## Product Interviews . . . Estimated revenue from Google ads

This article 4135 words Estimated reading 11 minutes

In our product interview series, we systematically discussed how to answer the estimates of product interviews.

In the past few days, a small partner sent a private message to the background of an estimate question, which is as follows:

**How do you estimate the revenue generated by GoogleAdSense?**

The known conditions are as follows:

The New York Times has about 10 million subscribers to Google's AdSense website, with 1 million in the 20th and 100.subscribers.

Google's AdSense revenue is about $17 billion.

Let's try to explore how to calculate the problem of this estimate.

//////////

**What is the interviewer looking at?**

The interviewer looks at your abilities:

Are you organized and logical in explaining your estimation methods?

Will you make your intentions understood when you make calculations?

Can you calculate quickly?

Can you explain your hypothesis convincingly?

**The key to answering estimation questions is to separate the "method" from the calculation.**

Separating methods from calculations prevents you from losing direction and logic in your calculations. Tree maps can help you clear your mind, and using them allows you to first explain the logic behind the method, and then focus only on calculations.

**How to answer a question**

In the articleProduct Interview Series - How to Answer Estimation Questions, we give a framework for answering such questions, which can be summarized as follows:

**Clarify the scope of the problem**Reach a consensus on the extent of the estimate through a conversation with the interviewer.

**Create the main equation**Introduce a master equation to explain your overall approach.

**State your assumptions****Draw a tree chart to break down the unknowns in the formula into simpler formulas**Use the results you want to obtain and the time range it covers (for example, one year) as the parent node. Break supres the parent node into child nodes (A, B, C) calculated using arithmetic operators.Explain the logic behind arithmetic operations.

For example, why divide AXB by C? Explain all assumptions using familiar scenarios that common sense or interviewers may agree with.

If you need to decompose the child nodes into other child nodes, repeat this step.

**Calculated from bottom to top**Starting at the bottom of the tree, calculates the results of each node branch until it reaches the top of the parent node (the result to get).

**Conduct a reasonable review**Check that your estimate is within the standard range by comparing it to a known baseline. If your estimate is not within the standard, check your hypothesis to determine which one is wrong, and then explain to the interviewer what might lead to miscalculation. As long as you have an explanation, your answer will be well received.

**Answers to questionsExample**

**Clarify the scope of the problem**

**Interviewer **: First, I want to clarify my understanding of how GoogleAdSense generates revenue.

**Interviewer **:Good.

**Interviewer **: Google AdSense is a Google advertising solution for publishers and is only available on third-party websites, right?

**Interviewer **: Yes. It does not include products such as Google's YouTube.

**Interviewer** : Ok, Google AdSense works by connecting advertisers and website owners through a web platform, serving both parties in a tie-up, and making your own profits.

**Interviewer **: Yes.

**Interviewer** : OK. I know that publishers get a portion of advertising revenue, so should estimates exclude this share?

**Pick interviewer **: No. Estimate the entire AdSense revenue.

**Interviewer** : Yes, thank you. Should I estimate my income for one year?

**Interviewer** : Yes.

**Interviewer** : Can you give me two minutes, let me consider how to solve this problem?

**Interviewer** : OK.

**Create the main equation**

↑ Main equation:FrameThe main formula for advertising revenue for one year

Note:

CPC (CostPerClick): Every time you click on a paid ad, the station's webmaster gets revenue when the user clicks on a CPC ad on a website

CTR (Click-Through-Rate): Click-through rate, a commonterm used in Internet advertising, refers to the click-to-reach rate of an online ad (image ad / text ad / keyword ad/ ranked ad, etc.), i.e. the actual number of clicks for that ad (strictly speaking, it can be the number of target pages divided by the number of ads shown (Showcontent).

CostPerMille is the cost per thousand people. CPM is a display paid ad that advertisers pay for as long as they show the content of the advertiser's ad.

**Interviewer** :

Well, on the whole, that's how I estimate it. Many publishers use GoogleAdSense to sell ad space on their sites and earn annual revenue from those ads.

Therefore, I will estimate total ad revenue for a year based on the average daily ad revenue X for each publisher in "Number of Publishers X."

Before explaining the details of how to estimate the number of publishers and daily income, I would like to confirm that my approach so far is clear?

**Interviewer **: So far, yes.

**State your assumptions**

**Interviewer** : OK. I know the New York Times has about 10 million subscribers, and I guess most of them are daily readers.

The 100th publisher has about 10,000 subscribers, assuming they read it once a month.

So, considering the bottom of the top 100 for minority publishers, I assume that these 100 publishers represent nearly 100 percent of Google AdSense revenue.

I say this because I don't count the number of personal bloggers who use Google AdSense. Do you agree with this assessment?

**Interviewer** : It seems reasonable, please continue.

**Interviewer **: OK. A number of large websites attract millions of subscribers every day, while niche sites attract thousands of monthly subscribers.

So, I think we can apply the 20/80 rule here, where 20% of large sites can generate 80% adSense revenue. Do you think it's reasonable?

**Interviewer** : Yes. So how will you use this information?

**Draw a tree chart to break down the unknowns in the formula into simpler formulas**

**Interviewer** : OK, because I'm using the 20/80 rule, I estimate that the top 20 percent of publishers (the root part) generate 80 percent of GoogleAdSense revenue.

Therefore, I will "represent the top 20% of the top 100 publishers" X "Average revenue per day" X "365 days".

