- il y a 2 jours
Carrefour Pragmactic innovation in Retail
Catégorie
🤖
TechnologieTranscription
00:01Sous-titrage Société Radio-Canada
00:30Sous-titrage Société Radio-Canada
01:00Sous-titrage Société Radio-Canada
01:02Sous-titrage Société Radio-Canada
01:37Sous-titrage Société Radio-Canada
01:42Sous-titrage Société Radio-Canada
02:16Sous-titrage Société Radio-Canada
02:18Sous-titrage Société Radio-Canada
02:21Sous-titrage Société Radio-Canada
02:24Sous-titrage Société Radio-Canada
02:30Sous-titrage Société Radio-Canada
02:32Sous-titrage Société Radio-Canada
02:34Sous-titrage Société Radio-Canada
02:36Sous-titrage Société Radio-Canada
02:40Sous-titrage Société Radio-Canada
02:44Sous-titrage Société Radio-Canada
02:53Sous-titrage Société Radio-Canada
02:57Sous-titrage Société Radio-Canada
03:09Sous-titrage Société Radio-Canada
03:12Sous-titrage Société Radio-Canada
03:27Sous-titrage Société Radio-Canada
03:29Sous-titrage Société Radio-Canada
03:42Et nous n'avons pas fait rien.
03:44Et nous avons fait un projet, et Nicolas va illustrer ça.
03:47Nous avons fait un projet, nous avons créé notre propre VC avec Daphne.
03:51Parce que je vendais des potatoes, je ne sais pas comment être un fan d'investissement,
03:54mais je sais comment être partagé avec un fonds d'investissement.
03:57Et donc nous avons créé notre propre VC pour aller à un projet,
04:01partagé avec le tech écosystème.
04:03Et ça nous donne un autre mindset totalement différent.
04:05Et maintenant, nous sommes conscients de la évolution du marché.
04:08Nous savons ce qui s'est passé et ma façon de le dire c'est que le partagé avec le
04:14tech écosystème
04:15est notre R&D à Carrefour.
04:17C'est comme nous construisons R&D.
04:19Non de l'entreprise tech, non de choses qui sont complexes,
04:21juste avoir la relation correcte avec le tech écosystème.
04:26Donc c'est le pêlant, construire nos capacités,
04:29partagé avec le tech écosystème.
04:31Et last mais non le plus, nous n'oublions pas les potatoes.
04:36Do you know how many potatoes I need to sell to earn 1 euro?
04:42I need to sell 10 kilograms.
04:45So do you know when I launch an IT project of 10,000 euros,
04:48how many potatoes I need to sell?
04:51It's 100 tonnes.
04:53So each time my team come with a project,
04:56we ask how many potatoes we need to sell.
04:59And each time a project is one day late,
05:03it's a number of tonnes of potatoes.
05:05So either you're on time with a project,
05:07either you go to the stores and you sell the potatoes.
05:09That's the deal if you work at Carrefour.
05:11And I will explain to you how it helps us,
05:14I think, go a little bit faster.
05:15So let's start with how we partner with the tech ecosystem
05:19and innovation with Nicolas.
05:22Thank you, Elodie.
05:24Let's be honest,
05:26there is no secret source for innovation,
05:28at least at Carrefour.
05:30We could design some roadmap for two years,
05:33we could even spend six months
05:36assessing the risk of such technology before launching,
05:39but we don't.
05:40what we did,
05:42we had three principles for innovation,
05:45test, test, and test.
05:47And sometimes we think,
05:50but after,
05:51because retail is moving so fast,
05:53so we need to be the first on the ball.
05:56The downside of having a testing-first approach
06:00is that sometimes we do not scale,
06:03but that's okay, that's fine,
06:05that's innovation.
06:06And I can give you a small example.
06:10A few months ago,
06:11we launched a new services
06:13with a project manager in my team.
06:16The objective was to deliver customers
06:18through Smart Lock.
06:20So to be able to deliver them
06:22even if they're not at home.
06:24So we spent six weeks designing the service,
06:27we select some beta customers,
06:30we quit them with Smart Lock,
06:32and two months running the service.
06:35Guess what?
06:35No scale.
06:38No scale because technology was not mature at all.
06:42Actually, just a few examples.
