Google Ads: Why Expected CTR is the Most Important Metric for Google?

Why Should Google Even Predict CTR?

Why should Google even predict clicks? Because choosing the right ads for the query and the order in which they are displayed greatly affects the probability that a user will see and click on each ad. This ranking has a strong impact on the revenue the search engine receives from the ads. Further, showing the user an ad that they prefer to click on improves user satisfaction. For these reasons, it is important to be able to accurately estimate the click-through rate of ads in the system. For ads that have been displayed repeatedly, this is empirically measurable, but for new ads, other means must be used.?

56% of Google revenue in 2024 came from Search Ads, most of it using a pay-per-performance model with cost-per-click (CPC) billing. To maximize revenue and user satisfaction Google must maximize expectations that users will click. Google can make expected user behavior predictions on historical click-through performance of the ad, but the new ad has no historical information, therefore its expected CTR is unknown.

In search, the probability that a user clicks on an ad declines as much as 90% with display position. Thus, it is most beneficial for search engine to place best performing ads first.

Most search engines today order their ads based on expected revenue where expected revenue = Probability of the ad click * CPC of the ad. Thus, Google cares a lot about your expected CTR because it’s revenue depends on it. Btw an ad with true CTR of 5% must be shown 1000 times before we are even 85% confident that our estimate is within 1% of the true CTR.

Every day new players enter the market and existing ones create new campaigns and add new keywords. As a result, there is a large inventory of ads for which the search engine has no prior information. An incorrect ranking has strong effects on user and advertiser satisfaction as well as on the revenue for the search engine. Thus, for ads that are new, or have not been shown enough times, we must find a way to estimate the CTR through means other than historical observations.

Search Advertising Framework

Whenever an ad is displayed it has some chance of being viewed by the user. The farther down the page an ad is displayed, the less likely it is to be viewed. Probability that an ad is clicked on depends on two factors:

• Probability that it is viewed
• Probability that it is clicked on, given that it is viewed

So CTR is the probability the ad would be clicked if it was seen, with discounting applied we can estimate the probability an ad would be clicked at any position.

It is easy to predict CTR if an ad has been displayed a significant number of times. Whenever the ad was not clicked, it may have been seen with some probability (see figure below).

Thus, the number of views of an ad is the number of times it was clicked, plus the number of times it was estimated to have been seen but not clicked. The relative probability of an ad being seen at different positions can be experimentally measured by presenting the same ad at various positions on the page.

But how to predict CTR for the new ads?

Predicting the CTR of an Ad

Let’s start with what we have, each ad contains (not limited to):

• Landing page
• Bid term aka keyword
• Headline
• Description
• Display URL
• Clicks
• Views

Furthermore logistic regression is used to predict real-value (CTR of an ad). In simple words, Google predicts the contribution of every feature (see features above) towards probability of an ad being clicked. I’d like to focus your attention on the Term CTR, because there are other ads showing for the same Term we can benchmark new ads against this data. In Google Ads “Expected CTR” and “Ad Relevance” components are heavily based on the Term CTR in my opinion. Expected CTR measures AD component contribution towards CTR, in other words how well your ad is crafted. Ad Relevance measures how good your ad CTR is compared to other ads showing for the same Bed Term (keyword)

Estimating Ad Quality

In the previous section we estimated the CTR of an ad based only on the terms. Within every term there is significant variation in ad CTR. For example, maximum CTR for an ad for digital cameras is more than 3 times greater than the average, surgery - 5 times greater. So can we use features of an individual ad to predict CTR even better? Yes, we can. The work of Jansen and Resnick? suggests that Web searchers consider the summary, the title, and the URL of an advertisement in deciding whether to click it, what else??

• Appearance: How many words are in the title? In the body? Does the advertisement have good capitalization? Does it contain too many exclamation points, dollar signs, or other punctuation? Does it use short words or long words?
• Attention capture: Does the title contain action words such as “buy”, “join”, “subscribe”, etc.? Does the body? Does the ad provide numbers (such as specific discounts, prices, etc)?
• Reputation (social proof, business assets):? Does the display URL end with .com (similarly for .net, .org, .edu)? How long is it? How many segments are in the display URL (e.g., books.com is generally better than books.something.com)? Does it contain dashes or numbers? Because good, short, .com domain names (such as single word names) can be expensive, some of these features can be seen also as estimating how well-established and/or large the advertiser is.
• Landing page quality: How quickly it loads? Is it covered with ads? Is it relevant to the search query and ad? Does it have images?
• Relevance: Does the bid term appear in the title exactly? Do any subsets of the term appear in the title? In the body? What fraction of the body?

Why Should You Care?

Because expected CTR is important for Google revenue you should check your Quality Score and try to get it to average or above average rating using info provided to you in this article, otherwise you will compensate with your or your clients money.?