How many reviews are left on Amazon on a given Monday?
Estimate how many total reviews are left across Amazon.com on any given Monday.
You'll get access to over 3,000 product manager interview questions and answers
Recommended by over 100k members
Clarify:
- US only or world wide -> Assume US Only
- Amazon Fresh / Whole foods to be included? - No just amazon.com retail reviews
Let's come up with an equation
# of household that makes purchases on a given day * % of users who write reviews
Let's dig deeper:
# of household that makes purchases:
- There are roughly 300 Mil people in US.
- # of households = 300/3 = 100 Mil households
Out of 100 Mil, I'll take 80% of the US household has ordered atleast once on Amazon.com = 80 Mil
I know Amazon's revenue for 2019 was roughly 300 Billion and 5% of that came from Amazon Prime but that is world wide. So I'll choose 2% from Prime, Which equates to 6B. Now let's devide 6B by 120 (Prime subscription) which gives 50Mil.
So out of 80 Mil that have used amazon.com, 50 Mil are Prime users and I would tag them as high usage (buys multiple times a week) and I would put 20 Mil as Bi weekly shoppes and 10mil as low usage (once a month)
Out of 50 Mil, I would say only 10% of the people write reviews and probably around the same for biweekly and I would not touch monthly or low usage as it would be negligible (Low engagement anyways).
Rationale for 10% : Most of the reviews are if the users are very happy or very unhappy about a product and Amazon does a good job of vetting products that are really crappy.
So that puts us at 10% of 50 Mil = 5 Mil and 10% of 20 Mil = 2 Mil and a total of avg 7 Mil reviews a week
Now, I know for the fact (I have read it somewhere) that majority of the delivery happens on Wednesday and Majority of the Orders happens on Thursday / Friday (Excluding any holidays / deals where there is a spike through out the week)
- Given the above information, I would say 60% of the reviews are written on Wed / Thursday => 60% of 7 Mil = 4.2 Mil
- I would then allocate 10% each of reviews are written on Friday, Sat and Tuesday => 700k each = 2.1 Mil and
- 5% on Sunday and 5% on Monday which equates to 350k each since there are very less deliveries on Sunday and Monday being a working day (busy) after weekend, I would keep it at 5%.
Self Critique: The highest number of ratings I have seen is 2000 reviews for a product and least is couple hunderds. The patterns I have seen is that less than 25$ item gets a ton of ratings and as the price increases the number of reviews decreases. Perhaps we could split the 10% rating for weekly and bi weekly based on average product price and then do the math.
Let's assume that the question is about Amazon India.
90% of the time reviews will be given by people who have purchased that item through Amazon recently. 10% could be by people who have already purchased it outside and wanted to share their experience.
So, we I derive the average number of reviews on a day from the average number of orders and the % of users who purchase and give reviews to products using the below formula:
# of reviews per day = # of orders per day* % of users who give reviews
# of orders per day = ((# of users who shop online in india * average # of orders per year ) * Amazon India's market share)÷365
In india about 1/3rd of the population is shopping online which is about 400 million
From avid shoppers who shop weekly to people who shop rarely, it would be safe to assume that a user shops twice in a year online. That would make it 800million orders. Amazon India market share is about 30% . So the number of Amazon orders = 240 million.
# of orders per day = 240 million÷ 365 = ~ 750k
Usually about 10% users come back and write a review of a product that they purchased.
# of users who give review per day= 10% of 750k = 75k
Usually people have more time during weekends and tend to right more reviews than average and Monday being the week's starting day, may see less traffic than weekends. If we assume Mondays receive only 80% of the average activity, the average number of reviews on a Monday could be equal to 60k
So the question is ‘How many reviews are left on Amazon on a given Monday?’
Step-1 : Clarifying the scope of Question:
Are we looking at a specific region like US, India? Say we need to focus on US.
Amazon has many product lines like Amazon Fresh, Amazon music, Amazon e-Book e-commerce etc. Are we looking at a specific line of business ?
Lets assume the interviewer says you can narrow down the scope as per your consideration. In that case, let us focus on Amazon.com and no specific business line.
There are many other e-commerce sites that Amazon owns like shopbop, zappos, and Woot. Are we also considering those additional sites?
