Empower Your Product Management with the Digital Market Funnel for Product Management

How to create outlook signature

hi everyone my name is Tanvi thank you so much for having me thank you so much for being here on a Thursday evening I what I'm hoping to give you today is a quick walkthrough of what it is like to be a product manager in a marketplace business how I got here and if I have time towards the end I walk you through some prep interview questions if you yourself would want to be a p.m. or interview as a p.m. in a marketplace business all right so product managing marketplace businesses what are marketplace businesses what is unique about product managing them how do i how do you identify a product manager in a market based business and then some of the key Ted components of what that entails let me start by telling you a little bit about myself so I was born and brought up in India I moved to the west coast east coast of the US to study computer science at Bryn Mawr College after doing that I had the startup bug I really wanted to work in the big economies face TaskRabbit was just starting out in the US and I decided I wanted to move to India to try and do that from Mumbai so started a company called chachi comm I did that for a year built the marketplace from the ground up recruited a bunch of service providers did a ton of fundraising build the website but then realized I was all of 22 and didn't know what I was doing so it was fun for a year but then I decided I needed a paycheck moved back to the US where I did PM at Microsoft kind of fell into it by accident but it was kind of a perfect fit from the get-go I started out on the windows team built Windows 8 worked on the browser then moved to Microsoft research where I worked on machine learning tech for language so I worked on translation technology and then I decided that I needed to be back in the gig economy space and uber worked out so joined the uber marketplace team first as a p.m. on their matching team and now I'm a p.m. on their shared rights pricing and matching team where I lead a bunch of pricing and experience related efforts so I'm hoping to share some part of the service marketplace experience okay today a little bit about why I am personally deeply excited by marketplaces and like why I get up in the morning excited to do my job I think we're in the golden age of service marketplace tech businesses for these three reasons one depth and scale of impact I'm really compelled by the flexible employment opportunities it entails whether you're driving a car or renting out your spare room the flexible employment opportunity that that marketplaces provide previously did not exist in our economy and that works on the opposite side to transportation accessibility to underserved communities which previously didn't exist I think both the depth and scale of impact in marketplace businesses is extremely unique tat this point of time in the global economy number two we are at extremely add a zero of where marketplace businesses are uber and lyft for example serve less than 1% of global transportation miles traveled in the world today there is a lot more to go same for many other untapped service verticals so I think the future potential of impact of building skill sets of growing and scaling service marketplaces is huge and third marketplace science if you go to any medium to large-sized marketplace business anywhere in the world it's just a bunch of crazy data scientists running extremely interesting experiments in essentially a pure snapshot of an economy of pure supply and demand I would recommend if people want to familiarize themselves a little bit more with marketplace economics and science I listened to the freakonomics episode by Steven Levitt who wrote the freakonomics book called why Eber is an economists dream and he talks a little bit about why to an economist a platform like Burberry supply-and-demand herbs are the in their purest form possible and that data is extremely accessible is super interesting as an academic and you get to do some of that as a PM in a marketplace business which to me personally is super exciting so yours what I'm hoping you'll get out of today one I'll tell you a little bit about what a marketplace business or a service marketplace businesses I double-click a little bit into the key tech components specifically matching and pricing with ride-sharing being my canonical example because that's my bread and butter but these generalized learnings can be transferred to other kinds of businesses 3 what a product manager in a marketplace business looks like and for if we have time towards the end I'll do a quick mock interview if you were to interview for a market based business what kind of questions could you expect and what kind of answers the interviewer is looking for all right what is the marketplace business so as simply as possible a market based business is anything that takes two groups of people supply and demand or maybe two plus if it's let's say uber eats our door - to supply demand and let's say provider like a restaurant and connects their needs and wants in a centralized fashion so I'll be speaking from an educated point of view about lyft and uber but there's eBay five were Angie's List handy even Craigslist is an example of a market based business and the stuff I'm talking about today kind of encapsulate all these examples when I think about market based businesses and specifically service market based businesses I mainly think about two dimensions within which they can be divided so my x axis in this case is is this business a matchmaker