Published May 16, 2023, 5:20 a.m. by Violet Harris
As a data scientist, I often get asked how I use data to improve Google maps. The answer is that I use data to improve the product in a number of ways.
First, I use data to improve the accuracy of the maps. We have a lot of data on where people are searching for things, and we use that data to improve the accuracy of the map.
Second, I use data to improve the usability of the maps. We have a lot of data on how people use the maps, and we use that data to improve the usability of the maps.
Third, I use data to improve the performance of the maps. We have a lot of data on the performance of the maps, and we use that data to improve the performance of the maps.
Fourth, I use data to improve the quality of the maps. We have a lot of data on the quality of the maps, and we use that data to improve the quality of the maps.
Fifth, I use data to improve the coverage of the maps. We have a lot of data on where people are searching for things, and we use that data to improve the coverage of the maps.
Sixth, I use data to improve the scale of the maps. We have a lot of data on the scale of the maps, and we use that data to improve the scale of the maps.
Seventh, I use data to improve the resolution of the maps. We have a lot of data on the resolution of the maps, and we use that data to improve the resolution of the maps.
Eighth, I use data to improve the design of the maps. We have a lot of data on the design of the maps, and we use that data to improve the design of the maps.
Ninth, I use data to improve the experience of using the maps. We have a lot of data on how people use the maps, and we use that data to improve the experience of using the maps.
Tenth, I use data to improve the safety of the maps. We have a lot of data on the safety of the maps, and we use that data to improve the safety of the maps.
You may also like to read about:
hey everyone it's dan the founder of
datantv.com ex google and paypal data
scientists in this video i'd like to
cover a topic based on product data
science into a question and the type of
question is the following how would you
improve a product this is the type of
question that's commonly asked in
product data science interviews this is
about putting yourself in the shoes of a
data scientist and company that's
interviewing you and then thinking about
how would you use data and extract
insights that can be meaningful in the
form of hey you know what i have an idea
for a new future or enhancement of a
current future that can help the product
grow
so let's
do a deep dive in terms of how do you
approach this type of problem
um so
i just want to first of all mention that
this is the type of problem because
it's um it's about how would you improve
a product and so you would have a
multiple companies asking similar
questions like this for instance google
might ask a question about you know how
would you improve um google maps and
this is a fairly a fair game sort of
question especially if you're
interviewing for the product analyst
role at google and uber could also ask
questions about you know how would you
improve the user experience on ubereats
you know
the difference is the key word but the
essence of what is the main aim of this
um this type of question is pretty much
the same it's about improving the
product essentially and then the meta
company or formerly known as facebook
might ask a question about should
create stories and so you can see that
this is also you know you can see that
they might also ask you a question about
hey you know what um we have an idea for
a new future should we actually build
this or not right um but regardless of
whether they actually mention a specific
product that they are or featured that
they want to build out the essence of
how you would approach this type of
problem is pretty much the same so i
want to mention a framework a five-step
framework that is useful in terms of
approaching this type of problem and so
the first thing is to define the goal of
the problem
so this is really where you bring out
you know what is the business objective
and what is the product experience of
this um you know of this type of problem
right and then the step two is walk
through the user journey so this is
really where you do like a high
you know in-depth details in terms of
what is the typical user journey you
know what are the types of users
that use this platform and what's the
user journey from the beginning to the
end right
as soon as they land on the website with
the platform you know what kind of tasks
can they achieve and what sort of
satisfaction um experiences do they do
they get out of this right um and also
talk about the monetization aspect of it
as well um and all of these are
basically a stepping stone for you to
think about
um you know what are ideas that you can
definitely propose to the interviewer
but also at the same time you're
demonstrating to the interviewer that
you have product knowledge about the
company that you're interviewing for and
that is really important
and then the third thing is to
brainstorm two to three ideas that can
be really helpful right so
based on the user experience of the
current platform or the future um you
know can you think of ways to enhance it
right um the last thing you want to do
is mention something that's already
there
and that can be somewhat costly because
the interviewer might think you haven't
really done the research
in terms of the latest sort of user
experience of the
of the product
um so the the third step is to
brainstorm two to three ideas um and
then
the fourth step is to validate uh ideas
using data so
um the last thing that you want to do is
just basically say hey you know what i
have an idea for building a new future
but you're not really able to back it up
with data and so this is where you kind
of put yourself in the shoes of a data
scientist and imagine if you have all
