- il y a 18 heures
From Sports to Scale: Rethinking Data-Driven Engagement
Catégorie
🤖
TechnologieTranscription
00:00Merci et bienvenue.
00:02Je veux juste faire ce qu'on va parler de ce qu'on va parler de ce matin.
00:07Sports ont toujours eu l'impression d'avoir l'impression d'être à l'économie, à la scale et à
00:13l'économie.
00:15Mais, dans le spectacle, sport a été l'une des plus puissants de tester l'engagement.
00:22Aujourd'hui, nous sommes unpackés comment les tactiques et playbooks pour les brands
00:27qui sont utilisés court-side, track-side et on-the-pitch
00:31pour reshape le futur de marketing partout.
00:34Donc, merci, Jonathan et David, pour joindre me ce matin.
00:37Jonathan, je veux commencer avec vous.
00:40IBM a leanedé dans le sport comme F1 et d'autres places, évidemment.
00:46Et d'un des sponsorships, je peux penser à beaucoup de tennis matchs
00:50où je vois l'IBM shield et logo.
00:56Mais, il a vraiment été une plateforme, comme je disais,
00:59pour l'innovation et la storytelling.
01:01Pourquoi pensez-vous que sport est un important catalyst pour cela,
01:05pour que l'on a général perspective et, vous savez,
01:07particulièrement d'une IBM lens ?
01:10Alors, first, Michael, merci pour avoir me.
01:13Nous avons été fortunate.
01:15Nous avons eu un partenaire avec Wimbledon, Augusta,
01:18pour le Master, et avec le U.S. Open Tennis,
01:21pour 30 plus years.
01:22Mais, je dirais, nous avons changé notre approche,
01:27récemment,
01:28par leaning en plus avec les organisations comme Ferrari,
01:31pour F1,
01:33et avec U.F.C.
01:34et avec Sevilla F.C.
01:37pour nous,
01:37c'est qu'il nous donne une grande opportunité
01:40pour nous donner notre technologie à la vie
01:42en un différent façon.
01:43Et pour moi,
01:44c'est pas juste de mettre un logo sur quelque chose.
01:47C'est vraiment de l'agir des organisations
01:50change la façon dont ils engageent avec leurs fans,
01:53change la façon dont ils operent
01:54en organisation,
01:55et, en plus,
01:57change la façon dont ils performent
01:59sur la course,
02:00sur la pitch,
02:01sur la track,
02:02et pour nous,
02:04c'est...
02:05c'est quand vous parlez de l'agir
02:06sur les plus difficiles,
02:07les plus gros topics de l'A.I.
02:09et large models,
02:11et ces sortes de choses,
02:12c'est peut être très abstracte.
02:14Et pour nous,
02:15c'est story-showing,
02:16c'est pas de story-telling.
02:17C'est vraiment de l'agir
02:18la technologie,
02:19c'est de l'agir,
02:20et de l'agir,
02:21c'est de l'agir,
02:22et c'est de l'agir
02:23de l'agir,
02:24quand ils sont les plus passionés,
02:26quand ils sont les plus passionés,
02:27c'est de l'agir.
02:28Mais je suis dit,
02:29c'est-à-dire que story-showing
02:30c'est vraiment de l'agir
02:32dans une autre façon.
02:34Il est exactement il même.
02:35David ?
02:35Il y a vraiment de dire,
02:36c'est qu'elle a été forte,
02:37is que je suis a fan
02:38de ces sports
02:40du temps de mon temps.
02:41Et l'intégration
02:43de IBM'es technologie
02:45dans le jeu,
02:48c'est-à-dire que cela,
02:49c'est-à-dire que c'est un peu
02:50differentiel de l'agir
02:52de l'agir,
02:52c'est-à-dire que c'est quelqu'un
02:52de l'agir,
02:54ce qu'est-à-dire,
02:55Je pense que l'IBM est partie des masters, partie du Wimbledon, partie du US Open.
03:02Je pense que c'est une manière incroyable de toucher les consommateurs dans une manière très visceral.
03:09C'est ce que nous essayons de faire.
03:11Dans un monde où vous essayez vraiment de parler de haut niveau stuff.
