Drivers are dropping out of a city on Lyft. How do you figure out what's going on?

I will start with getting clarity on what dropping out means and what metric is driving that. Is it:

• Daily Active Usage (DAU)
• Ride acceptance rate
• Driver availability during busy hours

Also seek clarity on the specifics of the city. Is it a large city or small city? Is it seasonally abnormal or just around the specific time. Was it a gradual drop or a sudden drop?

From an investigation standpoint I would split the investigation between internal factors and external factors. Internal factors will include:

Technical issues:

• Any driver issues with the app.
• Any new software update launched such as UX changes for a driver that were piloted here.
• Bugs with the driver experience that multiple users have complained about
• As it appears to be specific to a geographical region could it be anything with regards to geocaching.
• Any server side software issues that were first piloted in this region.
• Any infrastructure outages or outage in a specific region. 

Non-technical issues:

• Any policy changes on driver compensation announced by Lyft such as revenue percentage or tipping.
• Any bad press around Lyft
• Any changes in passenger behavior such as fewer rides being ordered that is having a domino effect on the driver side.

• Technical
  • Any outages for an infrastructure provider such as AWS in the region.
  • Any outages for mobile network provider such as AT&T or T-mobile.
• Non-technical
  • Any activity by competitors such as Uber that is causing drivers to migrate.
  • Any local regulations that are impacting driver availability.
    • Example - Surge pricing in NYC.
    • Example - New regulations around gig work in cities like Seattle.

Drivers are dropping out of a city on Lyft. How do you figure out what's going on?

Ok, so my general approach to this question would be to find out about the reason by asking you question to come up a reason. There could be different reasons for this problem.

So, first thing is clarify that I understand the problem right. When we say drivers are dropping out, how did you find this out? What is the metric which is getting affected here and since how long is this happening?

-              Communicated by the support team.

-              Number of registrations have decreased.

-              Assume that it is happening since past 6 months.

Also, what do we mean by drivers dropping out?

-              They could be not opening the app for taking rides

-              They are de registering and deleting the app

-              No new drivers are getting enrolled effectively decreasing the number of drivers on the app.

-              Is there a seasonality pattern here? has this happened in the previous years around the same time?

Ok, so, my approach to this issue would be two folds – one will be to establish a channel of communication with relevant teams to see if anyone else has more information about it. – Like talk to the support team and get data on driver complaints or concerns raised. I would like to analyse that data to see, if any insight can be gained out of it.

At the same time I would like to find out the reasons by eliminating possible causes

So, first I will check internally,

1.        Has there been any change in the drivers compensation  – No

2.        Is there a change in the communication done with drivers ? like the kind of training programs/ feedback system/rules to comply? – No

3.        Is there a change in the on-boarding or verification process of lyft for the drivers? - No

So, basically the idea behind asking these questions was to find out possible reasons keeping in mind the value which drivers seek from lyft and  if that has changed in any possible way.

So, If internally everything seems to be fine, we need to look for external reasons –

1.        Is there a new competitor in the market which can be trying to poach the drivers? – yes.

Ok, so this seems to be a possible reason that why we are losing drivers. Now, one approach would be to cheek the competition, find out the package they are offering and then come up with a plan to retain the drivers.

But it is still a good idea to see if there could be some other reason as well? Like verify the external factors –

1.        Any new policy regarding taxi services by the govt

2.        New job enabling plans by the govt

This can also bring up insights on the problem at hand.

A few clarifying questions -

1. What do we mean by dropping out of a city? This could lead to several scenarios -

   1. A particular city, where drivers are no longer interested in driving lyft?

   2. This is happening in a few or all major cities? Where traffic could be a problem and it’s difficult to complete rides and make money?

   3. Competitor (let’s take Uber) has figured out a better incentivisation formula for the drivers. 

   4. A new competitor platform (another ride hailing service) is launched

   5. Macro factors like - govt incentive schemes on unemployment allowances, licence fee etc.

Assumptions - 

1. Time-period - Drop off identified in the last 1 month only, before this the time period was normal

Diving deeper into the data - The deep dive on the data (collection, diagnostic and prescriptive analysis) will vary based on the scenarios mentioned above. So, I would like to start doing my analysis based on the data mentioned below -

1. Number of new driver registrations in the city across ride hailing platforms

2. Number of drivers drop offs in the city

3. Number of driver complaints in the last 3 months to lyft

4. Amount of money made (commissions/fee for service) by drivers in the last month and prior 6 months to analyse the trend -

   1. Decreasing trend would lead to couple of insights on whether the drivers are making money or not

5. In the last month, the number of drivers listed on competitors' platforms in the city. (This can be accessed via market research firms)

6. Launching a survey to both the drivers - 

   1. Retained drivers with the lyft - What is causing them to stick with lyft?

   2. Drivers who abandoned the service 

      1. we need to make sure that at 20-30% of the driver's responses to get enough sample size to be statistically significant

Solution strategy - Based on analysing the above data points and the scenarios, a strategy need to be developed to solve this issue.

