Filippos Protogeridis @protogeridis

Guide

Data for
Designers 101
Understand and leverage data to create better
experiences for your customers
Filippos Protogeridis @protogeridis

Everyone talks data.

Filippos Protogeridis @protogeridis

But I’ve only met a

handful designers that
truly know how to
work with data
Filippos Protogeridis @protogeridis

With this guide you’ll

learn how to understand
and work with data

Slide to continue
Filippos Protogeridis @protogeridis

PROCESS

01 Align

02 Track

03 Analyze

04 Understand

05 Action
Filippos Protogeridis @protogeridis

01

Align

The first step to working with data is to align on the

key success indicators, from a combination of

business and product metrics.

Filippos Protogeridis @protogeridis

Let’s take as an
example a consumer
digital banking app

(e.g. Revolut, Monzo)

Filippos Protogeridis @protogeridis

What we care about is

how many people
switch to us as their
main bank provider,
and whether they
Account switches

continue using us.

Monthly deposits
Filippos Protogeridis @protogeridis

Business metrics
Switches from other banks

Monthly deposit amount

These metrics drive success for the business,

but we can’t influence them directly easily.

Product metrics
Onboarding completion rate

Time to first deposit

Avg. deposit amount

These metrics may indicate problematic areas that

we can gradually work and experiment on.
Filippos Protogeridis @protogeridis

Business metrics are

hard to impact directly.

But product metrics are

much more easily

changed.
Filippos Protogeridis @protogeridis

In this case, let’s pick

onboarding completion
as our focus area, as it’s
a product metric we
can easily influence.
Filippos Protogeridis @protogeridis

02

Track
Once you align on your key data metrics, it’s time to
track their performance over time.
 Filippos Protogeridis @protogeridis

WHAT TO TRACK

Business How many people switch to us as their main bank? How

many people have made their first deposit? What’s our
metrics average monthly deposit?

Product How does our onboarding funnel perform?

Where is the biggest dropoff? How many people use X

analytics feature?

AB Tests Are we running any experiments?

How are they moving our key metrics?

 Filippos Protogeridis @protogeridis

In most cases, you should have one or more dashboards to observe business and product
metrics. Examples include Mixpanel, Google Analytics, Amplitude, Looker, Tableau and so on.

Download KYC Passed First Deposit

Account Types Created

 Filippos Protogeridis @protogeridis

You may also be using AB testing software e.g. VWO, Optimisely, ABTasty

New Onboarding Experience

Experiences Expected Probability to Conversions

Conversion Rate Improvement / Visitors
beat baseline

C Current Experience 1.39% Baseline Baseline 92 / 6,672

V1 New Onboarding Experience 3.20% 130.14% 100% 214 / 6,715

T Total 2.29% - - 306 / 13,387

Filippos Protogeridis @protogeridis

If you aren’t tracking that

data already, it’s crucial

that you engage with

your team in setting it up.

This is usually done by a product manager in startups

businesses, or a data specialist in larger teams.

Filippos Protogeridis @protogeridis

03

Analyze
Once you are tracking your data accurately, you can
start observing specific patterns, trends and
discrepancies.
Filippos Protogeridis @protogeridis

Time to identify potential

problem areas using data

You’ll usually want to involve a product manager and a

data analyst for this step.

 Filippos Protogeridis @protogeridis

WHAT TO OBSERVE

Dropoffs Where are our users dropping off the most

in our funnels? Why?

Discrepancies Are we noticing any metric that is different

from before? Why?

Trends & Is one action/trigger affecting another? E.g.

users that select a specific account type
Correlations
have a higher completion rate.

Time to complete Is it taking too long for users to perform

an action or complete a task? Why?