06:47One day, it was Saturday at midnight,
06:50Melanie and my team received a call from a customer,
06:53a beta customer,
06:54so Saturday at midnight saying,
06:55I'm just in front of my door,
06:57does not work.
06:58So Smart Lock does not work.
07:00Maybe it was drunk,
07:01who knows?
07:02but actually the lock did not work.
07:05So Melanie took his phone,
07:08called the locksmith that went in to the customer,
07:12unlocked the door,
07:12etc.
07:12and the day after,
07:14he just put back a normal lock.
07:17So the lesson is
07:18never be a beta customer of Nicolas.
07:20They wanted to change my door
07:22at the beginning of the study,
07:24but so I have this lesson,
07:26I will never be a beta tester.
07:29Fun fact,
07:30but true learning,
07:31because when you read the website
07:33from the provider,
07:36sounds amazing technology.
07:38When you ask for customer using it,
07:40sounds amazing technology,
07:42but it's not,
07:43at least for now,
07:44and the only way to understand it for Carrefour
07:46was to test,
07:47and this is what we did,
07:49and now we're still looking at the technology,
07:54but of course we stopped the test.
07:58So testing is good,
08:00but of course,
08:01what to test?
08:02So we have to understand
08:05and to identify some trends.
08:07If I get back three,
08:08four years from now,
08:10we have a fragmented ecosystem,
08:13we used to partner
08:14with a few startups,
08:15etc.
08:16but the truth was
08:18that we were missing key innovations.
08:20We were missing trends,
08:22and we were aware of trends
08:23after competition.
08:27What we did,
08:28we partnered with Daphne
08:30to launch a VC fund called Dastor.
08:33They bring the VC expertise.
08:35We are not a VC at Carrefour,
08:38and we bring the retail expertise.
08:41Daphne is not a retailer, never.
08:43So we have received many pitches,
08:46more than around 2,000 pitches.
08:49Of course,
08:49we have invested in 11 startups,
08:51but what we cannot see in the figures,
08:54actually,
08:55is our deeper understanding
08:57of the trends.
08:59On the trends in France,
09:01in the US,
09:02in LATAM,
09:02in Israel,
09:03around the world.
09:06What we cannot see in other
09:08is the way we work with startups
09:10right now.
09:11If I get back three,
09:13four years now,
09:15Carrefour was working with startups,
09:17like we worked with Google,
09:19or with P&G,
09:20or L'Oreal.
09:20So from day one,
09:22you can imagine that startups
09:24receive a 50-page contract,
09:27has to go through
09:28every single process
09:30of working with a big provider,
09:33so technical audit,
09:35procurement negotiation,
09:37financial analyst,
09:38etc.
09:38and it took months
09:40or even years
09:41to make a test.
09:43Now,
09:44thanks to Dastor
09:45and thanks to a better understanding,
09:47we have light processes.
09:49So we have ease
09:51of committing budgets,
09:52and thanks,
09:52Elodie,
09:53for giving me that freedom.
09:56and doing a test,
09:58a POC MVP,
09:59takes weeks or days
10:02and not months anymore.
10:05So just as a conclusion
10:07before giving the floor to Elodie,
10:10if you have to remember one sentence,
10:14we adapt the processes
10:15to the innovation at Carrefour
10:17and not the contrary.
10:19So that's the first pillar,
10:21our R&D with startups
10:23and ecosystem.
10:24And the second dimension
10:25is how we really build
10:27the capabilities,
10:28and I want to give you
10:30the illustration
10:30of how we manage
10:32to launch
10:33these new GNI solutions
10:35in a matter of weeks.
10:36What are really the recipe,
10:38what is the recipe
10:39to do that?
10:40So the first thing
10:42is you need to have
10:43the foundations,
10:45the pillar.
10:46So the foundation is key.
10:48If you don't invest,
10:49and if you look
10:50at the VivaTech,
10:51you see like
10:52everyone is about AI,
10:54and I feel more comfortable
10:55that we have invested
10:57some time,
10:58money and people
10:59in building the foundation
11:00of AI
11:01over the past years.
11:03Because if we had not,
11:04that would be another story.
11:06So really,
11:07building the basics
11:08of AI
11:09is absolutely,
11:11to me,
11:12key,
11:12and without it,
11:13you can never launch anything.