A few clarifying questions -

1. What do we mean by dropping out of a city? This could lead to several scenarios -

   1. A particular city, where drivers are no longer interested in driving lyft?

   2. This is happening in a few or all major cities? Where traffic could be a problem and it’s difficult to complete rides and make money?

   3. Competitor (let’s take Uber) has figured out a better incentivisation formula for the drivers. 

   4. A new competitor platform (another ride hailing service) is launched

   5. Macro factors like - govt incentive schemes on unemployment allowances, licence fee etc.

Assumptions - 

1. Time-period - Drop off identified in the last 1 month only, before this the time period was normal

Diving deeper into the data - The deep dive on the data (collection, diagnostic and prescriptive analysis) will vary based on the scenarios mentioned above. So, I would like to start doing my analysis based on the data mentioned below -

1. Number of new driver registrations in the city across ride hailing platforms

2. Number of drivers drop offs in the city

3. Number of driver complaints in the last 3 months to lyft

4. Amount of money made (commissions/fee for service) by drivers in the last month and prior 6 months to analyse the trend -

   1. Decreasing trend would lead to couple of insights on whether the drivers are making money or not

5. In the last month, the number of drivers listed on competitors' platforms in the city. (This can be accessed via market research firms)

6. Launching a survey to both the drivers - 

   1. Retained drivers with the lyft - What is causing them to stick with lyft?

   2. Drivers who abandoned the service 

      1. we need to make sure that at 20-30% of the driver's responses to get enough sample size to be statistically significant

Solution strategy - Based on analysing the above data points and the scenarios, a strategy need to be developed to solve this issue.

For the context and to clarify the ask : Is this happening only in a specific city and in a particular geo-location?

I am assuming this to be in a small city and in the US. 

Moving towards building the hypothesis:

1. Has there been any changes to the data pipeline captured and insights? No changes

2. Is this a Trend seen or overnight drivers dropping out. - Trend seen

3. Any Tech or UI changes which is instigating the drivers to drop out- All good with Tech and UI, no changes.

4. Any changes to incentives for drivers in this particular city - No changes

5. Competitor launch and caused the stirred - No 

6. Any city council specific regulations (new/changes?) - No

7. Any changes in passenger behaviours? Any change in the Demand side? - No, it is healthy

8. Are Drivers dropping off the platform or dropping out of this small city? - Drivers are not dropping off the platform but only out of this small city.

9. Have you seen the drivers hopping onto the platform from another adjacent city that is bigger in radius, better demand? - Yes

My hypothesis would be : Drivers are dropping off from this small city and catering through the adjacent bigger city, maybe because the incentives and perks are more lucrative and healthier demand (more rides) and better job prospects.

Here is my approach, first I would like to list out the assumptions which are ideally clarifying questions in a real interview, and then start solving them.

Assumptions:
1. The drivers are dropping from one particular city and the rest of the towns the drivers are still using Lyft and all the metrics related to usage are going normal.
2. Dropping off means stopping using Lyft as a whole ( If it is at a particular part of the user journey while using the app I think this can be the same across the other cities).

To find the reason I would first like to isolate the period in which this has happened and then understand if any changes during that period have affected the driver's decision.

Internal Issue: 
1. I will list down the changes made in the app or policy during or around that period and see if those changes can affect the behavior specific to that city. Again I want to reiterate my assumption that app changes can be universal to all the cities. 
An example of how policy change can affect the driver's decision is :
If Lyft has reduced the tariff for time spent on the ride drivers from cities where there is high traffic can find that not lucrative and may go for alternatives.

2. I will look if there is any change in customer sentiment on drivers through reviews in that time and will try to find the reason for the change in driver behavior.
3.  I will look for if there are any complaints from the drivers on company staff specific to that city.

External Issues: 
External Issues can either put drivers out of business or make them look for alternatives so I would check if the actual number of drivers has dropped in the city. 