just does connect to people and then leave them to conduct their business or is it a full-stack service provider from the intent expressed all the way to service delivered my y-axis is is this a provider that just aggregates supply that already existed or is it creating supply that previously didn't even exist so let me take an example Craigslist is your canonical example of something that's just a pure aggregator just a web page with a list of links and contact information and free-for-all it's simply a matchmaker Craigslist takes no responsibility for the sketchy things you find on their platform and you do not expect Craigslist to take responsibility for that Internet service on the other end of the spectrum let's take Airbnb or post mates Airbnb arguably created supply that previously didn't exist there was no way for you to put your spare bedroom on a platform and have it rented by reliable people so it created supply that previously didn't exist mostly and it's a full stack service provider not only do you see these listings but you can pay for them you can review them you can communicate with your hosts it's an end-to-end service and at the end of the day if you don't have a good experience with your Airbnb Airbnb is at fault on the other end let's take handy as an example handy is something that aggregates existing cleaning cleaning providers and is also a full-stack service provider but didn't create supply that didn't previously exist so generally I mean these are just broad strokes but generally when thinking about service marketplaces think about dimensions in terms of how deep into service providing is this marketplace going versus just aggregating and then is this marketplace creating supply that didn't exist which has a completely different set of type challenges versus is it just an aggregator of supply that already existed Andrew Chen who is kind of like the Pope of marketplaces wrote this really good essay called what's next for marketplace businesses and I found it as a really insightful because he identifies these four chapters in where marketplace businesses have over the last three decades so he starts with what he calls the listing era the Craigslist Yelp era just just as the internet was getting more popular and more and more people had access to post and add content Craigslist and Yelp kind of grew in traction giving people access to those platforms but they took zero responsibility for other services associated with that listing then was the unbundled craigslist era so for example something like an like Angie's List gave you a very specific specialized listing platform but again was not responsible for some of the other services associated with it then was the uber 4x era uber lyft Airbnb are really good examples of the uber 4x era where now we're starting to get into supply creation and not just supply aggravation and this is where payment systems and real time information and live location sharing all that tech starts to click into place and gets really good and now we're in the managed marketplace era and I alluded before to service verticals and just and getting those service verticals right the managed marketplace era is essentially all about taking a very specific service vertical let's take honor as an example which is a startup that serves senior care so it's a network of service providers who are specifically allocated and vetted to provide high quality senior care all the way from identifying a service provider to actually getting your service completed to reviewing and specializations and so forth it's not as much of a commodity as let's say an uber riders you don't really care who your uber driver is as long as you get from point A to point B safely when you start getting into the managed marketplace services you do care who is taking care of your grandparents and therefore there's a higher quality bar associated with it and therefore a lot more operations muscles that those are that are required so again when you think about service marketplaces and if you if you're interested in kind of the evolution of technology and evolution of these business models the Android chance what's next for marketplace businesses is a create read that double-click Center some of this so anyway at a high level any marketplace is going to reduce entropy between demand and supply and adding speed quality and reliability to that service by leveraging the power of that network as simply as that that is a marketplace with tech kind of powering the brain of it so you may ask all right service marketplaces no not new news taxis existed for a really long time hotels existed for a really long time what is the value add of putting things on a smartphone app I like to use Los Angeles's taxi network as a canonical example of what radical efficiency that service marketplaces bring can do to a to a market so free uber and lyft back in 2010 2008 Los Angeles was doing 8.4 million taxi trips a year once uber and lyft came into the market 2012 2013 everyone starts installing uber and lyft on their platforms taxi trips reduced down to 6 million so you might say ok well you know Burren left its cannibalizing the taxi business uber and lyft as of 2017 created hundred million plus trips annualized in 2017 not only did it surpass where taxis were just a few years ago it 10x2 that volume just by virtue of the fact that your ETA is very reliable your pricing was structured everything was faster than before and generally had some confidence of the quality control associated with your experience so yes while some of the taxi business was cannibalized the 10x