For the French retailer Monoprix, we envisioned an interactive shopping list based on a voice interface. Using customer and retail data we designed a unique user experience.
We will show you how we communicate brand identity, increase sales, and improve the shopping experience through voice.
4. A squad operating
on all marketing
needs
Technologies
Our highly accredited experts have 20
years’ experience in leading award-
winning global digital marketing
campaigns.
Business oriented, data raised, digital
native, marketing minded and IT
trained unicorns, they live to put
together the perfect team of our in-
house talents.
Our Full-stack Engineers will set-up
end to end MediaLake, with the right
trade-off between speed and
reversibility
AI specialists with PhDs in Machine
Learning developing unique
proprietary solutions to fit our clients’
needs and optimize in-house
processes.
Media & Activation
Creation
Big Data
Ecosystem
Artificial Intelligence
Strategic Consulting
Data Expertise
Award-winning creatives, able to design
concepts from branding to performance
based on data.
Consulting
Marketing
First class data experts from the
largest international pool of giant tech
certified tracking experts, analytics
specialists, data consultants and data
scientists.
5. Benelux
40 people
Switzerland
10 people
Dubai
35 people
Hong Kong
8 people
South Africa
15 people
ASEAN
15 people
Australia
10 people
France
250 people
UK
160 people
Italy
15 people
Spain
35 people
Germany
190 people
Nordics
55 people
China
80 people
USA
Brazil
5
A squad of 1000 people operating worldwide
6. Our credentials
Retail / E-
commerce /
CPG
Banking /
Insurance
Travel /
Hospitality
Automotive Media /
Entertain.
Cosmetic /
Luxury
Telecoms /
High tech
Other
7. 7
ECHO Awards -
VM in Digital
Performance 2016
B2B Agency
Awards
2015-2016
European
Performance
Marketing Awards
2016
European Business
Awards
2016
Online Retail
Awards
2015
Kenshoo Infinity
Awards
2017
Econsultancy
Top 100 Digital
Agencies 2015-
2016-2017
The Drum RAR
Digital awards
2016
Grand Prix
Stratégies
2017
Top 50 hyper growth
French companies
2017
Grand Prix
Data Festival Paris
2017
Agence innovante
de l’année
2017
Proven Track Record
Performance Best agency Marketing Innovation
Grand Prix Stratégies
du Digital 2018
Trophées
Marketing 2018
One Show
2018
Grand Prix de la
Vidéo Numérique
2018
Nuit des Rois
2018
Club des DA
2018
Grand Prix
Stratégies de la
publicité 2018
Grand Prix
Stratégies du
Digital 2018
11. Do you know?
How much longer a voice search is
in % compared to a text search?*
12. Source: Bruceclay.com
Voice searches are
significantly longer than
text-initiated searched
76%
This leaves more room for brands to
understand customer intention, claim
it and engage into a meaningful and
mutually profitable dialogue
20. Be part of the new customer journey.
Enhance ease of use
and demonstrate
innovation
Practical use cases :
● Track bill/ usage
● Report service issues
● Update meter readings
● Manage home
automation
New propositions around
the recommendations,
bookings and loyalty
Practical use cases :
● Room/ flight availability
● Book a ticket/stay
● Get a destination
recommendation
● Check loyalty points
ENERGYHOSPITALITY /TRAVEL MEDIA
“Bring laid back to lean
forward”
Practical use cases :
● Read articles
● Program
recommendation
● Schedule information
● Save for later
20
RETAIL
Offering services
adapted to each user’s
profile
Practical use cases :
● Track your orders
● Gift recommendation
● Add product to cart
● Get information and
book a show
22. Monoprix has very specific objectives for its voice app
Enjeux monop
Enjeu Monoprix
OFFER AN OMNICHANNEL EXPERIENCE INTEGRATING
OFFLINE AND ONLINE SHOPPING
REDUCE FRICTION POINTS DURING THE SHOPPING
JOURNEY
TRANSFORM RUNNING ERRANDS TO MOMENTS OF
PLEASURE
23. The smart shopping list must be able to follow the consumer
and adapt to changes in his daily life.
HAPPY PATH - DEFAULT USE CASE
TBC
User
FRONTS
MONOPRIX
App Monoprix
Google Home
App
Monoprix & Moi
Site
Monoprix.fr
Jon sets up his Google
Home and associates
his Monoprix account
with his Email
At Home On the Go /
In Store
At Home /
On the Go
Jon changes his
shopping list by
talking to his
Google Home
Jon leaves
home and goes
to work
TALK EDIT PURCHASE
Jon connects
to
Monoprix.fr
Once in his
Monoprix app,
Jon picks the
items from
his list he
wants to buy.
