Decoding Estimation Puzzles: From Google’s Queries to Bandwidth Forecasts and Troubleshooting Celebrity Website Traffic Drops
Calculate the number of queries answered by Google per second.
Okay for the sake of this question we will assume that we are looking at the entire population of Earth and on any mode of use be it mobile or web.
Population of Earth: 7B
Countries that don’t have google: China (Baidu) and Russia (Yandex 60%)
China population +60% of Russia population: 1.4B + 1M = ~1.5B
Internet penetration of the world currently stands at 65%.
Among these there would be different types of users who use search:
a. Sparse: Old people, Kids, Our Parents etc => 1 search/day
b. Occasional: Us, Students etc => 2–3 searches/day
c. Frequent: Researchers, Publishers => 5–7 searches/day
Above numbers are w.r.t the current scenario meaning, if we need a product we would directly go to Amazon, LinkedIn for Jobs and YouTube for videos etc, if that was not the case then above numbers would be higher.
Now, 65% of 5.5B = 3.3B (effective population)
Considering 50% for Sparse + 30% for Occasional + 20% for Frequent
=> 0.5 (3.3B) * 1 + 0.3 (3.3B) * 3 + 0.2 (3.3B) * 5
=> 1.65B + 2.97B + 3.3B
=> 7.92B searches per day
Finally, 7920M/24hrs/60mins = 5.5M/60sec = 91,000 searches per second.
That number doesn’t seem far off. Let’s check the internet now.
Looks like we are approximately correct!
If you wanted to build the world’s most popular mobile messaging product, and you need to estimate how much network bandwidth would be used in a year, how would you go about doing this?
Let us assume popularity here means an app with high downloads and high engagement.
World population: 7B out of which say 75% own a smartphone this excludes people who can’t afford one and children/infants who can’t have one. So, effective population: ~5B
Now, we would have 3 types of users in a messaging app:
a. Occasional: Few messages a day: ~2MB/day
b. Regular: Few messages and photos: ~12MB/day
c. Heavy: Photos Videos Docs and text: ~50MB/day
Total bandwidth per day= 20% population of Occasional + 50% population of Regular + 30% population of Heavy
= 1B*2MB + 2.5B*12MB + 1.5B*50MB
= 107 Billion MB/day
Multiply the above number by 365 is what will be our required bandwidth per year.
If you host a celebrity website which displays ads and suddenly notice a drop in traffic to your site/clicks on ads, how do you root cause the issue?
The way I would approach this is:
- I would know the whereabouts and public meetings of my celeb, I will try to find out if they said something in a public setting that would lead to a negative image of my celeb.
_> But this doesn’t make so much sense as only a substantial portion of public would not want to engage with the celeb page but rather, they would want to engage and post negative comments or find a way to dislike, this would increase traffic not drop.
- Then it comes down to figuring out if this issue is present only in a part of the site/type of ad or else on the entire site.
_> If this is only on a part of the site or the type of ad we would review things in reverse chronological order try to find the root cause.
_> If this is through out the site we would then find out whether this is occurring only for our celeb or all celebs in that domain in general.
- It is also very vital for us the check whether out host is up and if our analytics platform is working properly.
- I would also check if there is any search algorithm update by Google and the likes that affected our websites SEO badly.
- We can also use our analytics platform to find out through what all means was the traffic coming to our site and which sector is affected and then dive into the root cause of the same.
- We should also look at our recent changes which have been deployed if there is any technical defect or if there was some issue with the content pushed.
- Sometimes it is possible that everything is fine, but a competing celeb becomes more popular, and public adores them more resulting in the traffic drop.
- Is there any latest news that has grabbed everyone’s attention for the moment that is causing the traffic drop? (Cough! #pandemic).