kinds of data right and obviously you
can clarify with the interviewer and ask
the person hey um can i use any sort of
data source or are there particular data
sources i'm sort of restricted to right
and then think about you know what kind
of user events can you
um look into that suggest that users
want this right um and this is about
finding proxy behaviors that seem to
convey us a certain intent and i'll talk
a little bit more about how this works
um as of course like as i go through an
actual illustration in terms of how do
you approach this type of problem
and then the last part is you know now
that you have two to three ideas and
you're able to validate that hey you
know people are actually using you know
wanting to use this um you have to now
prioritize um the ideas based on the one
that the product team should focus on
right and this is really about putting
yourself in the shoes of a product
manager and then kind of thinking about
you know how would you actually
um among these three ideas which one do
you think has the highest sort of ri
um you know with kind of minimum effort
right and so this is about you being
able to prioritize you know what to
focus on and that's a pretty typical
behavior among uh you know
responsibility among data scientists um
you know especially when you have all
these competing ideas what do you focus
on right and that's what the interviewer
wants to see
um so we're going to go ahead and
actually
talk through this type of problem by
going through this exercise um how would
you improve google maps and
you know why don't you actually just
pause the video for just couple seconds
and just kind of think about how you
might respond to this question right
imagine if you're if you if you're a
candidate for the product analyst role
at google
or if you're interviewing for any sort
of product data science interview in
general and you have a product sense
interview coming up
um take take a moment right you know it
could be a couple seconds it could be a
minute or five minutes and just use the
google maps open up the google maps and
just kind of go through the user
experience and think about hey you know
what i wish there were these features it
can be really helpful that would make my
life so much easier right and then think
about how you respond to this as though
you were an actual candidate um so
pause the video and then you know resume
all right so let's
do a deep dive let's apply the framework
that i mentioned in this type of problem
right so the first thing is to really
define the goal of the problem and so
you want to what you want to definitely
want want to walk through um kind of
demonstrate you know some knowledge
about hey you know what i i understand
what this problem is asking right um i
often see as i you know coach um
candidates for interviews
i might ask questions about you know how
do you measure success for you know
platform or how do you um
you know validate whether an idea should
be whether products should be built or
not right
and
the kennedy often have these clarifying
questions and it's really important to
have clarifying questions but at the
same time it's important to be somewhat
intelligent in terms of what questions
you're asking right like you don't want
to ask questions about um you don't want
to ask a clarifying questions you know
oh what what do you mean by success or
how do you want me to measure success
when that is a very question that that
the interviewer is asking and i've often
seen this type of clarifying question as
in like um you know how would you like
um
you know what do you mean by improving
google maps right um
but you don't want to you don't want to
ask that when that's
basically what the interviewer is asking
right um rather the way to sort of frame
this is personal make start with some
assumptions and then clarify to see
whether you're on the right track so you
can kind of lay out a couple things
right so um if i was a candidate you
know responding to this my opening
response was
would be something on the lines of you
know thank you for this question um you
know i think this is very interesting
and very important because um improving
google maps is about really focusing on
three different areas or combination or
maybe one of these areas and
i know that improving google maps can be
in the form of you know improving the
user experience so maybe helping them
achieve a certain task um you know much
quicker it could be about helping to
promote businesses and obviously
businesses can promote themselves and
google maps um by dropping
by by letting users know that hey you
know this business exists on a map but
in a form like a pen and um like a
specific location
and then the last thing is you know
maybe it's it might be in the form of
helping google um google maps generate
you know advertisement revenue right
because that's one of the main
advertisement stream for google maps at
any time you go on the app itself and
sometimes if you click a specific
business you might see like an
advertisement banner in the search
result and so whenever user clicks it
that's um you know basically businesses
that are taking out these advertisements
or paying google for every click
um and so
this you know it could be in that form
as well or maybe there might be this
sort of one idea that might help all of
these three aspects right um so it's
important to kind of first of all
establish you know what does it mean to
improve goal improve you know improve a
product here and start that as like
you're clarifying assumptions right like
make these assumptions and ask the
interviewer you know um are there
particular areas you want me to focus on
or do you just want me to kind of do
like a general some exploration and then
conversion to one idea that i would
probably recommend to a product manager
right that's how you demonstrate
expertise you don't start by asking some
sort of general questions and expect the