incrementality that this business adds to this ecosystem is just unforeseen and that's the kind of radical efficiency that service market faces can bring to already existing verticals that are honestly disruptive alright so we talked a bunch about water market-based business is now let's double click and let's double click into any medium to large-sized market-based business and look at kind of the structure and the limbs that constitute of what makes it work again I'm speaking from my uber knowledge but every marketplace business I have seen in the Bay Area kind of follows the same structure so in a market based business you have a few core teams in the beginning they are kick-started by operations like manually people going out on the streets and doing recruiting and eventually they get replaced by tech so engineers PMS and data scientists so you usually have supply growth how am i recruiting my drivers where am i finding them from what is the registration process you have demand growth what kind of promotions can I offer refer to a friend and get $20 off you have a safety and quality team how do I make sure that the people on my platform are having a safe experience you have user experience in ops like doing the pricing and the actual UI associated with the experience and then each marketplace business will have kind of its heart in the inner marketplace team and the marketplace team in the beginning usually it's just Excel sheets with prices but eventually as it matures and the business grows it becomes a powerhouse of machine learning based tech that makes really smart and really quick real-time decisions so for example the magic that happens when you request a lift and you get matched with someone else or the surge price that kicks in when the market is supplied or under supplied all that happens within the marketplace team and that Tech has to mature really really fast like when you're talking about hundred million ride-sharing trips analyzed in Los Angeles multiply that by four hundred cities where uber exists so that tech has to be really resilient and really responsive because the decisions it's making is having ramifications in the real world look at Twitter anytime uber goes down for like 15 minutes and you will see the ramifications of the marketplace text team not team's tech not working all right so picture the marketplace team what does this team do the way we think about the marketplace teams is the supply journey and the demand journey demand in this case in the uber cases the rider supply in the uber case is the driver so let's talk to the rider first the rider first opens her app she looks at the price she looks at uber X uber pool premium maybe makes a decision based on those prices mixer request is on trip and then once the trip completes has a long term relationship with that platform the driver goes online when she's ready to start driving then once the driver is online has to make a decision should I accept this trip or should I hold out for a more lucrative trip okay she says yes she's driving - the rider is on trip driving the rider to point B and then has a long-term relationship with that platform each journey here has a piece of pricing and matching and long-term incentives tech associated with it so it starts with supply and demand pricing so in your shopping stage in your online stage you have to pitch to supply and demand with this is how much this trip costs so this is how much we think you're going to make in one hour of driving in downtown San Francisco then there's recommendations I think this is where you should be driving we think it's going to surge at this time maybe you should go down to the financial district then is the actual match user request here is supply how do I make the most Network optimized decision then you kick in rebalancing mechanisms okay are there too many requests are there not enough drivers on the street do we need to start surging how much should we search once the trip is over what is your long-term loyalty and customer buyback of what kind of promotions and incentives am i offering both drivers and riders to keep them on my platform and keep them away from my competitors and then quality control is this a well reviewed driver if the driver is carrying a 2 star review what's wrong should they be on my platform are there an unsafe driver how do I read out those bad experiences on my platform and all of these usually Matt one on one to a tech team within the market based business again at some level of maturity so I'll double click into a few of these pieces I'm not going to be able to speak in very much depth about each one but I'll call out the key parts first I want to talk about matchmaking the underlying principle of matchmaking a service marketplace business is matchmaking is unopp in yin ated it's not biased towards any one single actor on the network and it's optimized now matchmaking can be of two formats obviously there's recommendation based matches so I'm going on Airbnb I want to take a trip to Tahoe here are my dates here are the number of people who are going with me what are the top options on your network you can give me for what you think my price point is you're getting matchmaking right is just essentially getting search right how do i how do i surface the most relevant search results based on past experiences on each of these platforms for this particular person to increase the probability that they will convert to me what's exciting is real-time matching so I put in a request I have 10 people 10 drivers in the five Lock radius who do I match you with what are my ETA expectations