11:15And you know,
11:16I mentioned it,
11:17when your business
11:18is about potatoes,
11:19it's difficult
11:20to build MLOps.
11:21So you need also
11:23to be able to have
11:24vision and conviction,
11:26that is the way to go.
11:28That's the first thing,
11:29so the right foundations.
11:31The second is,
11:33you need to adapt
11:34to your business.
11:36At Carrefour,
11:37we will never be
11:38number one in AI.
11:39Let's forget about it.
11:42Not even number two,
11:43number three,
11:44number ten,
11:44whatever.
11:45But where we can be
11:46number one
11:47is the way
11:48we adapt AI
11:49to our specific business
11:51and customer needs.
11:53Because knowing
11:53our customers,
11:54this is what we do best.
11:56So what we did
11:57with Nicolas
11:58is to go to
11:59every line of business,
12:01customer facing,
12:02operations,
12:04support functions,
12:04and understand
12:06to what extent
12:07our AI,
12:08our Gen AI
12:09could really support them
12:10and change the way
12:11they were doing business.
12:12And this we know
12:13how to do
12:14because we know
12:15deeply their business.
12:17So that's really
12:17the second thing
12:18is how we really fit
12:21to our business.
12:23The third thing
12:24we wanted to do,
12:25we did,
12:26is partnering
12:28with some
12:29key tech partners.
12:31because for the same reason
12:32we cannot invent
12:34everything at Carrefour
12:35and we will never have
12:37the investment,
12:38the capacity,
12:39the capabilities
12:40to do everything.
12:41We need to partner
12:42and the way we partner
12:44is to learn from them.
12:46So how do they help
12:47our teams
12:48and I see some of our teams
12:49who have become
12:50the experts
12:51in their sector
12:52thanks to these partnerships.
12:54The only rule
12:55to partner with Carrefour
12:56is to be able to open,
12:58to train our teams,
12:59to bring the expertise.
13:01and the other way around
13:02we also share the results.
13:04We share the feedback
13:05from our customers.
13:07We open the doors
13:08to have direct feedback
13:09from our teams
13:10and customers.
13:11So partnering
13:12is at the core as well.
13:14And then you have
13:15the three ingredients
13:16that I mentioned.
13:17Building the foundations,
13:19be very close
13:20to our business
13:21and having the right partners.
13:24Is it the three ingredients
13:26that really made us
13:27go to GNI so fast?
13:32It's not enough.
13:33I want to tell you
13:35an anecdote about that.
13:36We were in February,
13:38we were with Google.
13:39We were having
13:40a very bright strategic session
13:42about the partnership.
13:44What should we do
13:45in the upcoming year?
13:46What would be the priorities
13:48and blah, blah, blah.
13:49Very interesting session.
13:51Lots of ideas.
13:53And at the end
13:54of the meeting,
13:56I just started
13:58to be very impolite
13:59like a retailer can be.
14:01Like putting on the table
14:02a very bad word.
14:03Like, what about ChatGPT?
14:05What shall we do?
14:07And then the answer was,
14:10don't worry about it.
14:11We at Google,
14:12we will do much better
14:14than ChatGPT.
14:16and I told them,
14:18I'm sure you will,
14:19but I don't care.
14:20I sell potatoes.
14:21I don't need
14:22the very latest
14:24LLM model, whatever.
14:25What I need
14:26is to go fast
14:27because every day
14:28I'm just losing,
14:29losing, losing money
14:30if I don't go fast.
14:31So we decided
14:32to change the whole story.
14:34We decided
14:35the next meeting
14:36would not be
14:37one month later
14:38with new priorities.
14:39We decided
14:40to focus our energy
14:41on Gen AI
14:43and building solutions.
14:44And that was really
14:45the kickstart
14:46is I came,
14:48I really became
14:49the retailer I was.
14:50I need to go fast
14:51because every day
14:53I'm just selling potatoes
14:54and every day
14:55that I'm not developing
14:56new solutions,
14:57I'm just losing money.
14:59And this really is
15:01what made us
15:02come out
15:02with some key solutions
15:04and I want to share
15:05with you
15:05the first solution
15:06we came out with
15:08and I guess
15:09because they are
15:09discussing together
15:10it's not working
15:11the way it should be.