03:14Et David, c'est un grand segue, parce que vous parlez de toucher les consommateurs.
03:18Et Zeta est known pour ses data-driven audience intelligence.
03:23Et toucher les consommateurs est une chose, mais toucher les consommateurs à l'heure,
03:29en l'air context, en l'air device, en l'air moment,
03:32c'est où la chose la chose.
03:36Je veux dire, vous avez 550 millions de personnes globalement opté dans notre data cloud.
03:45Most of them are here, at Divatec, by the way.
03:48We can look around the room and see, oh, I know you're in it.
03:51But no, in all seriousness, it allows us to ingest trillions of signals on individuals
03:59every moment of every day.
04:01And the ability to synthesize that into first and foremost intent,
04:07what does somebody intend to do next?
04:09Do they intend to get a new credit card, churn off a cellular platform,
04:13or get a white convertible in the state of Texas?
04:16The next question is, will they be approved for those products?
04:20The next question is, how do you best target them?
04:25So we have this theorem that is sort of like data over artificial intelligence equals intent.
04:32And the intent over the activation methodology equals intelligence.
04:38And if you can get to that level of intelligence, it becomes a flywheel.
04:44Because you might have one AI agent on the first component of the theorem,
04:49another AI agent on the second component of the theorem,
04:52and the third AI agent, which is really attribution,
04:55is feeding back real-time information for the other two.
04:59So I'll just end on, for this topic, the conversation,
05:03what I think a lot of people don't understand yet,
05:07is that everybody's started trying to get to one agent, right?
05:11How do we get to one AI agent doing one thing?
05:14What we have found at Zeta is when you string agents together,
05:19and yes, it's called an agentic workflow, but most organizations are not there yet,
05:24it's not one plus one equals two.
05:27The first agent fully integrated in communicating with the second agent
05:31is an order of magnitude greater.
05:35When you add the third AI agent, it's another order of magnitude greater.
05:42So you're 10 times greater at two and 100 times greater at three.
05:46We haven't gotten to four yet.
05:48We're working on it.
05:50What I was getting nervous about, my recent experience with agents
05:53didn't work out that well, so you're talking about different kind of agents.
05:56Yeah, we're not in downtown L.A.
05:58That was inside baseball.
05:59But guys, you're talking about the thing that is at the core, I believe,
06:06which is the separation of signal and noise.
06:09And this is really a question for both of you,
06:11because what's driving it is the signals you're getting,
06:14but if it's not properly interpreted, it's noise.
06:20Please.
06:22Okay.
06:23So, listen, I sort of joke,
06:26there's some data points that are worthless in 30 seconds,
06:30and some data points that are incredibly valuable for seven years.
06:34If you bought a new home,
06:36and you redo your mortgage every eight years,
06:39on the seventh year, our clients need to start marketing to you
06:44because you're going to be redoing your mortgage,
06:46moving, getting new insurance, buying a new car, all of those things.
06:50If you have purchased an automotive insurance policy,
06:55that signal is worthless.
06:57Yes.
06:58Because you've already purchased.
07:00So, to me, the difference between noise and true signals
07:04is timing of it and the ability to make it actionable.
07:10I don't know how you look at that, John.
07:11Well, I think for us, as we're looking at,
07:13when we're integrating, as you were talking about,
07:15if I just stick with the masters for a minute,
07:18each shot in the masters produces roughly 30 to 35 points of data.
07:25So, for us to be able to really get the insights out of it,
07:28it is really about lifetime feeding it in there
07:31and balancing out what's the most important piece there.
07:34So, if you think about 20,000 shots in a tournament,
07:38and each one of those produces 30 points of data,
07:41I can't do the math.
07:43It's 600,000.
07:45Thank you.
07:46Well, but, you know, being here at Viva Tech
07:48and the influence of publicists being felt,
07:51and you can't help it,
07:52and it's so warming,
07:55it reminds me of something Rashad Tabakawala,
07:58and many of you know Rashad, has said for years.
08:01Data is like oil.
08:04It's not worth much in the ground.
08:06It's when you take it out and refine it,
08:08and that's what we're talking about.
08:09The signal is one thing, the noise is the other.
08:13And a lot of organizations don't understand that.