Jon opens the
Monoprix app and
says: "Can you add
avocados to my
shopping list?”
"Remove
avocados from
my shopping list
and add
detergent"
"Summarize my
shopping list!"
Holy guacamole, yes!
Jon can map his
needs to actual
product SKUs.
The products added to
via the voice interface
appear as a popup on
the site. Jon can
validate his products
and order.
He opens
the M & M
app
23
25. Use available data to offer an intelligent and innovative
service.
25
▪ Create a smart shopping list
▪ Communicate the Monoprix brand and
highlight Monoprix products
▪ Highlight the innovative technological
lever and make the experience cool and
fun
91%
of Monoprix’s
customers
create
shopping lists
30%
on
smartphone
**
No easy
to use
tools
available
+
“Ok Google, add
sugar to my
Monoprix shopping
list”
26. The success of the smart shopping list experience depends on
the fluidity of four main features.
26
CREATING A LIST AND
ADDING ITEMS
REMOVING ITEMS
SUMMARIZING AND
VALIDATING THE LIST
ITEM RECOMMENDATION
1 2 3 4
BASIC SERVICE ASPECTS
INTELLIGENT
EXPERIENCE
27. The mise-en-place: setting up the workspace.
27
DATALAKE
API BACKEND (Artefact for Monoprix)
4M CLIENT DATA
PRODUCT DATA
MODEL
TRAINING
EXTERNAL DATA
ALGORITHMS
MAPPING SKUS
PRODUCT
RECOMMENDATION
GOOGLE HOME
VOICE
INTERFACE
MOBILE APP
E-COMMERCE
PLATFORM
Chatbot
Dialogflow
TECHNICAL
SETUP
DATA
INTELLIGENCE
DIGITAL
ASSET
NECESSARY ELEMENTS
28. Artefact accompanied Monoprix during the implementation
of these features along three major components.
28
Product
vision
29. Component 1 - Data: Adding intelligence to the shopping list.
29
Data cleaning
and
transformation
Product
mapping
algorithm
Product
recommendation
algorithm
30. 30
INPUT “Président butter”
FINAL SCORE
A final bonus is added to consumers'
favorite products
PROPOSED SKU
TEXT ANALYSIS
DATABASE SEARCH
PERFORMANCE
BOOST
butter Président
The most purchased products globally
benefit from a bonus factor
Name :
Brand:
Full name :butter
Président Président + butter
PERSONALIZATION
Data - SKU mapping algorithm.
31. Data: Calculation of product recommendations.
31
02/03
03/03
05/03
06/03
09/03
10/03
1
2
3
4
5
Purchase data Most frequent associations
Frequent
Pattern
Matching
Algorithm
(FP Growth)
The objective of this phase is to extract the relevant information from the purchasing data, in
this case the frequent product associations.
32. Data: Calculation of relevance scores.
32
This score makes it possible to eliminate the products that are bought very often and that
therefore go up in the frequent associations, without being relevant.
Number of baskets containing Number of baskets containing
without
Number of baskets containing
without
Number of baskets with none of them
Do we recommend with ?
Log likelihood ratio relevance score (compare to threshold)
33. Data: DSS flow for the recommendation system.
33
Ingestion and cleaning of
purchasing data
Calculation of frequent
product associations
Calculation of the relevance score
of the product associations
Step 1 Step 2
34. Component 2: Tech - Chatbot and recommendation system
overview.
34
Voice-based
product vision
ChatbotAPI
35. Tech: Chatbot.
35
Google Cloud
Function
< >
Search through
information, perform
calculations, manipulate
a database ...
User
Google
Home
Input
(voice)
Output
(voice)
Text analysis:
understanding of
the query and its
parameters
Text request
Response
Request
Reply
APIs Monoprix
Monoprix
Database
1 2 3
456
36. Tech: Recommendation system.
36
E-commerce
Mobile app
Google Home
Recommendation
tables are stored
in Elasticsearch
Calculate
recommendations
in Google Cloud’s
DataProc
Store raw data in
Google Cloud
Storage
The API queries the
Elasticsearch table
for
recommendations
Extract 2 years of
shopping history
from internal
databases
Customer
“Ben ik
iets
vergeten?
?”
BACKFRONT
“Melk”
1 ms response
time
37. Component 3: Creative - Creation on all fronts (bots,
applications and websites).
