the end it’s the advertisers responsibility to learn, experiment and use a networks capabilities to the full extent. Marketer’s who fail to do so, pay inflated click prices.
The Advertiser’s Perspective on Quality Scores
Up until now we have looked at ad relevance scores from the perspective of ad networks. By now it should be clear that ‘relevance’ is not only about user experience but also one of the most important factors in monetizing ad impressions effectively.
Does that mean advertisers, just like ad networks, should be strive for maximum click through rates? The short answer is no.
An advertiser’s goal is to maximize the return on investment of an ad campaign. This typically means generating as many sales or leads as possible from a given ad budget. Advertisers key performance metrics are typically be cost per sale or cost per lead.
When it comes to ad quality scores, you often hear the following advice:
Optimize quality scores (by increasing click through rates) in order to pay a lower cost per click. The reasoning behind this is that more clicks from the same ad budget should equal more sales from the same ad budget. This is not necessarily the case.
Advertisers need to know their ‘ideal prospect’ and craft ads that maximize the response rate of these prospects. That is not the same as maximizing the overall click through rate of the ad.
I will go into more detail on how to manage and optimize quality scores in a future blog post.
If you have any questions, please post them in the comments.