how do I make that decision in 20 seconds and then having the driver on the other end accept that request I specifically call out that matchmaking has to be network optimized and I think it's let's walk through an example of why it has to be network optimized so consider the scenario on my right which is a purely greedy algorithm where the passenger number one comes onto the platform first and she says okay I'm going looking to go home give me the closest possible car something cool there is a car that's one minute away I will give you that car passenger number two shows on the platform five seconds later passenger two now wants the closest possible car but the closest possible car to passenger two is this car five minutes away I made the most optimal decisions for each given passenger but if I add up all the ETS on my network I have an aggregate six-minute ETA in a network optimized algorithm you're looking for the best possible values that minimize network level eta is in this case so passenger number one shows up first and passenger number two shows up second but instead of making greedy decisions for each given passenger at any given time I'm looking at my network and essentially solving a giant matching algorithm across all the different nodes in that snapshot of the world and in this case I give two minute ETS to both my passengers yes passenger number one got one additional minute of waiting but passenger number two her waiting shrunk by three minutes giving me four minutes at a network level and imagine ubers systems and lyft systems doing this every every second every five seconds at citywide levels so you're solving this algorithm in very short time spans for millions of data points of drivers and riders at any given spot to make this network optimize decision and the reason this algorithm has to be uh opinionated is I cannot say let's suppose I really like Rider one rider one is an extremely loyal customer and she takes ten trips a week I want to get rider one a minute ETA each time and the reason I can't do that is my rider number two who might be taking trip number two on uber ever in her lifetime it's going to get it should eat here so any kind of greedy decisions at an real-time match making algorithm has a real cost associated with for all the other participants in that network at at any given time and therefore matchmaking has to be uh opinionated and network optimized now I will flip to talk about pricing because I think pricing here is equally interesting because remember that pricing here is not just pricing your uber trip for the rider but pricing heuristic finding that equilibrium between demand and supply so demand in this case the rider is saying hey is this the best return on investment for the dollar amount and giving into this transaction am I going to pay eight dollars for this trip is it worth it could I walk could I take the bus like this make sense the supply or the driver is thinking is this transaction the best return on investment for my time I'm going to spend some time driving to the rider I'm going to spend a bunch of time on trip and then I'm going to be taken somewhere where I may or may not find a subsequent trip is this a good investment I don't know and you have to make that transaction by transaction level decision so pricing in a marketplace is not simply setting prices for one part of the marketplace over another part of the marketplace it's about finding the right price point where this trade-off makes sense so that this transaction is converted from a shopping session to a completed transaction there are some things that kind of put their foot on the scale in one way or the other so seasonality for example it's raining it's really cold it's snowing demands elasticity to pay any given price is now called into question because the weather sucks similarly on the supply side there are competitors I could be driving for lyft I could be delivering food under - I could be shopping on is this particular platform worth my time compared to the competing options I have on the table and of course there are other software things like what is the brand associated with any given platform versus what is the user experience of using this app and all the stuff takes place some weight in demand versus supply pricing but inherently at the end of the day pricing has to solve that real-time decision of is this good return on investment for each side of the marketplace and finding that perfect price point so to do that to find that perfect price point I need to go figure out how much things cost and then how much would people pay for something so on the cost side I'd like you guys to think about let's say an uber pool trip you might have taken recently so if you take a $10 over pool trip you're paying the driver $10 but you don't get matched there's no one else going in the same direction as you you're comfortable you're not sharing the car with anyone no detours that's great from your perspective from ubers perspective I payed driver 1 $10 and driver 2 $10 and I might not be charging writers that much money if this trip were being were to be matched and if I went from pick up one to pick up two to destination one to destination two instead of paying to drivers and aggregate $20 I'm paying one driver $16 in order for me to be able to set accurate rider prices for uber pool I need to be able to predict that I think this trip will match and because this trip will match I think it's going to cost $16 and therefore this rider is occupying half of that $16 trip and therefore I should charge that rider eight dollars as opposed to ten