What feature would you add to a self-driving car to make it a more attractive purchase?

I wasn't asked this product design question in my Google interview and I didn't read the aswer provided so I was unaware of the clarifying answer about safety concerns before I answered the question.   Feedback welcome....

What feature would you add to self-driving car to make it a more attractive purchase?

I.                    Clarifying questions?

1.       I want to make sure that you and I are on the same page.  When I think of self-driving cars I think of a few types

a.       Completely autonomous self-driving cars where the car does all the driving from random start point A to random destination point B

b.       Completely autonomous self-driving cars where the car does all the driving from fixed  start point A to a set of fixed destination points B (I’m aware that Las Vegas has some Taxi services that will take you from the airport to your hotel. 

c.       Cars that driver assist such as auto stopping, keeping you in the lane, collision avoidance – but will not take you from point A all the way to point B without a driver intervention.

I’d like to focus on the group “a” the Completely autonomous self-driving cars where the car does all the driving from random start point A to random destination point B

2.        When you say more attractive what do you mean?  I'd like to assume that this is open ended.

II.                  Constraints

1.       I’d like to understand if I have any constraints.

a.       Do we have deadlines, or resource constraints that I should be aware of?  Otherwise I’m going to assume that this is car like a Tesla or other major car manufacturer or tech entrant like a Google with sufficient funding and skills to implement even a moonshot idea as long as it remains economically viable to do s. 

b.       I’d like to improve the self-driving car as we understand it today not reinvent the entire car platform, is it OK if I stay inside the framework of car instead of trying to reinvent the concept of a car?  Is that OK? 

III.                 Goals

1.       The stated goal was acquisition of new purchasers

IV.                 Personas

1.       When I think of potential personas of people that want to buy a self-driving car it could be broken down by

a.       Users that want to use drive time for other uses these could include

                                                               i.      Productivity

                                                             ii.      Rest

                                                           iii.      Leisure

b.       Users that believe that self-driving cars are safer

c.       Tech enthusiasts

2.       If we contrast these personas

a.       Other use of time seems really appealing and provides a positive incentive to purchase the car

b.       Safety is also really appealing but there will be strong counter forces those that believe self-driving cars aren’t safe and high safety seems like table stakes for the market.

c.       Tech enthusiasts want the cutting edge and will likely purchase the product already – creating a feature for them won’t move the entire market. 

I’d like to focus on the 1st group – users that want to utilize drive time – I’ve already identified some of those segments.  If we contrast these

a.       Productivity is incredibly useful and impacts a broad set of potential users

b.       Rest – we all need more rest but just having the self-driving feature, seats that recline already go a long way

c.       Leisure – while many folks take a scenic drive it’s a smaller population than those that use the automobile to go somewhere

On balance I think that  Productivity is the biggest impact  – as drivers we already try to make our time productive by taking phone calls and checking emails/text even though we shouldn’t  I’d like to focus on that one.

V.                  User Journey and pain points

1.       There is somewhere you need to be

a.       How long will it take?

b.       What time should the user leave

2.       You get ready and collect your things

a.       You may forget things

3.       You goto your car and get in

4.       You drive to your destination

a.       Even in the passenger seat there is not enough room to work

b.       I don’t always have the same type of connectivity in a car

5.       You arrive at your destination

a.       Can I find parking

b.       How far will I need to walk

c.       Will I remember where I parked

6.       You park

7.       You exit your car

a.       Did I lock it

VI.                Prioritize the pain points.

1.       How long will it take and what time should I leave – these seem like pre car activities and are handled well by Google assistant and maps – low priority

2.       You may forget things - these seem like pre car activities and are handled well by Google assistant and maps– low priority

3.       Even in the passenger seat I don’t have enough room to work – we have all tried to send emails from a car and not having a good work environment makes it really hard. – high priority

4.       Connectivity – we all have dropped calls in the car – this is a high priority but may be outside of the cars control (e.g. rely on an outside network).  – high priority

5.       Parking and parking related activities – its last mile of the trip and particularly important – but it only impacts productivity during those final minutes – medium/low priority when related to productivity

6.       Did I lock it – also important but doesn’t really tie into making you more productive - low priority

We have two high priority items

1.       conducive workspace

2.       connectivity

I’d like to focus on the work-space – it’s in the car and it doesn’t rely on outside factors like the quality of cell reception  Connectivity is important but there are already cars with Wi-Fi networks in car and with 5G much of the connectivity issues will likely self-resolve. 