08:16They think of data for data's sake.
08:19Really, if you're not synthesizing it
08:22by using, you know, machine learning,
08:25artificial intelligence,
08:26other layering points on top of the data,
08:31because back to Jonathan's point,
08:33yeah, it's great to get 30 data points per swing.
08:36Speed of swing might be an interesting topic in one moment.
08:41It's got to be relevant.
08:43But it's got to be relevant, right?
08:44Because the ability to put it together
08:47with that particular golfer's, you know,
08:51time on the fairway with his speed swing
08:54is going to be much more valuable to the audience
08:57than one data point just sitting in a vacuum.
09:00And if you take that out,
09:01just out of sports for a second,
09:04only 1% of enterprise data is in a model today.
09:09So companies are just getting started in this
09:12and really getting the insights
09:14out of the data that they've got.
09:15There's such a green field ahead.
09:18And one of the things I talk about,
09:20sorry to keep going, Michael,
09:21is 44% of the Fortune 100 largest companies in the world
09:26use the Zeta marketing platform to do this.
09:28is when we go into these companies,
09:31the vast majority of them have 16 different
09:34disparate data ecosystems that do not speak to each other.
09:39So the first thing we're trying to do
09:41for the largest enterprises in the world
09:44is get them to talk to one another.
09:45Well, just get all the data into one repository,
09:47like put it into a consumer data platform
09:49where it can begin to actually look at,
09:53the CRM system can talk to point of sale,
09:55which can talk to loyalty,
09:57which can talk to the credit card program.
09:59We should talk about Red Hat at some point, too.
10:00Yeah.
10:01That's a separate conversation.
10:02Well, we're happy to have that.
10:04Listen, we can have it now.
10:06You know, you'll have some partners,
10:07but it'll be okay.
10:08Might be a Reg FD issue, but...
10:10You think?
10:11But let me come back to this
10:13because with the increased privacy pressure
10:16and people are concerned always
10:19and more so maybe on this side of the pond
10:21as opposed to the other side of the pond,
10:26especially in these emotionally charged moments
10:28at sports and whatnot,
10:30are you feeling any restriction
10:32from that privacy pressure?
10:33Because trust and transparency
10:35are two of the critical words
10:37that we have to focus on all the time.
10:39For us, you know,
10:41we're a 113-year-old company
10:43and trust is at the foundation
10:45of who we are at IBM
10:47and so is transparency.
10:49And we really take it seriously
10:51and we've been focusing on it.
10:52But a fundamental of doing business with IBM
10:55is we don't own the data.
10:57We never take our clients' data.
11:00They keep the keys to their data.
11:02We bring our AI, our technology, to them
11:05and help them pull it all together
11:08the same way David's talking about it.
11:09But that allows a significant level of trust
11:13to be put in the place.
11:14But we take that very, very seriously.
11:17Without a doubt.
11:18I think any company that doesn't
11:20isn't going to win in the marketplace, right?
11:22So when we build a consumer data platform
11:25for our enterprise clients,
11:27that is their data.
11:29We enrich it with our data.
11:31But in a fashion,
11:33you're putting your hands on their keyboard,
11:34but it's their keyboard.
11:35Correct.
11:36And if they were ever to terminate us,
11:39I mean, I will point out
11:40we had a net retention rate
11:41of 115% last year.
11:46But if anybody was to fire us,
11:48their data would be their data.
11:49It would go back to them.
11:50Now, the interesting thing is
11:52when we do the data enrichment,
11:54where we match our data to their data,
11:57we remove the piece
11:58personal identifiable information
12:00and we replace it with a Zeta ID number.
12:04So Joe Smith becomes Zeta ID number 13578,
12:08much like Meta and Google do.
12:10And that's how we activate to that consumer.
12:12But when it comes back through the CRM system,
12:15they're able to de-anonymize
12:17who that was who purchased.
12:18It's funny, one of the simplest things
12:21we do for large enterprises is
12:22we start by looking at the entire marketing landscape
12:26and saying these 45 million people
12:29are your customers.
12:30We should not waste money marketing to them.
12:32And I don't know anybody else who can do that.