If the number of drivers has dropped in the city:
 I would look for the reasons for it this can be due to some changes from the government side or some other campaign which is impacting our business or an increase in crime among drivers in that area. 
For example, the city's adoption of public transport might have increased due to an increase in routes or other reasons. 
There might be a movement running for citizens to adopt cycling or self-transport 
There can be an increase in carpooling culture ( this can also be through a competitor)
A new employment opportunity/ public welfare scheme was created which might have pushed the drivers to leave their jobs.

If the driver market is intact:
I would look for reasons for it, this can be due to our competitor or an unsolved long-term problem for the drivers in that city. 

For example, Our competitor might have launched a city-specific pilot which might have motivated either riders or drivers to switch platforms.
Some local competitors might have created an app specific to the city.
Drivers unions might have boycotted the app for not solving long-standing grievances.

In this solution, I have tried to understand the driver's motivations and address the core reasons for drivers dropping off like a drop in demand on the app, finding an alternative, or not finding it lucrative to isolate the issues. Most of these would include direct talks with drivers as the final step.

@bijan please review this whenever you have time as I believe this answer is differently structured from the popular answers and  I am doubting the approach
 

Context

Lyft is a ride sharing app where users can request a ride and drivers will fullfill it. Lyft's mission as a company is to reconnect people through transportation and bring communities closer together. Lyft needs drivers to do this and that is why we are concerned. 

Clarification

Before we dive into figuring out what is going on, let's make sure we understand the question correctly. 

1. Which city is this?
2. What do we mean drivers are dropping out? What metric or set of metrics are we using to determine this?
3. What is the magnitude of the decrease in drivers? 2% is fairly different than 20%. Going to assume this is statistically significant.
4. Going to make the assumption that our data is accurate.

1. Around what timeframe did we observe drivers dropping off? Did it happen gradually or was there a sudden drop off?
2. Is the drop off uniform across the city or are there certain areas or neighbordhoods were it is either more or less severe?
3. Is this drop off uniform across user segments and demographics? For example, are new drivers dropping off faster than experienced drivers? Lyft Pool vs Lyft Luxe drivers?
4. Is there any discrepancy between iOS vs Android? What about particular OS versions, or versions of our own app?

Internal

1. Could there be any recent product changes that are negatively impacting drivers? Maybe we are beta testing a new ride matching algorithm in this particular city and our drivers are having to drive farther to pick up riders.
2. Are there any bugs in the app that could be causing drivers to drop off? The bug would need to be city specific so it could be in an app version being beta tested in the city or something specific to the city itself. For example, maybe we recently updated our city street maps and the data for this particular city was old resulting in confusing directions for drivers.
3. Are we running any city specific sign up campaigns? Something like a earn double on your first 30 rides would cause an initial uptick in the number of drivers but then result in drop off overtime.

External

1. Competition - Have any competitors increased their marketing efforts in the area? This would draw drivers in the city away from our platform.
2. Lack of demand - Has the number of riders dropped off? This would result in less frequent rides and a longer distance to pick up riders for our drivers.
3. Driver classification changes - Has there been any local ordinance in regards to whether or not drivers are classified as contractors or W2 employees? This would affect their take home pay and desire to drive for a ride sharing app.
4. Crime Spree / Decreased Driver Safety - Has there been an uptick in muggings or crime in the city? If drivers don't feel safe they are less inclined to drive.
5. Inclement weather? Wildfires, hurricane? - Could there have been a local natural disaster that would cause people to flee from the area?

I have specified different branches that the problem statement could be lead to during the interview below:

Clarifications:

1. What is the definition of the drop ? - Assume -number of drivers marking themselves as available has come down, consequently overall number of available hours has come down.
2. Any  recent change in data definition ?
3. Is it a sudden sudden drop / gradual drop, Is the drop expected as per regular volatility
4. Confirmation of driver user journey
   1. Onboarding & approval
   2. Driver app install
   3. Login
   4. Mark as available
   5. Receive new ride alert - accept & Complete ride / Cancel ride, ignore / reject alert
   6. Mark as not available
   7. Log out
   8. uninstall

1. Any change in data pipeline , can we rule out all issues related to data sanity?
2. Where is the driver drop offs ? Lower logins - Yes
3. Lower acceptance for rides -Yes

Sanity questions

1. Any increase in crash reported in app ? - No
2. Any new feature rollout in the platforms?
3. Funnel wise drop offs as per user journey?
4. Check performance reports ( any issues in health / latency metrics limiting usage?)