VII.               Brainstorm

While we got to the brainstorm around the idea of workspace, I’d like to expand the definition for the brainstorm to be a more a conducive work environment

1.       Built in electronics and systems for handling productivity

a.       phone systems,

b.       screens

c.       google assistants for productivity answering emails and text.  These could be used in either drive or self-drive mode. 

2.       On-star for productivity – a personal assistant on call to help you with productivity

3.       Adjustable interior

a.       space that transforms a seat into an office with a desk, power, and data ports

b.       space the transforms into a collaboration space, seats can be turned to face each other modular work surfaces that

4.       Remove distraction mode – blacks out the windows and sets internal lighting for productivity

VIII.            Compare and Contrast the ideas

1.       Built in electronics – much of this tech already exists and it is handled by our phones, and computers.  If we build it into the car our update cycles are likely much longer, and the tech may become obsolete before the car.  Goes counter to tech like android auto where the car is a screen and the 3rd art device is the driver.  Impact low, implementation low (already exists)

2.       On star for productivity – having an on call personal assistant is nice but it creates ongoing labor costs and likely isn’t scalable.  Also there is nothing uniquely self-driving about this feature.  Impact low, implementation low, not scalable.

3.       Adjustable interior – having the right space to really be productive is very valuable.  Creating interiors that are more modular is fierily easy – airplanes already have seats with desks in the work arm rests – adding power and data ports is already existing tech,  some Minivans already have seats that slide, turn and can be easily removed. Impact high, implementation is low.

4.       Remove distraction mode – I know when you get seasick looking out at the horizon is the best thing to do and being below deck is the worst place.  While removing distractions sounds good it might make the drive experience worse.  Some jurisdictions also have tint laws which this feature might violate.  Impact low – implementation low (they already have photoelectric glass that changes tint.

When I look at these ideas a straightforward modification of creating modular space for productivity would have a high impact for the set of users that buy self-driving cars for productivity.  Simple changes like adding power and data ports, a seat that slides back far enough from the steering wheel, an armrest desk would go a long way to turning a self-driving car into a mobile office.  Additional features like rotating seats so the front and back seats can collaborate would take it even further. 

These same features likely make the self-driving car better for users that want to rest and those that just want to watch tv or play games.

IX.                 Define Metrics

1.       If we sell the productivity pack as an add on then we can measure the acquisition rate of users that bought the car for productivity – Percent of cars sold with productivity pack

2.       We can track total cars sold pre productivity pack vs total cars sold post introduction of the productivity pack – this is a poorly designed a/b test because other factors may affect the time sequenced total sales of the car.  While not a clean a/b test it’s still a useful metric.

3.       If the car is connected (IOC “Internet of Car”  - OK that is a bad joke) then we can measure the different configurations that drivers use this is like a car configuration engagement metric that will help us understand which configurations to focus on going forward.

X.                  Limitations

a.       Jurisdictions still need to embrace driverless cars – let alone driverless cars where the driver is working on a laptop or turned around

b.       Looking a screen or being turned around may increase car sickness

XI.                Summary

One of the persona types that would buy a driverless car is those that want to use drive time more productively.  Productivity may mean working, resting or leisure, I looked at working because it impacted the broadest set of users – although we discovered that the solution for workers likely also helps other user types.  I looked at a host of solutions but settled on a easy to implement cabin configurability feature that allowed the driver to use the space more effectively.  The main goal was user acquisition and we can measure this by the number of purchasers that select the productivity option kit, time sequenced total sales of car pre and post productivity option kit, and finally we can measure the engagement with these configurations.  We have known limitations about legality of self-driving cars and a driver that may be out of classic driving positions and some users may experience car sickness.  Even with those limitations drivers are already trying to use their drive time for productivity and a self-driving car that is configurable for productivity would be more atractive to those users. 