12:35We take about 50% of the marketing cost
12:38out before we even start coming up with intent
12:42by simply matching deterministic IDs
12:45to individuals in the ecosystem
12:47and not waste money on them.
12:49I remember the product Polydent as the example.
12:53You still use Polydent, don't you?
12:55Yeah, but the idea was 8% of the population was toothless
12:59and I'm spending a lot of money
13:01against 92% of the people
13:03who have no need for what I'm doing.
13:05And depends are another big thing for you.
13:07I haven't gotten there yet.
13:08Well, depending on the day.
13:11David, speaking of that,
13:13and, you know,
13:15the biggest word for everybody
13:18should be metrics.
13:19Is this working?
13:20Am I getting the return on my investment?
13:24What is your impression
13:27as to what is broken
13:28from a consumer's perspective,
13:32meaning a brand in this case?
13:34I think, and Jonathan, I'm sure,
13:36has an opinion on this as well.
13:38I would tell you that the vast majority
13:41of attribution is done
13:43on what I call last-click attribution,
13:47where you might run 12 ads
13:49and the 12th ad that consumer purchases,
13:53you would look at it as an enterprise
13:55and say the first 11 were a total waste of money,
13:57stop all of them.
13:58And the 12th was the most valuable ad unit
14:01you've ever run.
14:02when you try to replicate that,
14:04it doesn't work.
14:06So one of the things
14:07that we've developed
14:08through our attribution capabilities
14:10is literally a touchpoint
14:12through every component of the journey
14:14for the tracker,
14:15for everything.
14:17And if you,
14:18by way of example,
14:20we map into meta.
14:22There are not a lot of companies
14:23who do that.
14:23We pull no data out of meta,
14:25but we can target into meta.
14:27And when a consumer clicks on a meta ad
14:30and comes out,
14:30we can then track them
14:32and we can do the attribution on that.
14:34And we can then see
14:35if a person who clicked on a meta ad
14:37and then we run a connected TV ad that night
14:39calls their call center the next day.
14:42We're looking at the return on investment
14:44across all of that.
14:46And I think until most organizations
14:49can get to true attribution modeling,
14:53where the investment is done that way,
14:55because the other large companies just say,
14:57oh, we're going to spend 15%
14:59of total revenue on marketing.
15:01And if we can get that to 14.5,
15:04we make another billion dollars
15:05in net income this year.
15:07It's going to have to get deeper
15:09and it's going to have to get
15:10into a more micro environment.
15:12Well, you know,
15:13we would go down the wrong road here,
15:15but your point is well taken
15:16because I've asked this question frequently
15:20of chief marketing officers
15:22and people in that role.
15:25What's success for you?
15:27How does your CEO look at the success
15:29of what you're doing?
15:31And the core question I always ask is,
15:34does your CEO really understand your job?
15:37You know, what?
15:37I'm lucky I work in a company
15:39where my CEO genuinely understands
15:41and values what marketing
15:43and communications brings to the business
15:46and recognizes that we can't,
15:47as a business,
15:49we can't move forward
15:50without the work that we do.
15:52And he's also very clear to me
15:53and my whole team.
15:56Your sales is your number one client
15:59and you need to be fully aligned.
16:02And our head of sales and I,
16:04there's 99 plus percent of alignment
16:08before we even talk
16:09because we know where we're going
16:10as a business.
16:11But the point that you were talking about,
16:14about how you tie all that together
16:16from a B2B perspective
16:17in the tech space right now,
16:19it's getting even more challenging.
16:20Correct.
16:21I'm looking at buying groups
16:22of 12 to 19 people
16:24for every different buying decision.
16:27And that group changes
16:28in every company
16:30every time they make
16:31a new buying decision.
16:32You might get three or four that stay,
16:34but this could be
16:35a six-month buying decision,
16:36an eight-month,
16:37and they drop in and out of the funnel
16:39and engage at different times.
16:40We should talk
16:41because we build clusters
16:42for enterprises
16:43that allow us to track
16:44who's coming in and out
16:45of the cluster
16:46as they're in and out.
16:48And we're seeing
16:49all of their online research
16:50and searching
16:51in addition to that.
16:54Jonathan,
16:55AI is obviously
16:56a big part
16:57of the DNA of IBM.
