this has got to be one of the trickiest questions you'll get in a product interview how would you reduce churn for product X so in this video we're going to tackle this question and I'm going to show you the structure some common mistakes and tips so you can Ace this interview so first we're going to talk about what is the product we're talking about Uber has a multitude of products of course it's Flagship product is the Uber rides product but it also has ubereats and now it's launching into other verticals where it's trying to deliver
your things from your favorite grocery store or Pharmacy so what product are we talking about let's say the interviewer tells you we want to focus on the Uber ride hailing app great now the next step we have to figure out well who is churning when we talk about Uber we have the driver and the rider and this could be a question that I posted an interviewer or you know I'm going to make the assumption that the driver turn is actually more problematic than Rider turn because think about it one driver supports multiple Riders versus if
one Rider is training it's not really going to have that worse of an effect on other riders or drivers so if I had to prioritize one that I think is the most problematic to solve it would be the drivers now we've honed in on who we're driving down churn four let's understand this user's user flow now if I'm thinking about the driver the first thing you're going to do is sign up for the Uber driver position which requires things like a background check requires getting their car are registered and to get a specific license from
certain governments and after all of that is completed they'll start doing their first rides and after the first rides they will get raided and this rating will help Uber determine do they keep this driver on the app or potentially fire them and lastly the rider will get their first paycheck the first tip I have for you here that the user flow is going to help you build your hypotheses on how to reduce churn why because understanding the user flow will help you build empathy towards the user that we're talking about and that will help you
develop hypotheses which will help you then develop Solutions now that we've defined the problem a little more let's go into what data you would need to actually solve this question so my main tip here is you want to ask for what data will help you make progress on this question but note you might need to make some assumptions because the interviewer might not necessarily give you data or have data ready for you so you want to show the interview that you can know what to ask for and then you can make a best guess and
this is going to show your product intuition so let me dive in some examples so here the first piece of data I want to understand or even construct is let's define churn when we talk about churn in this case for drivers I'm going to assume churn could be when a driver hasn't come back to the app for let's say two weeks or turn could be if the driver hasn't taken any clients or customers on a ride after two weeks so up to me to Define that and here I'm going to prioritize those driver who have
not taken any rides for the last two weeks why because a driver can come back into the Uber app but if they're not accomplishing the key thing that is tied to Uber's Revenue it's not going to matter if they're spending a lot of time on the app or Not Another insight as I'm thinking about defining churn is that churn for drivers is actually expected if you think about the nature of the driver work a lot of these drivers are using Uber driving as supplemental income this might mean they might be driving just on weekends to
supplement their existing full-time job but there are also other drivers who are full-time drivers like taxi drivers so that's to say churn is expected for drivers but we have to figure out what churn we can actually solve and what churn is having the largest negative impact on our ecosystem another tip here is when you ask for data that will lead you to insights that allow you to make these assumptions for example when I was asking to Define churn in my own head to myself that made me realize that churn is actually expected on the platform
for drivers based on what the product next I want to segment the data that I have so here in a real life scenario I would segment which part in the driver's user flow are we seeing users inactive after so for example we might see that after the payout period after week one or two that's when drivers are becoming inactive and churning and that might lead me to a hunch that maybe in the first few paychecks drivers are realizing they're actually not making enough money or what they expected another segmentation that I might want to take
is based on personas so I mentioned there are different types of Uber drivers those that are full-time drivers that maybe transition from a taxi company those that might be doing driving every weekend as supplemental income to their full-time job those who might be only participating in driving to during surge pricing or huge events where they know they're going to make a ton of money and those who might be coming back once every year once every six months when let's say they lose their job or they go in between jobs so another Pro tip here is
you really want to use user empathy to help you make assumptions so here I talked about the different types of drivers and that was me putting myself in their shoes to then be able to segment out the different personas another cut of data that I may ask for is what is the term looking like is it a one-time thing or is it a pattern that's happening after every three months or two months how long is it taking for drivers before they churn and here I have a hypothesis slash assumption that drivers are turning pretty early
on let's say after a month once they realize they're not going to make as much income as they thought and a last piece of data that I might ask for is to interview users an easy way here is talk to your customer service reps we're talking to these drivers more regularly and at a larger volume that can give you a sense of what they're hearing is going on with why drivers are leaving so one assumption I might make when I hear from customer service reps is those drivers that are leaving are having negative experiences is