interviewer to provide hints for you
um so once you have
established this step the next step is
to essentially walk through the user
experience right and this is something
that um
you know even if it's the type of app
that you've often used maybe it's like
uber or instacart um or whatever it is
um it's it's often easy to just kind of
passively go through like what the core
features are
um i've used google maps often and even
for me when i'm doing this sort of
research um you know in the preparation
for the interview i have to pretend as
though like i'm an actual like ux
designer or uh or you know basically
like put it place myself in the mind of
like a product manager and really
deconstruct the app right um so one
thing i recommend is that if you have a
product data science interview coming up
right whether it's for uber
whether it's for um you know google or
whatever it is right um actively use the
app right and
create a word doc and create an outline
of hey um here's how here's what are the
core functionalities of these features
and here's what the user experiences are
from the beginning to the end right and
that is gonna be incredibly helpful that
map in your mind is gonna be incredibly
helpful whenever you need to address
questions like this
um so
um so how i would respond like you know
further build on my response would be
like okay first of all i just want to
walk through the user experience here
because i think this is important as a
stepping stone um for me to think about
what potential ideas there are right um
what ideas could perhaps enhance google
maps now i know that um
at its core there are two types of user
bases so there's the regular users like
myself the commuters who loves to
navigate through uh things right like i
need to go from point a to point b i
need to browse places right like maybe
i'm trying to look for new eateries um
like music venues to go to
um maybe maybe i just want to do like a
specific search right there's like a
search bar i can search like um places
to eat or something like that right so
those are one of the core
functionalities and of course there are
a lot more right um but generally just
follow the rule of three like you don't
want to like overwhelm the uh interview
with so many things and the next thing
you know you're not really ever able to
cover the main things right so just
start by focusing the core
functionalities um and then kind of
provide some details from there right
the other aspect of
of the user
user group could be like businesses
right so businesses can definitely
establish their location on google maps
so they let other people know that hey
um you know this steakhouse exists in
san francisco you know maybe you should
definitely check it out
um and they can also take an
advertisement as well right um you know
as a way to promote um you know maybe
like uh like a particular service or
something like that right
um and i'd often see it like every time
i look up like you know restaurant in
the search banner i often see like
advertisements from like postmates or
ubereats because they obviously have
some partnerships
um you know with these
with these restaurants in terms of
delivering you know food to consumers
right and so they would
take out these advertisements um so
these are the user experience these are
one of the core sort of user experience
on google maps and once you have
demonstrated this you show that into you
show to the interviewer that hey you
know the product knowledge but also at
the same time um you know like this is
also going to help you come up with
really cool ideas
then the the third step is really the
sort of like entrepreneurial side of
things right this is where you have to
like come up with
um ideas that have been really helpful
helpful for
improving google maps and so right off
the bat
so i can think of three ideas right and
and it can really start from your own
personal experience but also it can
start from by saying hey you know um i
would actually do some investigation
data and then kind of extract what
potential behaviors users are expressing
and then from there come up with some
ideas that can help remove these
friction points that they're suffering
through right
um so there's all these different angles
in which how you would approach this but
in terms of like what you do afterwards
it's exactly the same and i will show
you what i mean by this in a second but
basically the three ideas i have are
this so and i really wish google maps
would actually um build it so if anyone
from the google maps team is watching
this um please add these features
um
so the first thing is eta planning so
what i mean by this is like you know how
like when you're searching to like when
you're trying to navigate to a place
like
it shows the eta in real time like the
current moment of what the eta is
but what if you're doing like a plan for
like a trip elsewhere right and you want
to see what typical eta is um given
various time of the day right it's like
i live in san francisco and sometimes i
like to go to santa cruz um and like
some i notice that there's a lot of
variability like sometimes there's
traffic like heavy traffic on like
thursday afternoon i'm like what the
heck is going on like who goes to santa
cruz on thursday afternoon and stuff
like that and
and it kind of makes you wonder like
you know it would be nice if you can
kind of like look at trends right
historical trends of like what the eta
is like pick out a day and a time and
then um and then the location and then
be able to see you know what has been
the general sort of eta for this time at
any point in time right and that could
be really helpful for people trying to
make trips like any certain commute if
especially if you move to like a new
city and you want to do a little bit of
exploration outside of the city or
through the city um you know like that
sort of look up will be incredibly
helpful and i do see that like there's
