Guesstimate problem for Product management interviews

Top-down approach to find the number of reviews on Zomato

Product management interviews have always been challenging (and daunting, for a few). While interviewing the candidates, the interviewer always tries to judge a (future) product manager on different criteria such as clarity of thought, proper structuring of the case based questions, ability to communicate, basis for the answers effectively and so on.

Guesstimate has been one of the all time favourites of product management interviewers. It not only helps them to separate the wheat from the chaff but also gives them a glimpse of your ability to solve (read approach) a situation with a limited understanding in a defined time frame. For this blog, let us try to solve one such interesting guesstimate.

How would you find the Total number of reviews on Zomato?

In order to answer this guesstimate, the very first thing we need to do is to clarify the question itself. Here, for instance, a clarification question you can ask is if the interviewer is interested in knowing the total number of reviews posted in a year or since the inception of Zomato.

Clarification / Scoping the problem

Clarifying questions establishes a dialogue between you and the interviewer, gives you a definitive direction, and signals to the interviewer that you are able to visualise multiple scenarios. Once you get some direction and understand what the interviewer is looking for it would be easy to lead and hypothesise the numbers. For the sake of this product management guesstimate question, we will consider that the interviewer wants us to find the total number of reviews on Zomato posted from the beginning of time ie from when Zomato came into existence.

The next thing you should do is to update the interviewer which approach you think will work best for the question presented and why. For this guesstimate we will be using the top-down approach (in this you start with a top ceiling number, say the population and keep on segmenting, breaking it down further as you move ahead towards the solution). However, you are free to try and solve the same problem using a bottoms-up approach if you know certain numbers in the food-tech space.

For the sake of simplicity, we will break the approach in 8 simple steps:-

STEP 1: Once you have clarified the basic assumptions and decided the approach, you should ask yourself where can one find the people who write the reviews on Zomato.
We can take into account that in recent years Zomato has gone international and is now present in many countries. So, it will be safe to assume that currently Zomato is providing its services in 15 countries with India being its biggest market. Once you have defined the number of countries, you will move to the total population of these countries. Let us assume that the current population of India is 130 crore and the average population size of rest of the 14 countries is 5 cr.
Thus, total population living in these 15 countries are: 130 + (5 x 14) = 130 + 70 = 200 cr
(Note how we have assumed the numbers in a way that we arrived at the population rounding to the nearest 100. For next few steps, this will help us in performing mathematical operations faster).

STEP 2: Now, let us run our population through a couple of segmentation filters which would help us identify the right population size, who might be using Zomato.
The first segmentation can be on the basis of area. Since, food delivery apps are popular only in urban cities, for this guesstimate we would assume that nearly 30% of the population in these 15 countries are living in cities and the rest 70% either in rural areas or sparsely populated areas where food delivery app services might not be available. So, total population of the urban cities where any food delivery app could be providing its services are: 200cr x 30% = 60cr
(You can present the rationale that since India forms are significant portion of 200 cr population, for the sake of simplicity we apply 30:70 ratio of urban:rural divide over the complete population size).

STEP 3: Second level of segmentation which we can use is Age group. It can be the most relevant factor in deciding which type of population generally orders food online. We know that food ordering apps are quite popular among teens, college goers and working professionals. Do clarify this with the interviewer if you are taking a good enough assumption. Now, let us assume that people under the 15 to 50 age bracket within cities have a higher number of college goers and working professionals. Also, it would be fair to consider this age bracket will form nearly 60% of the total population living in urban areas. So, the total number of population under 15-50 age bracket (who order food online) will be: 60cr x 60% = 36cr

STEP 4: Before we move ahead let’s apply one final segmentation filter on the 15-50 age bracket. How can we figure people who order food online and those who don’t. Since this data is almost impossible to assume accurately, we would ask the interviewer if it’s safe to assume that only 50% of people falling under the 15-50 age bracket prefer to order food online. So, total number of population who orders food online (via different food ordering apps) are: 36cr x 50% = 18cr

STEP 5: Now, we are aware that food delivery is a fiercely competitive space with 3-4 major players present in each country. Hence, let us assume that Zomato, being a giant player, has captured nearly ⅓ rd of the market share. Thus, the total of people using Zomato app at any particular given point of time out the total people who order online could be: 18cr x ⅓ = 6cr. This 6cr is the approximate number of people who are using Zomato to order food.

STEP 6: To arrive at the number of reviews we should now understand how many orders are placed via the Zomato app in a year. Based on the frequency of ordering, we can divide our target audience in three buckets -

• High - those who order 10 meals per month from Zomato
• Medium - those who order 4 meals per month from Zomato
• Low - those who order 1 meal per month from Zomato
  

This would give us the average orders per month per person, which would be: (10+4+1)/3 = 5 orders per month per person. So, on average at any given month the total number of orders placed on Zomato should be: 6cr x 5 i.e. (number of people who use Zomato x avg order per month per person) = 30cr per month or 360cr per year (30cr x 12 months). For the sake of simplicity in future calculations we would consider the number to be 350 cr.

STEP 7: Now that we have the total number of food/meals ordered from Zomato in a year, we should move to the total number of reviews posted in a year. For this we can safely assume here that after ordering the food, not everyone gets enthusiastic to post a review. Once again we would divide the population in 3 buckets based on the frequency of posting the review:

• High - review posted 20% of the time (or at every 5th order)
• Medium - review posted 2% of the time (or at every 50th order)
• Low - review posted 1% of the time (or at every 100th order)
  

So, the average review per order would be: (20+2+1) / (100 x 3) = ~ 24 / 300 = 0.24 / 3 = 0.08 reviews per order or 8 reviews per 100 orders
The total number of reviews posted in a year = total number of orders in a year x no. of reviews per order = 350 cr x (8/100) = (700/2) x (8/100) = 7 x 4 cr = 28 cr

STEP 8: Now that we have the number of reviews posted for a year, we need to find out total reviews that have been posted on Zomato till now.
Zomato has been present for more than a decade but in its initial years (as a restaurant discovery app) the number of reviews posted might have been fewer. Reviews posted per year would have increased only in recent years (with the popularity and adoption of online food delivery). And the online food delivery picked up around 2015, so to calculate the total number of reviews present let us take the last 5 years in account. Hence, total number of review on Zomato = 5 x 2 cr = (10/2) x 28 = 140cr
(Please note that here I’m assuming that each year we would have the same number of reviews since we have not considered the (fewer) reviews present before 2015).
Final take: Since guesstimates don’t have any right or wrong answers all that you need to present in a product management interview is how logically you are structuring and how rationally you are approaching the problem presented. We hope that you would have found this blog useful. PM school wishes you best for your next product management interview.

I hope this will be helpful to you, if you want to prepare for Product Management role by Microsoft, Adobe and Accenture Product Managers you can join our latest cohort batch.

Program Link