Product usage leg

1. Platform specific dropoffs / version specific -  Iphone vs Android
2. Driver product usage funnel analysis ?

Localized issues ?

1. New competition in the area
2. Local campaigns by competitors ?
3. Road blockages / Traffic issues leading to reduced driver interest ?
4. Local regulations implemented by city ?Network related issues ?
5. Any specific micromarket / location ?
6. PR issues ? / accidents / news ? Any negative social sentiment ?

Demand related issues ?

1. Reduction in ride requests ?
2. Decrease or increase  in distance per ride

3. Reduction in specific category - Pooled rides,Large capacity cars / PremiumSmall capacity cars

Driver related issues ?

1. Car type
2. Male / Female ? - Any safety issue ?
3. Retention related issues / incentive pay misses ? 
4. Incentive changes / Price reduction ?
5. Demographics
   1. High tenure
   2. Low tenure
   3. New signups

Competition behaviour

1. Price / Incentive reduction
2. New competitor launch ?
3. Campaigns ? / one time bonuses ?

How I understand the Lyft app in this specific case. We are talking about the ridesharing experience, where drivers help riders to get from point A to point B. Obviously, we are not talking about rentals, e-bikes, or scooters. 

Clarifying questions:

• What do you mean by drivers are dropping out? Is this something specific?

• By ‘Driver” do we mean new drivers or existing drivers?

• Are we talking about one specific city?

• Do we see this happening across all of the platforms?

• Is this happening for a while or this dropout is sudden?

Internal and External factors

Internal:

1. Driver related

New drivers:

• Are there any changes in the number of requests to join Lyft?

• Do we see any changes in the flow of submitting a new driver?

• Do we see any changes in the process of approving new drivers?

• Do we know anything about the success of the first ride? 

Existing drivers:

• Do we know any information about incentives changes for drivers?

• Are there any data/changes about avg distance to the rider in this city?

• Do we know anything about launching marketing campaigns for the drivers in this city?

• Are there any changes in payment procedures for drivers? (payment success)

• Do we see this drop across all types of drivers (electric, gas, economy, premium)

• Are there any rollouts in the updating maps or pickup points? Does the map provider work as expected?

• Do we have any drivers’ feedback?

2. User related:

   • Do we have any information about the demand of the users? Is it the same? Do we see a decrease in the number of users in this city?

   • Do we know anything about experiments on the user's side? (Price change, user-driver connecting algorithms, etc.)

   • Do we see an increase in the cancelation from the user’s side?

   • Do we know anything about user satisfaction in this city?

External:

• Are there any new local regulations in this city? ( new taxes for drivers, license regulations)

• Do we know anything about the activity of the competitors in this city? (marketing campaigns, increasing the incentives for drivers, launching new services)

• Do we know anything about season activity? Can we compare existing data to the previous year’s data?

• Are there any celebrations or events in the city?

• Are there any natural disasters in this area? 

• Do we know anything about transportation limitations in this area/city?

I got the answer that a big music festival is still happening in the city and a major part of downtown is closed for drivers. As a result, passengers can’t set the right pickup location, and drivers have to manually contact passengers. As a result, most of the drivers switched to competitors app that was ready for this event. 

Action items to fix:

• I would contact the responsible PM for the mapping service and talk to him about possible ways to fix this for the rest part of the city event.

• If the problem is possible to fix with maps I would contact the marketing team and discuss communication plans for drivers.

• Also, I would suggest finding a representative in the city to avoid this in the future. Also, I would make sure that we have a responsible person who can manage this in the future, as well I would suggest reassuring that communication with city representatives in the other locations are set. This will help to avoid the same problems in the other locations. 

Any Google problem solving interview question starts with clarifying what the interview is looking for. 

Clarification

• By 'dropping out', do you mean the DAU of Lyft drivers who successfully provided rides in this city decreased?  [No, the DAU of Lyft drivers who can accept rides]
• Is this a large city or small city, asking because sometimes Lyft drivers will operate cross the border of Citys [Small city, so cross border operation is possible]
• Is the drop significant, outside of normal fluctuation? [Yes]
• Is the drop a sudden drop or a gradual drop? [Gradual]
• Do we see similar drops in adjascent cities? [No]

• Acquisition problem, where the driver attrition rate looks normal, but new driver recruitment rate dropped in the greater area.
• Engagement problem, where the frequency of drivers becoming available to ride in this city dropped.
• Retention Problem, where drivers are leaving the Lyft platform at a higher than normal rate.