some aspect of this um on google search
itself whenever you search like a
business um like a restaurant or
something like that they do have like
these bar chart that shows like how busy
um a restaurant typically gets
um so i certainly wouldn't see why like
google maps cannot have something like
that for eta as well and then the second
thing is street parking information so
how often
do you drive to like an urban area like
a city or a new city right and you're
struggling to find a parking spot right
and you don't really want to pay for
like a parking garage because they can
be someone expensive especially in a
city like san francisco where you're
paying like 20 an hour or something like
that you know
um and so a person like me would
typically look for street parking but
then like when you look it up on google
maps like there's no information about
like what are areas where i could do
street park water areas i can't if i
knew that information in advance i would
probably go to that street right
otherwise i have to drive around
and
you know when you drive around san
francisco my god it's a pain in the butt
there's all these runways and hills and
um there's all these signs like
sometimes it's comp like they kind of
conflict each other like the one sign is
saying you know don't park at from four
to six and then another sign is saying
you know uh you're allowed to park and
it's on the same street and so it's like
really confusing and so it's you know
it'd be incredibly useful if google maps
can help find that information right
and then the third thing is you know
local um this is this can be really
helpful for local uh you know venues
right so i i love going to happy hours i
love doing these special events that um
that music like theaters and restaurants
and bars hosts and stuff like that and
it'd be nice if the advertisement
display that shows up in businesses um
are actually from these local businesses
promoting some events right and this can
be really helpful in terms of like users
looking for these type of events but
also venues as well to attract new
customers to bring back current
customers um and also it can be helpful
for google because it can help them
generate advertising revenue so you
start by like laying out a couple ideas
right and then eventually you say okay
um what i would probably do is um do
some like validation to see
whether people would be using this right
and so
um the next step is basically you know
you know validating using data right so
in terms of the eta planning like how
can you validate this idea so you can
look at um search histories right search
routes where like users have searched
for a particular location but they did
not physically move right and this could
be indicative that hey you know what
they're doing some planning sir right
they don't want to just kind of get a
general understanding of how long it
takes to actually commute somewhere
right um
so if there's enough volume with this
type of search maybe it's indicative
that hey you know what we could
definitely build out this type of
feature that can be really helpful for
users right
and then the in terms of street parking
information this could be like in the
form of people who search you know who
are um
basically they search for parking spots
right and and they parks cars on the
street instead of like a parking garage
um and this could be
you know this once again like look at
the volume of this type of search right
um
and that could be indicative of whether
this
whether there's potential demand for
this type of future and then the last
thing is you know users who search um
events like happy hours and and shows
right so you can
so google has a way to like track you
know from people who are using like
google maps to just a regular google
search as well right so
so see like are there type of users who
are searching for particular location or
like a business and then they happen to
search like you know
like uh happy hour at this restaurant
called a bar called bargerack or
something like that and and see and
basically find that history and if
there's enough volume for that you know
maybe there's um
you know it's suggested that hey you
know what this could be really helpful
for uh you know for
for the the type of users that i just
mentioned earlier
um so the last thing is to prioritize
which idea to build and
one way to approach this is basically
using this prioritization framework it's
called impact versus effort analysis and
this is often used in product managers
um
and consultants as well and you can
bring this out as a way to think about
how you would actually value which idea
to focus on
um you know and and that's really really
powerful way to show to the interviewer
that you know you know how to basically
um basically define a scope of a problem
and then be able to figure out like what
to focus on right
um and so this this impact versus the
effort analysis basically for each idea
you're evaluating whether
um you know whether this has low medium
or high impact and whether it has a low
medium and a high effort and ideally you
want to go pursue an idea that's going
to return high impact with low effort um
and as opposed to a high impact and
high effort and of course there are some
exceptions to that but generally you
know that's what's um you know that's
what's desired right because you want
something that's gonna give you the
highest sort of ri you know minimum
effort and maximum return and so
um so you can kind of approach each of
the idea based on this and and and can
say you know um
based on
the ma you know based on this analysis
um
um i can say that you know eta planning
would be low impact and low effort and
and and you can and and basically you
would obviously have to kind of justify
in terms of why that is right so if i
were to sort of approach this like i
would say like you know um so
in terms of just sort of looking at the
impact here like um