dollars in an unmatched trip being able to have that kind of predictive power before the session is actually converted and therefore before I can actually tell you if this trip is going to be matched is a really important part of estimating how much something will cost and using that to set prices the second part of using historical data to set prices is getting a sense of how much is something worth so imagine if I was an Airbnb PM I would start with let's say an a/b experiment where I say okay I'm going to put this home I a 250 dollar a night price point I have another treatment at $200 a night and a third treatment one hundred and fifty dollars a night what is that price point at which certain percentage of my customers start converting so that I'm not giving things away for free and I'm not taking a loss to myself but I'm finding the right price point between how much something is worth or demand and how much something will cost and using that to set customer prices all right so we talked about matching and we talked about pricing the third part and an extremely important part especially for full-stack service providing marketplaces is quality control so everyone's had that awful experience with a really rude ride-sharing driver or a really bad experience in an air B&B where your room smells bad and your your hosts wouldn't pick up their phone and these kinds of things are sticky and stick with the reputation of the platform so all the horror stories of like people trashing their Airbnb zoar women having safety issues and ride-sharing those kinds of things stick with the brand and really hurt the reputation associated with those brands and those are p99 cases obviously those 1% of cases or something goes wrong really sticks on that brands name and customers associate that kind of service with the brand even though an employee of the brand was not providing the service so quality control mechanisms are like a very critical part of informing future decisions so for example each platform has their favorite version of how do I call out the best performers on my platform so you must have seen Airbnb is super hosts or lyft and uber premium drivers like highly rated drivers door - is recommended restaurants all of these are the platforms way of saying hey I trust this player on my platform and their quality their service quality is extremely high here is my stamp against this person early on in the lifetime of a marketplace business those things can be done manually but at a critical mass those things have to start being automated so looking at driver reviews looking at driver five-star ratings using that to inform pricing matching and then transactional decisions are really important so that that virtuous cycle of quality control continues to persist to keep your best actors in the platform and avoid situations like this so that you're reading out the bad players as soon as these kinds of activities start to rear their ugly heads so uber lyft Airbnb all have teams in place that are essentially running a bunch of machine learning and sending a lot of ops power behind reading reviews quality controlling five-star ratings looking at drivers whose ratings have dropped off suddenly and using that to make sure these participants are weeded out of the system before a bad incident happens and lastly rebalancing mechanisms so this is the last thing I want to double click into because I think it's really interesting which is how do we make sure in real time that the marketplace is healthy that demand and supply is more or less balanced such that we are ensuring a certain quality of making so imagine you are at a Beyonce concert at 10 p.m. century feeling at 10 p.m. thousands of people are getting out of this concert hall so rider requests suddenly shoot up when rider requests shoot up driver utilization in that neighborhood shoots up because now all the empty drivers are being sucked into century if you think as soon as that starts to happen eta is start sucking because now you have to pull drivers from further and further away to start fulfilling those requests when ETA starts sucking rider requests per session start tanking riders start canceling EJ's are not worth it it's not worth their time to wait they'd rather take the bus it's just chaos so write a cancellations go up and then driver cancellations go up because the driver doesn't want to have to drive 25 minutes to that concert because he's gonna spend 15 minutes finding their passenger was probably drunk so what unbalance looks like is just a terrible set of cascading events that complete takes a healthy marketplace and in a 10 or 15 minute time nine makes it completely unhealthy because now we are just in this cycle of requests and cancels and requests cancels and cancellations are really expensive for the network because cancellation is essentially a driver accepting a trip making progress towards the trip and then two or three minutes into the trip now having to be cancelled and not being paid for their effort and that is just unhealthy for the health of the marketplace because now the driver has made progress that they were not paid for alright so now let's say surge was invented what does this identical scenario look like where search exists and I want to say that search gets a really bad name because everyone's like oh we're a surging lifted surging they're just kind of their price gauging me they just want to get more money out of me what that kind of narrative misses is that surge is not so much trying to get riders to pay more but surge is essentially defensemen against an unbalanced marketplace so