11050_Peter_Zemsky_AI_Business_School_3_Retail_FINAL_V2
0 (0 Likes / 0 Dislikes)
[Driving Business
Value from AI]
>> Peter: Retail is
a very important setting
for understanding the business
value that's coming from AI.
From search
to e-commerce,
digital has been
disrupting the sector
and AI promises
to accelerate.
At the same time,
retail as a sector,
prone to intense competition
and tight margins.
A strategic approach to AI
focused on value
is really critical.
Fabletics, a born-in-the-Cloud
fashion e-tailer,
illustrates this well.
Founded in 2013 and operating
across ten countries,
they focus on women's
sportswear and accessories
in the growing
ath-leisure category.,
Leggings are a key product.
Like for
most retailers today,
Amazon is a key
and demanding competitor.
In the old physical world,
specialist stores like this
could carve out niches next
to general department stores.
Not true with the endless
options
available with e-commerce.
Here, for example,
is Amazon's dedicated page
just for women's leggings.
They themselves
have the depth of offer
you'd expect
at a specialist,
including a set of
partner brands
attracted by the large
user base of Amazon
and, of course,
the site has access
to Amazon's full suite of
e-commerce capabilities,
prime delivery search,
easy access to user ratings.
Now,
as a key differentiator,
Fabletics has pursued
a membership model,
where users get
monthly shipments.
They've also pursued a
celebrity endorsement approach
to branding with
co-founder Kate Hudson,
among others.
In summary,
the company is an early mover
in an emerging new category,
and has leveraged e-commerce
for a fast global entry
with already over
1.3 million members.
As one of their strategies
for avoiding direct competition
with the online
cost leader Amazon,
they pursue a
subscription model,
a model which shot to
prominence in the retail sector
with the success
of Dollar Shave Club,
which Unilever
purchased a few years ago
for a reported $1 billion.
Our question is
how can AI contribute
to Fabletics developing
a winning value proposition
in this demanding space?
Let's consider key elements of
their baseline value creation.
First, they're
looking to attract members
with a strong,
celebrity lifestyle brand
and the authority
on the category
that comes from being
a dedicated specialist.
On the cost side,
they have benefits as well.
First of all, from
lower marketing expenses
on their existing customers,
and they get
supply chain efficiences
from the more
predictable shipments
from a membership model.
However,
like any business model,
there are negatives, as well.
For the users, they risk
getting unwanted shipments.
They have until
the 5th of each month
to decline a monthly
order if they don't want it.
and the company
has the heightened risk
of returns of merchandise
that customers don't like,
which we know are a
key cost driver in e-commerce.
Also, it's worth noting
that initial
customer acquisition costs
can be very high
because you're asking
members to commit
to ongoing monthly orders.
So, how can AI drive
a leap in value creation
for a company
such as Fabletics?
Well, the core application
of machine learning
is to make predictions
based on rich data sets.
This is critical for retail,
whose core value creation
comes from connecting customers
with the right products.
Machine learning,
for the first time,
makes possible
fast, high quality
and mass personalization
of these kind of
product recommendations
and general selections.
I think the thing to note
is this kind of capability
is especially useful
in a subscription model.
The personalized selections
only deepen the members'
connection to their brand,
and moreover,
the better match
the selection
to each individual,
the lower their anxiety
of getting unwanted shipments.
On the cost side,
a KPI of the algorithm will
be the rate of costly returns.
And finally,
supply chain management
can be improved by
machine-learning predictions,
not at the individual level,
but on the aggregate ordering
of the entire membership,
based on their latest data;
for example,
on their website browsing.
Thus, we see here
another AI use case
which is compelling
for the retail sector.
But what about execution?
We know that the key
to effective machine learning
is building
the right data sets,
and this typically
needs attention
from across the organization.
At Fabletics,
the data gathering
actually starts
from the first click.
A user attracted by one
of their new member promotions
is of course asked to enter
a host of sizing information,
which you see here.
However, note that
this is only on screen five
of the sign-up process.
The first screens
ask the new member
about the type
of exercise they do,
then where do
they like to exercise
and even some
initial color preferences.
Why? Well, this is
exactly the sort of information
that can be useful to seed
machine-learning algorithms,
and will help drive
high quality predictions
and recommendations
when combined with data
on the member's
browsing history
and, ultimately,
on their possible returns.
One can simply
not underestimate
the importance of such
an integrated approach
to data gathering and storage
for AI strategies like this.
So, there are really
two headlines so far
from the Fabletics case.
First,
the ability of high quality,
personalized recommendations
to drive value
creation in retail.
It's a huge development.
The second, the need to
have a culture of data capture
across your organization.
And let's now move on
and consider one of the
biggest trends in retail today,
omni-channel.
Should this born-in-the-Cloud
online retail
open physical stores?
Does AI help make
this significant investment
a value-creating move?
Well, we know from Apple
that retail stores
can be a great way
to reinforce a brand.
We know from
retail banking and telcos
that physical locations
can be critical
for customer acquisitions.
So these benefits
are highly relevant
to Fablitics' value creation,
and they've actually
opened 24 initial stores
already in 2017.
However, true to their
culture of data gathering,
this was an integral
part of their store designs
from the very beginning.
For example, through the
use of employee handhelds,
they're able to track
what customers
are selecting to try on
and observe which items
are then purchased
and which are left sitting
on the changing room floor.
On a website,
when an items fails to sell,
it's often very hard
to disentangle why.
Is it the color?
The picture that you displayed?
Is it the price?
You can get, often,
much richer information
inside a store.
As the Fabletics
president summarized,
more data does not
equate to more insights.
We needed better,
more contextualized data.
This is the reason
we decided to open stores.
Their data
flows both ways.
When a member
walks into a physical store,
the sales associate
is fed user preferences
and recommendations
based on their prior
online shopping behavior,
and when a customer
tries on an item
but does not
complete the sale,
it can be added to
their online shopping basket
for later consideration.
Store inventory,
obviously critical
for the success
of physical retail,
can be optimized
based on analysis
of the data of
your local members,
social media sentiment
and in-store
heat-mapping data.
Now, one challenge
of extricating these
AI-infused digital strategies
is to cost-effectively
build the range
of organizational
capabilities required,
especially for
a brand like Fabletics
that is still busy scaling up.
So in the
case of Fabletics,
they overcome
the scale requirement
through their parent company,
the TechStyle Fashion Group,
which is able
to provide them
a common set of digital
tools and infrastructures,
not just to Fabletics
but to its range of
online fashion companies.
This is really
not an area of strategy
where you want
to go it alone,
especially as
a smaller-sized unit.
We've now seen several
other key takeaways
from this case.
Fabletics is a great
example of brick mining,
using physical stores
to augment and enrich
your online customer data.
Fabletics demonstrates
the power of AI predictions
to drive not just
customer recommendations,
but also predictions
to improve inventory
and supply chain management.
You see as well that AI
strategies cannot be separated
from the rest of
your digital initiatives.
This is logical because
it's actually proliferation
of digital data coming
from across your enterprise
that is fueling the
value creation from AI.
Finally, as attractive as these
AI and digital strategies are,
you need to be careful
about finding ways
to share the fixed costs
to assure that
you're driving a strong ROI.
This can either
involve shared efforts
across a
larger corporation,
as in the Fabletics case,
or it can involve teaming
up with external partners.
Retail,
a dynamic sector
for digital and for AI, indeed.