• feature launch or bugs, e.g. causing frequent app crashes in this area.
• Hw outage,e.g. related to load balancer in this city. 
• Policy change, e.g. end of incentive program.

• Demand, e.g. drivers move to other cities due to no business here.
• Environment, e.g. flood, public events, traffic congestion, or even government regulation
• Competition, e.g. Uber announced incentive program in this city.
• PR, e.g. there might be a 'Delete Lyft' compaign going on in this city.

Solution:

Summary

First, I’d ask some clarifying questions:
– What do you mean by “dropping out”? Is it:
. Decrease in new driver sign ups?
. Decrease in returning drivers? (# times opened app per day)
. Decrease in hrs spent driving per day?
– What city is this in? Is there a trend in the metric, or sudden drop? Are there other cities with the same issue around the same time?

Depending on the two answers above, I’d list out the steps + categorize them as:
1) Company related
– Did we make changes to the app?
– Has there been bad press recently?

2) Driver experience
From the perspective of the driver + their motivations for using a particular app, I can think of:
– Money: are they making more money vs other apps ($ made per week) – assumes competitor data is available.
– Money: are we taking more cut from drivers that made them want to quit?
– Safety: have there been reports of drivers getting attacked (ex. this happened in Brazil + France when Taxi unions attacked ubers in the wild)
– UX: do we have qualitative or quantitative info on driver’s experience driving with our app?

3) Competition
– Did competitors recently launch their app in the city?
– Did competitors launch a new program (business model, better treatment, event with better pay rate)
– Supply x Demand health – are there enough requests to keep drivers in the app? Would look at ratio of Drivers to Requests made over period of time.
. Supplier: Drivers, Demand: Riders

Depending on what extra data we get, I’d drill down on those points further.

I assume “dropping out” refers to deactivating such that they do not drive for Lyft anymore. First of all, I would cross-check the source of the red flag data to be sure it is accurate.

Assuming the trend is verified, I’d ask these clarifying questions:
– What % of drivers are deactivating?
– Over what period of time has this deactivation been occurring? When did it begin?
– Did the change occur suddenly or gradually over a period of time?
– What changes occurred at the time the change occurred that might affect app performance? E.g. a code release or vendor change
– Is the change occurring within the entire driver population in the city, or among certain drivers only? e.g. only drivers that signed up in the past 6 months are affected
– What feedback have we heard from the drivers in the affected group? Have they provided reasons for deactivating?
– Have there been similar drops in other Lyft metrics in the area? e.g. has ridership experienced a similar trend?
– Has this trend appeared before in this city or another city? if yes, what was the cause?
– Has there been any server downtime or other errors reported?

Using this information, we want to determine if the problem is *internal* or *external* in nature.
An internal problem could be a bug, server issue, or a new feature that had unintended consequences. An external problem could be related to competition, bad PR, a natural disaster, a firmware change, or new regulatory or legal change.

If the problem is identified as internal, meaning a bug or the like, I would work with the engineering team to isolate the problem and push out a fix. Depending on the severity of the problem, I might ask to interrupt the current sprint and push this out as a hotfix, or if less severe, could include the fix in the next sprint plan.

If the problem is a feature that is causing issues, I’d work with Eng to roll back the feature and if possible, spec out a new feature that does not have a similarly detrimental impact.

If the issue is external, such as a hurricane, change to the law, or new competitor entering the market, I’d work through the normal product development cycle to plan for an address the problem. If the issue is temporary or seasonal, there may not need to be further action taken as the drivers will likely return without further intervention.

Drivers are dropping out of a city, how do you figure out what’s going on?

There are multiple reasons why there could be a drop of drivers. We can break it down into company, customer (both driver and passenger), product, competition, and evengovernment.

• 1) Which city is this specifically? And, how much of a percentage decrease are we talking about?
• 2) Is there some economic reason for this happening in that city that is outside of our control (government is increasing tariffs to be a driver or taxes for a 1099 employee is too much)?
• 3) Have we gotten bad PR?
• 4) Have our business model changed? Have we taken more % of the drivers pay?
• 5) What do you mean by dropping out? Are they deleting the driver app and/or resigning as being a driver? What specific metric are we talking about?
• 6) Is the percentage / # of driver drop similar MoM or YoY (seasonality)?
• 7) Are there recent changes made in the app that make it difficult to drive?
• 8) Are competitors (Uber, etc.) having a stronger play and have the incentive to drive for them and quit Lyft?