16:59Yep.
17:01How do you see it
17:02playing out
17:03in terms of audience engagement
17:06and really sort of
17:08at the highest level,
17:09not just in sports,
17:10but across the board?
17:11I know we wanted to focus
17:12on sports this morning,
17:14but...
17:14For me,
17:16there are three legs
17:17to the stool on this one.
17:19First is the content
17:20that we're creating.
17:21And I think AI,
17:22we've already seen
17:22it's making a big difference
17:24in the content
17:24we are creating
17:26and the speed
17:27at which we can create.
17:28And it's giving our creatives
17:30the ability
17:31to be more creative.
17:32Previously,
17:33before we started rolling out
17:34AI in the organization,
17:35there's a team
17:36of about 300 designers
17:38inside of IBM
17:38on my team.
17:3980% of their time
17:41was on derivative assets.
17:42So AI is allowing them
17:44to be much more creative
17:45because it's giving them
17:46more time.
17:47Second is about placement.
17:49We're getting much more focused
17:51and targeted
17:51with our placement,
17:52not just in buying ads,
17:56but also what events
17:57we're showing up at
17:58and how we show up
17:59at those events
18:00and the stories we're telling.
18:01We're doing half the number
18:02of events today
18:03that we were doing
18:03two years ago.
18:04we're getting 5%
18:06more client signals
18:07and client interests
18:07out of half the number.
18:08That's interesting.
18:09So that goes to efficiency,
18:12obviously.
18:13Yeah.
18:13You're saying you're doing
18:145% less?
18:17No, we're doing
18:17half the number of events,
18:19but we're getting
18:205% more client interest
18:22and signals
18:22out of half the events.
18:24And then the last piece
18:25is really about
18:26figuring out
18:27where you need to be,
18:29when you need to be,
18:30and how you need
18:30to deliver it.
18:31Do you attribute
18:31that to AI?
18:33Yeah.
18:34A lot of that
18:35goes to AI.
18:36But then the last piece
18:37is making sure
18:38we're delivering
18:39personalized
18:41and relevant content.
18:43Because I have the view,
18:44a lot of people
18:45want to talk about
18:45personalized,
18:46I think it's only
18:47half the journey.
18:48But I do have to
18:50inject something here,
18:51because I used the expression
18:53with the question
18:53I asked you, David.
18:54Right place,
18:55right context,
18:56right device,
18:57right person,
18:57all the right things
18:58that we all know
18:59the holy grail.
19:01But I always feel
19:02I have to add this
19:03in this conversation.
19:05We are in the business
19:06of marketing
19:06on a very high level.
19:09And there is room
19:10for surprise and delight.
19:12There is room
19:12for that person
19:13who wasn't in the market
19:14for that car
19:15or didn't, you know,
19:17just need to refinance
19:19their home.
19:20But they see
19:21an interesting offer
19:22and they say,
19:22you know what,
19:23I didn't think
19:23I wanted a car,
19:24but now I do.
19:25Or I didn't think
19:26I needed a mortgage,
19:27but boy,
19:27at this rate,
19:29sorry, please.
19:30And literally,
19:31you're always going
19:32to have brand-based
19:34marketing that will
19:35handle that, right?
19:36So, you know,
19:38even at Zeta
19:38and I know at IBM,
19:40we're not suggesting
19:41all brand-based
19:42marketing go away.
19:44We're simply saying
19:45you should be getting
19:46it to clusters
19:47of people
19:48that have a higher
19:49level of intent
19:50to purchase your product
19:51and a higher propensity
19:53to be able
19:55to buy it.
19:56By way of example,
19:57just to put it out there,
19:58we have one
20:00insurance company
20:01that's a customer
20:02in the United States.
20:03They're one of the largest
20:04automotive insurance companies.
20:06They have gone
20:07from what started out
20:08as a $500,000 test
20:10on our platform
20:11to what will probably
20:12be a $100 million
20:13client this year.
20:16And we've lowered
20:17their cost
20:18to create a policy
20:19by 356%
20:22over the five years
20:24that we've worked
20:24with that client.
20:25I hope it's my carrier
20:27so the flow-through
20:28of that savings
20:28will reduce my premium.