let's say Ryder requests at 10:00 p.m. go up because everyone's exiting the Beyonce concert driver utilization goes up surge kicks in so surge looks at the real estate real time state of that market place in 10 minutes and says okay Ryder request is up all my drivers are utilized something's wrong I'm going to kick in search when search pricing kicks in Ryder's ETS go down immediately because the number of riders are requesting trips at those search prices gets dramatically cut down because willingness to pay has dropped all right so now sir just kicked in all those offline drivers were sitting on their couch are like oh I gotta get a century fueling its 3 X surge right now it's gonna be totally worth my time so online drivers it dramatically increases in volume Ryder ET is still down because a bunch of these drivers have just showed up online all right so now write a request for session stable because you're not seeing a ton of cancellations riders are coming onto the platform looking at the price and now deciding whether or not to convert and now that the market has stabilized surge auto corrects says okay now I've entered a healthy state of supply and demand things are looking good cancellations are looking stable I don't need to search anymore I'm gonna shut down in both these scenarios riders entered the ecosystem and completely overwhelmed it and overwhelmed the marketplace but when surge kicks in and when surges doing its job well it's able to quickly in some cases in matters of minutes correct the state of that marketplace so that riders and drivers are able to have continued to have a good experience on the platform without those p90 cases where people get like 20 minute eight years alright so we've talked about matching pricing quality control and surge I would say those are the four kind of most important and key from a technical perspective the key challenging tech components of running a market based business but now I want to flip and talk a little bit about what a p.m. does in the marketplace business and how is the p.m. adding value to marketplace business all right so a typical marketplace PM is a people person yes mostly grace for that stakeholder management more is more accurate you will see that these marketplace PMS the ones I have interacted with who are really successful and impressive a 50 percent data scientists like what is going on how can I D bug this issue what experiment should I run and 50% ops like what do I need to do to just get down there I get my hands dirty you roll up my sleeves and get things working again and you want that balance of like 50% science and 50% just like execution and getting your hands dirty a market-based p.m. owns the bottom line of her business areas you're not optimizing for features you're not optimizing for button clicks you're optimizing for the business unit that you're responsible for and lastly you can deal with local markets specific complexity laws and regulations in each market you have your business and look very different how are you optimizing for a tech that is as common and unified as possible but also complies with local competitive and local regulatory requirements so a lot of people especially when I take interviews and prep other people for interviews will ask me okay well I have been a software PM can I switch into being a marketplace PM do I need a business degree yours the high level chart I would draw for software pm's looking to interviewer transition into a marketplace PM role so as a software p.m. for people in the room and I know some of your PM's as a software PM your key stakeholders are your engineers your designers PMS from other teams but as a marketplace PM your stakeholders it's just ten xed so you're looking at local ops in every market you operate in you're talking to finance you're talking to me go you're talking to policy data scientists are your best friends and you're constantly making sure the decisions you're making are applicable to make the 400 markets you operate in software PM's are responsible for product metrics like what is my conversion funnel how many people are clicking on this page how many sales have we clocked in and so forth market based PMS are looking for business metrics like what are my Billings what I meant net inflows what do my profitability numbers say and what do they say it at a market level your service area for a software PM is global you're shifting your software everywhere in the world and then you have localized features like Windows 7 looks the same everywhere but obviously you localize the feature to that market as a market-based PM your tech and your your business is extremely hyper local but you're building global features where possible and again as a software PM you own features as a market face p.m. he own business articles alright so now that we talked a little bit about what market-based p.m. sorry and I hope I given you guys a sense of what that entails I want to talk briefly about a marketplace product interview scenario how much time do I have ok all right so let's do sample question 1 and I'm borrowing from some of the interview prep I've personally done in some of the interview prep our interviews I've taken in the past at uber so let's walk through sample question 1 so let's say ETS for uber X in Bangalore have suddenly shot through the roof in the last week you're the PM for this market you're you're the PM for this ETA feature community bug what's going on and come up with a remediation plan so I tell you first what the bad answer sound like so the bad answers and they look at this problem space saying hmm I wonder