this is how rank it so i think the eta
planning is low street parking
information is medium and event planning
is high
even banner is high um and the reason
why i ranked it this way is because um
i
think that even though there are types
of users who may want to do some ata
planning i just don't believe that there
would be high enough volume and
obviously we can kind of validate it
through data um and and at the same time
this is not like a must-have feature
right it's it's like nice to have but it
wouldn't be like the main sort of um
functionality that people would often
use and so that's why i think the impact
is low um i think the street parking
information is definitely
you know relatively speaking compared to
eta planning it's higher than low it's
it'd be medium um
you know because people commute a lot
and
you know this this type of information
can be really helpful and remove a lot
of friction points for a lot of
commuters out there especially if
they're looking for parking spots
um but i don't believe it will be as
high as
the event banners right so because this
not only helps
users find locations or events that they
want to go to but also helps businesses
and especially during the pandemic time
you know as a lot of these venues are
you know reopening they're they're
trying to attract customers regain old
customers through these um sort of happy
hours and events and so they're
perhaps a really good medium a way for
them to like you know um send out any
sort of events that they're hosting
right and so these event banners can be
incredibly helpful but at the same time
it could be additional source of revenue
stream for google ads so we're serving
all three aspects of this and given that
that's why i think the impact is impact
would be really high
but in terms of the level effort that is
required
um i believe that eta planning is low
because it doesn't really require much
sort of
major revamp on the engineering besides
um you know looking at historical trend
and aggregating it and then
you know basically storing it somewhere
and displayed it on um you know on
google maps uh so there wouldn't be much
kind of sort of cross-functional um you
know um coordination that is required to
build out this functionality now in
terms of street parking information i
could definitely see it being high
because you have to be able to extract
information at a street by street level
and sometimes that can change and um i
could think of like obviously this might
require um
a coordination with the
with some software engineers and machine
learning engineers um as a way to parse
this information in the form of like
some urban database like a municipal
database or like computer vision right
because you know google maps um
or google earth has all these different
images of you know like the street by
street and they might they could perhaps
extract you know what is the latest sort
of street information uh from these
signposts that are on you know that are
on the street um that lets them know
whether a car can be parked or not right
um but that kind of
effort would be um would be hard in
terms of um the engineering effort that
is required to build something like this
um and and similar thing i would say the
level effort that's required to build
out this event banner would obviously
require coordination among the
advertisement team um
the uh basically you know attracting
like new customer like advertisers for
this type of banner uh that might take
some time
um it it takes time to basically do a
bit of like a redesign on the um on
where this type of advertising gets
displayed and stuff like that so i
believe that the effort itself would be
fairly high um
and in terms of which one i think we
should pursue is that um i think the
idea should be even better because as i
mentioned earlier um you know this not
only even though the effort in
you know is is as high compared to like
other uh features per se in terms of the
impact that we're getting from this
would be really um you know really
powerful for
uh you know users who are browsing for
venues to go to uh businesses and
advertising revenue and so i think um
going pursuing this idea will be
would be important and
you know you just ended it with that and
of course you mentioned additional sort
of caveat saying um you know of course
like obviously this requires
conversation with product managers and
engineering managers and eventually kind
of converge toward which idea to focus
on and it'll be it through meetings and
kind of documentations and you would
obviously flesh it out but in terms of
just like the information i currently
have i think this is the best idea to
pursue um and so you mentioned that as a
caveat but also at the same time you're
showing that you have real life
experience in um and being able to um
you know
basically approach this type of problem
and that's really what the purpose of
these interviews are so
so that can be really helpful
so there you have it guys you know so
those are the five steps um
in terms of how would you approach a
problem about improving your product
now if you if you like this type of
content make sure you check out
datantb.com
because it contains courses um contains
a b testing product cases question banks
and like mock interview videos that are
really helpful for your interview prep
but also at the same time there's also
coaching services so you can um pair up
with an instructor like myself and then
i can give you like a personalized mock
interview for the meta um data scientist
interview google data scientist review
pretty much any data sentence interview
um and so make sure you check out
datantv.com and i will see you in the
next video
bye
2CUTURL
Created in 2013, 2CUTURL has been on the forefront of entertainment and breaking news. Our editorial staff delivers high quality articles, video, documentary and live along with multi-platform content.
© 2CUTURL. All Rights Reserved.