20:30Well, their profit's
20:30gone up a lot, too,
20:31so their stock is going up.
20:32And then there's that.
20:32I'm not sure how much
20:33they're passing through.
20:35But what I would tell you
20:36is that the algorithms
20:38get smarter.
20:40Yep.
20:40that doesn't mean
20:41this same company
20:43that spends,
20:43pick a number,
20:44$2 billion a year,
20:46they might spend
20:46$100 million with us.
20:48The other $1.9 billion
20:50might be going to linear
20:53and they might be
20:55building brand campaigns there.
20:56I just want to add
20:58one postscript
20:58to what you're saying, David.
21:00This was something
21:01that we discovered
21:02with a financial services company
21:04and I came up with a word,
21:06I don't know,
21:06now seven years ago,
21:08but it's more relevant now
21:10than it was then.
21:11And that was the combination
21:12that I put together
21:13of brand and performance
21:14and I called it
21:15brand-formance.
21:16That's a great word.
21:18The idea of brand marketing
21:19shouldn't be out here
21:21without the benefit
21:22of the data
21:23and all the things
21:24that make performance marketing.
21:25Why not?
21:27And it's a great point
21:29and to bring it back
21:30to the original topic
21:31of sports.
21:32Oh, sports!
21:33For my view,
21:34part of the reason
21:35we're where we are
21:36in these sports,
21:37with these sports franchises
21:39is because we can do
21:41surprise and delight there.
21:42We can show it
21:43to people
21:44at that higher level
21:45that weren't expecting,
21:47hey, I can use that
21:49in my company.
21:49Well, and by the way,
21:51we work with the NBA,
21:53Major League Baseball,
21:54UFC.
21:55I mean, we work with
21:56most of the main leagues
21:58at Zeta
21:58and I would tell you
22:00they want surprise
22:01and delight, right?
22:02So they're looking
22:03to get better data
22:05for their partners,
22:07their big sponsors,
22:08so they can drive
22:09sales there.
22:11A thousand percent.
22:11And that's one of the places
22:13we really play.
22:14And when you think
22:15about that,
22:16the phenomena
22:17that still exists
22:18in the States,
22:19the network up front,
22:20which just happened
22:21in May,
22:22and many of our...
22:23I had two people
22:23cancel dinner on me
22:25because of that.
22:25Exactly.
22:26Because they had to be
22:26at the network up front,
22:27and I was like,
22:28they still do those?
22:29Yeah, well,
22:30but when I was asked
22:30this year,
22:31interestingly enough,
22:32what was the takeaway
22:34from the up front,
22:36I said it reminded me
22:37of that scene
22:38in The Graduate
22:38where at his graduation,
22:41the friend came up
22:42and said,
22:43I have one word for you,
22:43plastics.
22:44My answer to the press inquiries
22:47was I have one word
22:48for you from the up front,
22:49sports.
22:50And I might modify it
22:51to say women's sports,
22:52but there wasn't much else
22:53that was talked about
22:54because the idea
22:56of programming,
22:57we all know,
22:57changed,
22:58but sports is center.
23:00And, you know,
23:01Michael,
23:01we did a big project
23:03for the Olympics
23:04last year
23:05when the Summer Olympics
23:07last ran
23:08on how to drive engagement
23:10for female sports
23:12with females
23:14in the United States
23:15of America.
23:16We came up
23:17with a very simplistic
23:18campaign and audience.
23:20We were able
23:21to increase
23:22using the Zeta
23:23marketing platform
23:24by 26% viewership
23:27for women's sports
23:29by women.
23:30As a subset
23:30of the Olympics
23:31by women.
23:33And it was just,
23:34it was not even
23:35that hard.
23:36That's great.
23:36Because the demographic
23:38was so interested in it,
23:40they just didn't know
23:40it was available.
23:42Well, guys,
23:43according to the clock,
23:44I think we've run out
23:45of time.
23:46There's a lot more
23:47we could cover,
23:48but I really appreciate
23:49the time,
23:51appreciate everyone's
23:51attention.
23:52I want to thank you,
23:53Jonathan and David.
23:54Thank you.
23:55You did a great job.
23:56Thank you.
Commentaires