those who are calling in about maybe an accident happening or a negative experience with a rider so all of these questions that I'm asking and the data that I'm asking for even though I might not have exact data unless the interviewer gives it to me is already allowing me to make some assumptions and come up with hypotheses so these are my three hypothetical hypotheses the first being a lot of turn is coming from these event-based occasional drivers these drivers are driving when there's a large event let's say a sports game or a huge concert that's
happening where there might be surge pricing and hence it's natural for them to churn because they are the type of people to work during certain occasions and this group we might ask ourselves is there a way to optimize it so they don't turn as often or do we let them be and instead focus on the group that should not be turning which are those full-time Taxi drivers my second hypothesis which is more relevant to full-time taxi drivers are that drivers are not earning enough to stay on the platform and that's why they are trading they're
looking at other things other job opportunities that they could take that will earn them more and my third hypotheses based on talking to the customer service reps I mentioned is certain drivers turn after they have a really negative experience I remember one time talking to an Uber driver who was in a neck brace and said that he is going to leave the Uber platform because Uber didn't provide him any insurance after he got into a really bad accident so I would not be surprised if a lot of Uber drivers are leaving because of negative experiences
with writers to move to the last step I first need to prioritize of these three hypothetical hypotheses which am I going to prioritize to solve and here as you know with any decision in the product interview start with your criteria and here I have two and first is the size so which of these are happening the most frequently second criteria is which has the most impact on our revenue of course Uber as a business cares about the revenue that drivers are earning because they take 20 cut from what drivers earnings are so if I start
off with size frequency I'm already going to take out the negative experience one because I think that one makes up the minority of cases if Uber designed a good experience I think at this stage many years later it should have gotten that right and hence that leaves me with the first and the second and between these groups which do I think has the largest impact on Revenue I'm thinking the second because the first I mentioned are these occasional drivers who despite what we do are on their own terms deciding when they come earn on Uber
versus the second hypothesis is that drivers not earning enough apply to this full-time taxi drivers who could be turning because they feel like they're not making enough money on Uber so I would say our top priority are these potential full-time drivers and less so the occasional drivers because they're the ones that are going to be servicing most of our customers and staying on the platform long enough to bring a huge lifetime value to the ecosystem now that we've defined Which hypothesis we can get into solutioning so a tip here before I go into some potential
Solutions is when you look to think on Solutions dig deep with the five Y's for example I might be tackling why are drivers not earning enough and that leads me to break it down into two things the first being drivers don't know where to go to earn and a second there might be mismatch expectations from drivers how much they think they're going to earn versus what is realistic if I'm tackling the first challenge drivers just don't know where to go to earn a couple of ideas that I have to tackle this the first one being
customize optimize tips for each driver based on their performance Behavior Etc so for example whoever might notice that these new drivers are not earning that many tips so once they identified that Uber can then send a notification to the driver to say do you know that you could be earning let's say two thousand dollars more a week and here is a list of things how and some best practices for example I would say tips makes up a good amount of how much an Uber driver can earn it could be a ten dollar ride and someone
because they had a good experience might tip another three to five dollars and this could be a great effort because it shows drivers that Uber cares about the drivers which will keep them and retain them on for much longer a second idea I have is let's just prioritize new drivers so when the algorithm goes to choose which drivers to get matched with writers let's get the new drivers to get matches more this way when new drivers join the platform they get the sense that there is a lot of business to be had another solution could
be hotspot Maps so here Uber would show drivers potentially the opportunity to go to different cities that are nearby to drive where they're higher earnings or it might show more locally based on an hourly or daily weekly what are some of the popular areas to go to if I'm tackling the Second Challenge which is helping drivers better match their expectations I may come up with a solution what I call an earnings calculator which based on inputs from the driver where they're adding in this is the number of hours I expect to work on these certain
dates in these specific areas it returns back to the driver here is how much you can earn in a week or month and that way drivers don't have this unrealistic expectation of how much they're going to earn obviously we should show them a range rather than an exact amount because it really depends on market conditions so these type of questions what would you do to reduce churn are in the category of debugging questions so you might receive other questions like we saw this data go up or go down what should we do so check out
these other videos where I show other examples of debugging questions so that you can Ace interview and I will see you guys in the next video