37
App
E-commerce
platform
Voice
interface
39. Challenges and opportunities
VAST PRODUCT CATALOGUE - REQUIRES SMART
CHOICES
OPPORTUNITY TO MARKET NEW OR SPECIAL
PRODUCTS TO THE CUSTOMERS
UNNATURAL CONVERSATION AND LANGUAGE
CONSTRAINTS
OPPORTUNITY TO COMMUNICATE YOUR BRAND
FLEXIBLE SCORING AND RECOMMENDATIONS
39
“New product,
currently on sale,
you’ll love it!”
“Butter added to
cart. Continue yes or
no?”
“Sure thing! Ready
to curry on?”
“Monop, add some
curry to my cart”
40. Questions? Join us at stand #28 for
more information and individual
demo’s!
40
Artefact Benelux is hiring!
Check our Utrecht vacancies at jobs.artefact.com
42. Questions? Join us at stand #28 for
more information and individual
demo’s!
Artefact Benelux is
hiring!
Check our Utrecht
vacancies at
jobs.artefact.com
42
43. The Monoprix app in action.
43
+30 000
Products bought
via the app
+6 000
Unique Users
Results
Notas do Editor
Welcome everyone to this session!
Today we will present our Monoprix client case “Building voice-based intelligent shopping lists” which is according to us, and hopefully you’ll agree after is, a perfect combination of utilizing new technology (voice), big data infrastructure and data science to create a unique experience for our clients customers.
And that’s is exactly our business as marketing engineers.
We represent Artefact which is a digital agency that combines performance marketing and technology.
But let me first introduce ourselves….
We do that with over 1000 colleagues world wide.
We are very proud to also have been awarded for our work and client collaborations. During this session we are going to address one of these award winning. The project that we did for the French retailer Monoprix. But first...
Let’s do a quick pop quiz. To get into the ‘mood’ first 2 quick questions. We have prices so be sure to respond quickly if you know the correct answer
So here comes… what percentage of searched will be voice searches in 2020?
Source is ComScore. Comes from a report issues in 2017 that has been quoted pretty much in every article on voice search.
Then something that has more to do with the characteristics of voice search.
Voice searches are significantly longer than text-initiated searched
This leaves more room for brands to understand customer intention, claim it and engage into a meaningful and mutually profitable dialogue
Already 20% of all searches in the US are claimed to be voice searches. So the future is now. But how did it all started?
Well let’s go back to the 90’s and the beginning of the internet.
For those in this room that were already born and had access to a computer, you probably remember that ‘surfing the web’ meant clicking on hyperlinks in very heterogeneous directories.
Then moving into the zero’s our desktop suddenly cleaned up dramatically leaving just one modest search box where we could enter our search query
A decade later we immersed ourselves in vast social networks where we discovered old friends and made a lot of new ones.
But even though there was definitely an evolution going on in the information exchange with the internet (also the arrival of images and video’s and new UX elements) search was still pretty much done by textual input.
But now, approaching the 20’s that is changing fast. Assistant didn’t only arrive on our phones and tablets and even in operating systems on desktops, but we now have smart speakers that sit in our homes and with what you can engage into a conversation.
Smart speakers installed in our homes have already reached the 100M, and the penetration is expected to increase significantly.
In the Netherlands the launch of Google Home with a Dutch assistant later this year (before Christmas) which will undoubtedly be of influence of the smart speaker penetration in the Dutch homes.
This new customer journey, where you engage in a conversation in your home or on the go on your phone with brands, will have a big impact on marketing, customer services and business in general.
This new technology gives brands new options to engage in a meaningful and mutually profitable dialogues with their customers. What that dialogue is will be different per market.
Travel: booking and travel information
Energy: especially with the connection with home devices IoT, track usage, bill, update meter readings but also manage home automation, so you know your house is comfortably heated when you arrive in the evening after work.\
Media: listen to music, recommendations, info about schedule, start a movie
So for retail a practical use case would be tracking orders, getting product information and ordering products.
That bring us to how we applied voice search for Monoprix.
Monoprix approached Artefact with these objectives.
They wanted to:
REDUCE FRICTION POINTS DURING THE SHOPPING JOURNEY
TRANSFORM RUNNING ERRANDS TO MOMENTS OF PLEASURE
OFFER AN OMNICHANNEL EXPERIENCE INTEGRATING OFFLINE AND ONLINE SHOPPING
So what does an omnichannel, friction-less customer journey look like?
And how would we then make it fun? Here is where Artefact comes in...
Monoprix wants to offer its clients a unique experience. This is reflected in their company motto which means something like “ No to habits”
So wanted a unique experience for their customers and knowing after research that 91% of their customers make shopping lists of which 30% were done on a smartphone, with no tools available, there was definitely a need and thus a market...