15:12Problem of network.
15:12But the first solution
15:14we came out with
15:16is the chatbot
15:17Opla
15:18and maybe
15:19the chatbot
15:21will work behind me.
15:23It's always very
15:26risky to do
15:27this kind of demonstration
15:28but you can see
15:29the chatbot
15:30it's basically
15:31helping our customers
15:32to answer
15:34to the first question
15:35they have
15:36and you have
15:36and I have
15:37every day
15:37which is
15:38what should I cook
15:39for dinner
15:40and even more
15:42what should I cook
15:43for dinner
15:44with a family of four
15:45and with 20 euros
15:46and please help me
15:47and people were coming
15:48to CAFOR
15:49and now they come
15:50to Opla
15:51and that's the first
15:53very broad solution
15:54on GNI.
15:56and then the second one
15:58which you can see here
15:59at Viva Tech
16:00and we will demo
16:01is just behind me
16:03which looks like
16:05another campaign
16:07of CAFOR
16:07one of these
16:09thousands of campaigns
16:10that you see
16:12every day
16:12CAFOR claiming
16:13that we have
16:14because it's true
16:15we have the best products
16:16with the best prices
16:18for the best customers
16:19with the best promotions
16:20and we do that
16:21every day
16:22so the only difference
16:23with this one
16:24is also true
16:26it's I think
16:28probably as good
16:29as what we do
16:29which is fairly good
16:31because we're not marketers
16:32the very big difference
16:33is it was 100% produced
16:35with AI
16:36with the AI of Google
16:38so now I want to show you
16:39how we do that
16:40and Nicolas
16:42will take the risk
16:43to demonstrate it
16:44in real time
16:45right Nicolas
16:46before short video
16:47a short video
16:48and then you will
16:49take the risk
16:49and then I take the risk
16:58Carrefour a created
16:59with Google Cloud
16:59an experimental tool
17:01that uses
17:02the generated artificial intelligence
17:03to make an omni-canal communication
17:06allowing the access
17:07to the good quality products
17:08and at the good price
17:10thanks to this tool
17:11we use the artificial intelligence
17:13to create scenarios
17:14for our communication campaigns
17:21so now we're trying to use
17:23we will try to use
17:24this new marketing studio
17:27creating in a real time
17:28a campaign
17:30hopefully it will work
17:31hopefully we will come out
17:32with a campaign
17:33Nicolas
17:34yeah it works
17:35oh fucking awesome
17:35ok
17:36yeah ok
17:37just like text them over
17:39yeah I'll text them over
17:39what you see there
17:41it's a marketing studio
17:42we've built
17:43in four weeks
17:45and the objective
17:46is to generate
17:47in a second
17:48an overall
17:49and holistic
17:50marketing campaign
17:51when I say marketing campaign
17:53it's basically
17:54every asset
17:55of a marketing campaign
17:56for Carrefour
17:57social networks
17:58so you will have
17:59YouTube script
18:00etc
18:00you will have
18:01Twitter script
18:02you will have
18:03even write your script
18:04etc
18:05with video
18:05and with image
18:07not video for now
18:09just
18:09filling some criteria
18:11so basically
18:12you said
18:12it's a
18:13between 20 and 25 years old
18:16people I want to target
18:17you say
18:18the campaign
18:19will last two weeks
18:20I want to be around
18:22inflation
18:22I want to be
18:24more on
18:25on social network
18:26than radio
18:27etc
18:28and in generate
18:29a campaign
18:30in a second
18:31now
18:31and it's not
18:33because of
18:33of tool
18:35it's because of network
18:36let's see if it works
18:37so for instance
18:39here you will
18:39you can see
18:40different campaigns
18:42if I click on
18:43Carrefour products
18:45of course
18:51that's why it took
18:52four weeks only
18:52it still goes wrong
18:53sometimes
18:54it's because
18:54we did not think
19:05thank you
19:07so just get back
19:08on what I'm saying
19:09so you can click on
19:11Carrefour products
19:12and you have
19:13sub campaigns
19:13here
19:15if you click
19:16for instance
19:17on
19:18organic food