what's going on I wonder if our ETA models are broken I would look at any service outages are some of our servers down what's going on like they look for bad answers look for tech solutions or tech features a good marketplace answer with will first right off the bat be aware that there are two sides of any given marketplace so in the case of this ETA question in Bangalore the first thing your head should go to is okay here's our up does that mean that our supply utilization has gone down are there fewer drivers on the streets in Bangalore this week than there were last week if yes why is that if not let's look at driver acceptance rates our fewer drivers accepting trips than before if yes then why is that okay does it mean that my incentive dollars have dropped or have you made a change to the pricing algorithm or have you given out a rider promotion where a demand has ten xed such that we've flooded the market with more trips than we were used to last week or has something weird happened where our matching algorithm isn't working so instead of doing Network optimize decision-making am making greedy decisions those lines of thinking indicate to me that the candidate is not just looking at this as a tech problem with an engineer hat on but as a marketplace PM who is the business owner aware of both sides of the marketplace and how they're interacting with one another second a good answer to that question calls out the balance between network effects and unit level optimization so when I'm looking for solutions for sample question one I am NOT looking for answers around how do I make that one ETA faster or how do I make it is shorter for people in that one neighborhood but I'm looking for at a network level what is broken about this market that causes the ETS to be degraded as opposed to user level or rider level decision making third talk about sustainable solutions I think that's healthy practice in any p.m. interview and for focus on testable hypotheses so what can I be test really fast so good answers I've seen four sample question one before have indicated things like what would happen if let's say we made accept the acceptance screen on the driver app in Bangalore look a little bit different such that you could see how much money you were going to make on this trip would that change driver accept rates behavior and then a B those different screens to see if accept rate smoked so focus on very testable hypotheses that are predicated on one part of the marketplace or the other making a certain set of decisions so let's repeat this exercise for sample question - so you are a p.m. on Remo automated cars just came online you have as the p.m. and access to a fleet of a thousand vehicles you're responsible for what the go-to-market plan is how do you get these thousand fleets on the market how do you convert them to a ride-sharing business bad answers here until things like I'm going to make a beautiful app and I'm going to send all these fleets on the streets of San Francisco and as people request cars on this app I am going to send the closest possible car those are very software centric answers a good marketplace p.m. candidate starts from looking at both sides of the marketplace and essentially does a business evaluation what are the scenarios which I want to serve with launching this product am i going off to commute am i going after food delivery am i going after I don't know school buses then that candidate hits picks one of these very specific use cases let's say we're going after commute at a.m. rush in San Francisco sounds good - what are the existing solutions that San Francisco currently has for commute and what additional value-add am i adding three what is the cost to serve these fleets and am I going to be able to make the unit economics work so that each trip is priced at a certain point where I'm making money on each of these trips for what are what is my matching and pricing algorithm going to look like where am I going to deploy my fleets where are they going to be parked so that they're quickly able to respond to requests five what are my KPIs how many requests per car how many trips per hour each of these fleet members that are going to fulfill what are my KPI is how do I ensure I meet those KPIs and then kind of starting to double click into demand then starting to look at lifetime value how many repeat trips am I going to look for at a per passenger level what is my price point how do I make sure that my self-driving vehicles are always working what happens when one of them breaks down or runs out of gas like those are the kinds of considerations that are more business centric that you're looking for in a market-based p.m. interview and bad answers and especially answers that are coming from people who are used to more classic software PM interviews usually entail user experience or usually Intel features that kind of often misses the point all right so I've talked a little bit about market based product interviews and I'm happy to take questions later if you guys have more specific tips you're looking for about market based product interviews but I'm at the end of the prepared content I have which is I would recommend this reading list for anyone who's interested interested in marketplaces so here are the essays I personally used to prep myself for my uber interview and they event a really long way in particular I'll call out number three which is Alvin Roth's who gets what he's an he's a Nobel prize-winning economists or physicist one of the two and he's written a great book on how market based businesses at scale are successful you