Say we will focus on the Amazon.com for now.
Lets take a moment to think about the calculation and list down the equation.
Step -2 : Main Equation
To be able to calculate the number of reviews on a Monday, we will need to calculate the number of orders received on a day multiplied by the percent of buyers who share a review on the site about their experience with the whole buying process or about the product feature that they bought.
#Number of Reviews = Total number of Orders online * percent of shoppers giving a review online
Step – 3 : Let us dive in further Break down each component.
#Total number of Orders online = # of users who shop online in US * percent of Amazon’s market share in US
ð #of users who shop online in US = 70% of US population
ð Market share of Amazon in e-commerce in the US = 50%
Step-4 – Calculations and Assumptions
ð Assuming US population is close to 300 million, the equation comes to
ð 350 million * 70% * 50% = ~120 million
Facts say around 25% of US online shoppers have shopped at least once a month. Assuming there would be frequent shoppers as well as casual shoppers who have shopped online once a while, it is safe to assume there can be 3 shopping orders per person every year.
So #Total number of orders online = 120 * 3 = 360 million
#Total number of orders online per day = 360 million/365 = ~1000 k
Assuming there will be 10-20% of users who will leave a review,
#Total Number of Reviews = 1000k* 20% = 200k per day .
Assuming Monday is a busy work day, people usually would have provided their reviews on weekend. So there will be considerably less reviews on Monday.
#Total number of Reviews on Monday = 5% of total orders received = 1000k * 5% = 50k
Step-5 checks and revisit edge cases if any– We are not considering the Prime users who usually shop multiple orders in every week. We assume Monday is a working day for most folks. What about the return reviews as some of hot products like clothes will be tried and verified on Monday? That can add up or compensate the review rate to some extent say by another 10% . So the reviews could be around 55-60 k on a normal Monday at Amazon.com shopping site in the US.
Clarification questions -
- Any specific region that we are looking at? Say US only
- Any specific channel? Say mobile and desktop
- Any specific time frame? Say one year
Assumptions -
- Reviews on Monday will be on orders done during the weekends
- Majority reviews are given by people who have ordered a product from Amazon
- Products are delivered in 1-2 days
- Rating is given before review in the product flow
- Amazon.com has a logged in user base of 1 million, of which say 5 lakh place orders
- Each user places, on an average, 5 orders
- Total orders in an year - 25 lakh
- Assuming majority (60%) orders are placed over the weekends, as that is when people have time
- Total orders placed over the weekends - 15 lakh
- Assuming each order will not get a review -
- Split of type of users who order - Motivated (20%), Demotivated (60%), Dissatisfied (20%)
- From the above lot, only the "Motivated" and "Dissatisfied" will be eager to give rating
- Only 10% of the users who give rating actually write a review
- Total ratings (on all Mondays) - 40% X 15 lakh = 6 lakhs
- Total reviews (on all Mondays) - 60K
- Total reviews on any given Monday - 60000/ 52 = 1153
Top Amazon interview questions
- A metric for a video streaming service dropped by 80%. What do you do?50 answers | 135k views
- Design a library for the future.25 answers | 23.5k views
- Your new feature boosts Amazon Search by 10%, adds 2s to load time. What do you do?19 answers | 36k views
- See Amazon PM Interview Questions
Top Estimation interview questions
- Calculate the number of queries answered by Google per second.45 answers | 78.5k views
- How do you decide how many cash registers you need for a new Walmart store?33 answers | 39.4k views
- How many dentists are there in New York?33 answers | 36.7k views
- See Estimation PM Interview Questions
Top Amazon interview questions
- How would you improve Amazon?14 answers | 35k views
- How would you determine if a specific block in your neighborhood is suitable for a new grocery store?14 answers | 13.4k views
- You have a grocery delivery service that delivers food within 24 hours. How many trucks would you need to operate it?12 answers | 29.8k views
- See Amazon PM Interview Questions
Top Estimation interview questions
- Estimate the number of Uber rides.30 answers | 51.3k views
- Estimate the number of bicycles required to start a bike sharing operation in a big city.18 answers | 15.8k views
- Estimate the number of WhatsApp chats occuring in India.14 answers | 13.3k views
- See Estimation PM Interview Questions