19:21you have all
19:23the assets
19:24of the campaign
19:24that has been
19:25generated
19:26in a second
19:26so
19:27Facebook posts
19:28Twitter posts
19:29even with the image
19:30YouTube scripts
19:32Facebook
19:34Instagram
19:35radio scripts
19:36Twitter
19:37and Google Ads
19:38of course
19:39last but not least
19:42that's not perfect
19:43sometimes
19:44sometimes when you
19:44generate humans
19:46you have seven fingers
19:47you have 11 fingers
19:49on the same end
19:50but that's the first milestone
19:52and there will be another one
19:55of course
19:55we will improve
19:56the marketing studio
19:59adding some video
20:00even maybe producing
20:03some marketing plan
20:05not campaign
20:05but plan
20:06etc
20:06and
20:07that's the first step
20:10should we
20:11take the risk
20:12of generating
20:13LOD tweets
20:14yeah
20:15let's try to tweet
20:15with Google Marketing Studio
20:17anyhow
20:17I will
20:18ask Google
20:19if anything happens
20:20so I know
20:22they will
20:22support me
20:23on my tweet
20:26so we still
20:27have to work
20:28to do
20:29if I want to
20:30keep developing
20:32this tool
20:32actually
20:33it's being used
20:33right now
20:34by our marketing teams
20:35they are learning
20:37to launch
20:38their own
20:39Gen AI campaigns
20:40and are using it
20:41as you can see
20:41the UX
20:42is really being
20:44designed with them
20:45so let's try
20:46for a tweet
20:48let's try for a tweet
20:49so
21:00I need to give you
21:01my brief
21:02brief
21:03the sentence
21:04so the simple
21:05brief would be
21:06that
21:07K4 can help
21:12K4 can help
21:13its customers
21:15I need to have
21:16a robot
21:16just to translate
21:17but that would be
21:18can help
21:18its customers
21:20to fight
21:21inflation
21:24thanks to
21:24technologies
21:25let's find
21:26what should be
21:27the answer
21:40now there's time
21:45Google told me
21:46you should never
21:47do real-time demos
21:48it doesn't work
21:49even at Google
21:51so
21:53I need to try
21:54I'm K4
21:54never mind
21:55I sell potatoes
21:56I try to do tweets
21:57and sometimes it works
21:58sometimes it does not
22:01let's try again
22:16to fight inflation
22:20technologies
22:26ah
22:27it works
22:29so
22:29here you can
22:32target audience
22:36audience
22:37what should I do
22:38okay next
22:39channel
22:40summary
22:46problem
22:48problem
22:49problem
22:49of networks
22:55so
22:55maybe
22:56we will try it
22:58with Google
22:59to
22:59do the tweet
23:01I think
23:01you know
23:02Google
23:02he's just tweeting
23:03on his phone
23:03trying to
23:04show that it works
23:07so the idea
23:08for us
23:09is really
23:10is it about
23:11replacing our
23:12marketing teams
23:14potentially
23:15one day
23:16about replacing
23:17our agencies
23:19potentially
23:20I told that
23:21to publicists
23:21I don't believe
23:22it right now
23:24I don't care
23:25what we need
23:25is just to explore
23:27I think we will
23:28not replace
23:28we will just
23:30change the job
23:30of people
23:31it will not be
23:32about going back
23:32and forth
23:33with the agency
23:34it will be about
23:34doing that
23:35and challenging
23:36the robot
23:36here we have a tweet
23:37yeah we have
23:38finally
23:39much better
23:40so we have a tweet
23:41from the Google
23:42studio
23:43so that was the demo
23:44we wanted to share
23:45with you
23:45we are happy to answer
23:46to some questions
23:47we have one minute
23:49and as you can learn
23:50from here
23:50Nicolas was telling you
23:52test, test, and test
23:53sometimes we fail
23:55like the lock
23:56and sometimes we win
23:57like for Gen AI
23:58but at least we are
23:59going forward
24:00and we
24:02never forget
24:03that we need to be
24:04we need to invent
24:05the business of tomorrow
24:06it's not only about potatoes
24:07it's also about digital
24:11thank you
24:11thank you
24:12thank you
24:12thank you
24:12thank you
24:12thank you
24:12thank you
24:12thank you
24:13thank you
24:13thank you
24:14thank you
24:14thank you
Commentaires