My approach to this product design question would be to 1) validate feature objective 2) identify users and pain points that need to be addressed with the feature 3) list potential solutions and 4) discuss MVP of feature to be designed along with trade-offs

Response:

Interviewee:  Can you further elaborate what you meant by 'make it more attractive' purchase 

Interviewer: Users may be afraid and so not keen to purchase by self-driving cars. What product feature can we build to overcome this fear?

Interviewee: To clarify specific user sentiments and objectives in mind for this feature. Can I assume that this feature will be important to the following segment of customers - Users that are not early adopters of new technology, not first followers or in other words can be classified as tech laggards. This customer segment is typically broader than the early adopter set so the feature(s) must appeal to a wider audience set - more mass-market type feature.  

Can we also assume we test this feature in key markets, iterate and adjust before launching across all markets

Here are some assumptions about user sentiment/pain points and hence the feature design principles to be considered ...

1. Insufficient evidence, need actual not just rhetorical proof of safe driving history for such cars. Implies that risk can be alleviated through trusted sources of data - ( example: test trials, industry-specific awards/certifications, etc)

2. Inability to understand complex new tech and adopt new driving styles. Can imply that feature should be easy to understand and learn. Affects perception of how drivers of self-driving cars will be effective on the road overall

3. Loss of control in a self-driving car and that driver is at mercy of tech. Implies feature needs to hand control back to the driver when needed, possibly cater to the individual specific driving needs

Brainstorm possible solutions: 

1. test data and results proving the reliability of self-driving cars are at par if not exceed user-driven cars

2. additional tech safety features: car self-driving mode customizable by user to include specific manual override controls, in-built car safety features ( defaults to manual in event of tech failure, corruption), external safety features such as auto-hazard light enablement etc

3. dedicated self-car lane control - specific to self driving cars that regulates the speed of all cars in this lane, thereby reducing external risk factors - driver errors, car malfunction due to tire puncture etc

Prioritization approach:

As I think of driving user adoption, I am drawn to examples of how inventors in the past accomplished this - example: carriages to adoption of cars, airplane travel etc. Each time a new tech was invented, mass market adoption was possible when trust, safety and control features were demonstrated consistently.  Features like airbags and lane control have been developed to reduce risk of injury and prevent accidents from occurring successfully. With this in mind,  I think it fair to consider feature 2 or 3 as opposed to focusing on generating successful test data ( this can be secondary evidentiary data that not a feature in of itself)

Further feature 3 includes ability to influence and launch dedicated lanes for self-driving cars.. this is a longer term solve potentially requiring co-ordination of product development with local state and road/DMV authorities

Hence, will focus on feature #2:

In terms of ability to develop a testable feature , we can consider car safety features- a car assistant type solution that users are already familiar interacting with that provides an easily adoptable interface to operate and demystify the complex car tech enabling the self-driving car. 

The safety features that default to manual override where each user can specific risk tolerance differently. 

Example: if Risk is calculated on a scale of 0 to 5, if I am risk averse I can set manual override when the probability of risk of an accident is low <1, vs. my spouse might have higher risk tolerance. the car assistant will compute the  risk or probability of error based on an algorithm that uses a variety of critical performance indicators generated by the car in run time ( eg: mechanical performance indicators - fuel levels, air pressure in tires  as well as tech performance indicators - lane control sensors, GPS navigational ability, etc

Trade-offs - the added safety features may be cost-prohibitive to include and not make this suitable for mass-market consumption

To summarize, design a car assistant enabled safety guide that simplifies user training and adoption of car tech. Also enhances the overall driving experience by enabling safety and control features that tailors drive experience to suit each individual risk tolerance