Next, I'll break down daily ad revenue into click-through and display revenue, two of the main ad types offered by Google AdSense.

Then, in order to calculate the ad revenue for a day's click-through ads, I will "Number of visitors per day" X "Number of visitorclicks" X "Percentage of ads clicked by users", i.e. "Click/Click Rate (CTR) " X "CPC" (third level of branch).

To calculate the number of ads that visitors click on in a day, I will calculate "Number of pages visited in a day" X "Number of ads per page" X "Percentage of clicks on ads".

To calculate the number of click ads visitors see in a day, I will calculate "Number of pages visited in a day" X "Number of ads per page X "Percentage of click ads."

- Complete estimated tree map

So far, do you feel anything wrong?

**Interviewer** : No, you go on.

**Interviewer** : OK, next I'll explain how to calculate the revenue from daily display ads.

To calculate a day's display advertising revenue, I will "daily visits" X "The number of display ads they see per day" X "Cost per thousand impressions".

To calculate the number of display ads visitors see in a day, similar to the method of counting click ads, I will "The number of pages a person visits per day" X "Number of ads per page" X "percentage of display ads".

Do you feel anything wrong?

**Interviewer** No, I know it very well. Let's move on to the overall estimate.

**Interviewer** : OK.

**Calculated from bottom to top**

Good. Starting with the lower left corner of the estimation tree, I'll first estimate the number of pages that readers visit each day. I will use myself as the ordinary reader of that estimate.

Since I haven't read The New York Times, but like reading a page every day, I browse about 10 unrelated pages a day. Therefore, I will use 10 as the number of pages visited per day.

Interviewers use typical scenarios that might be familiar with to explain these assumptions.

Next, the interviewer writes down the estimate in the upper left corner of the node. Repeat this process to calculate each node.

Now, I think you see at least 5 ads per page. So, I think the number of ads per page is 5. Suppose I see 50 percent of the ads I see that are clickable and 50 percent are display ads.

Therefore, I set the percentage of clicks to 50 percent. Good. Now, I can multiply these numbers to get the number of click ads visitors see in one day: 10X5X0 . 5 s 25 " .

- Calculate daily ad clicks

Now, let's go to the parent node of the number of visitors clicking on the number of ads. Of the top 20 publishers, The New York Times is one of the most popular in the United States, with about 10 million subscribers.

Therefore, I would consider this to be the maximum range of subscribers. The rest of the top 20 publications will also have millions of subscribers, but less than the New York Times.

Therefore, suppose the number of subscriptions ranges from 1 million to 10 million. Since no other data is available, I will average 5 million subscriptions for these sites.

Since these subscribers are likely to visit their subscriptions every day, I assume that this is also the number of visitors per day.

Continue to calculate daily click revenue. Will assume that the CTR is 0 . 5%, while the average cost of CPC is $0 . 5。

Therefore, we can put "Number of visitors per day" X "25 clicks on ads per day"X "Click Rate" X "CPC" calculates daily click-through advertising revenue, i.e. 500WX25X0 . 5 %X05 s 31 . 25W/day.

- Daily Ad Click Revenue

Similar to click ads, to calculate the revenue from daily display ads, we calculate the number of display ads that visitors see, and on this branch, all nodes have the same estimate as click ads, because the number of pages that readers visit each day is the same regardless of which ad is displayed on the page.

The same applies to the number of ads per page. Also, I assume that 50% of the other ads are display ads. As a result, visitors see 25 ads per day.

Now, let's look at its parent node, and based on what I know about the average 1,000 impression costs of display ads, my estimate is $1.

Therefore, in order to generate daily display revenue, I multiply these three factors, namely "5 million visitors per day" X "25 display ads" X "0 . 001" s 12 . 50,000.

- Daily Display Advertising Revenue

Based on the above results, the total revenue of daily ads is calculated, so the total revenue of daily ads =31.25W+12.5W=43.75W。

Now, we have all the numbers that can be estimated each year 80% of GoogleAdSense revenue . . . 20X43.75WX365 . . . $ 3.2 billion .

Total Revenue : $ 3.2 billion X1 . 25 - $4 billion.

- 80% of advertising revenue

**Interviewer** : OK, what do you think of that? Do you think it's reasonable?

**Conduct a reasonableness check**

**Interviewer** : Well, I think the estimate is much lower than Google AdSense's actual revenue.

Google NetworkMember's revenue is about $17 billion, so my revenue has been reduced by about four times.

**Interviewer** : How do you explain it?

**Interviewer** : I think my assumption is that only subscribers are daily readers, which is too conservative.

In fact, most of the people who access these publishers may not be subscribers. They're probably visitors who don't want to pay for subscriptions, but they want to read the publisher's content.

Therefore, I suspect that the ratio between non-subscribers and subscribers accessing these publishers may be 4:1, which explains the difference in data.

**Interviewer** : Yes. So, how will you change the calculation?

**Interviewer** : The ratio of the daily visitors to the sum of the number of daily subscriber visits and the number of unsubscribed visits is approximately 1: 4 ,This explains the difference in data.

Interviewer: Yes. So, how will you change the calculation?

Interviewer: Change the number of daily visitors to the sum of the number of daily subscriber visits and the number of unsubscribed visits, and their ratio is approximately 1 : 4。

- END -

Product Interview Series - How to Answer Estimation Questions

Product Interview Series - How to Answer Estimation Questions

Product Interview Series - How to Answer and Improve a Product

Go to Discovery - "Take a look" And browse "Friends are watching"