Filippos Protogeridis @protogeridis

Dropoffs
A big dropoff at a certain funnel step
may imply the following:
Not ready to commit / Need more time
Low confidence in purchasing
Lack of enough information
Usability issues
Filippos Protogeridis @protogeridis

Discrepancies
A discrepancy between different points
in time may imply:
A bug we released but didn’t notice
External factors such as seasonality or a market
change
A new update we launched that isn’t performing well
Filippos Protogeridis @protogeridis

Trends & Correlations

Correlations are harder to validate, but
often help us shine light to interesting
behaviors:
A specific choice in the flow is affecting a later
behavior
A specific action in the flow leads to better
performance
A specific user type is more likely to complete a flow
Filippos Protogeridis @protogeridis

Time to complete
Long times to complete a certain action
aren’t always negative, but may need
investigation as they imply:
Usability issues are making the task difficult to
complete
We are requiring too much information from a user
There isn’t a big enough reward to complete the
flow
Filippos Protogeridis @protogeridis

Work with your team to

identify the most
impactful problem to
solve.
Filippos Protogeridis @protogeridis

EXAMPLE

“We have noticed a big drop-off in

the date of birth step of onboarding.”
10% Drop
40% Drop

7% Drop
10% Drop

Email Date of birth Country Address Plan type

Something must be off

Filippos Protogeridis @protogeridis

04

Understand
We know something is off, but we aren’t sure why.
Let’s look at some simple way to answer that.
Filippos Protogeridis @protogeridis

Low effort
In-app surveys

User recordings

Will give us an indication on why something is

happening. What are people’s behaviors? What are
they expecting at each step?

High effort
User interviews

Diary studies

Will help us really understand the reasons that may

be influencing specific data points.
Filippos Protogeridis @protogeridis

Use a number of
observation methods to
understand why a
particular metric is
affected.
Filippos Protogeridis @protogeridis

EXAMPLE

We noticed a big drop-off in the date of birth step.

We ran an in-app survey that asked users what’s

stopping them from continuing, and also asked to get
on a 30’ call with them to further discuss.

After seeing the results of the survey and speaking with

users, we realised that we were asking people for their
personal information too early, and without giving
enough reason.
Filippos Protogeridis @protogeridis

05

Action
It’s time to come up with solutions that may help us
solve the problem areas we have identified.
Filippos Protogeridis @protogeridis

Now that we know the

root cause of the issue,

we can ideate on

solutions.
 Filippos Protogeridis @protogeridis

IDEATION METHODS

Journey mapping Understand the user’s key questions,

motivations and frustrations at the
respective point in their journey.

Brainstorming Brainstorm with your team on a whiteboard

and explore as many ideas as possible,
before you rank and prioritize.

Sketching Use sketching techniques such as Crazy 8’s

to explore innovative, radical solutions.
Filippos Protogeridis @protogeridis

OBSERVATION

“We are asking for the user’s date of

birth very early in the flow, without

giving them a reason.”

HYPOTHESIS

“We believe that by moving the step

later in the flow we can increase the

completion rate, as users will have

more confidence.”
Filippos Protogeridis @protogeridis

Time to launch an

experiment and see

whether our key metrics

are affected.
Experiments such as AB Tests only work with high-traffic

products. In low traffic product, you may need to simply

launch and observe what happens.

 Filippos Protogeridis @protogeridis

Onboarding: Move DOB later in the flow

Experiences Expected Improvement Probability to Conversions

Conversion Rate beat baseline / Visitors

C Current Experience 1.39% Baseline Baseline 92 / 6,672

V1 DOB moved later 3.20% 130.14% 100% 214 / 6,715

Hooray!
Filippos Protogeridis @protogeridis

Let’s be realistic though:

Not all experiments work.
In reality, most won’t.
Filippos Protogeridis @protogeridis

It’s important that we

learn from both.

Two or three failed experiments will often lead us to

the one that completely changes the game.

Filippos Protogeridis @protogeridis

And that’s it!

To summarize...
Filippos Protogeridis @protogeridis

In summary

Align on your key metrics (business & product

Make sure you are tracking them accuratel
Identify problem areas with your tea
Validate them by using qual/quant method
Ideate on solutions as a grou
Launch, experiment and learn
Filippos Protogeridis @protogeridis

Was this useful?

Repost to your network.

I’m Filippos, a product design leader based in London. I

share career advice I wish I had 12 years ago when I was

starting off in UX.

Follow me for weekly posts on getting started in product

design, transitioning from junior to senior, as well as pro

UI/UX tips.