Latest / The Yield Doctor with James Deaker / AI Won’t Fix Bad Data — Identity Resolution & Enrichment Drive Revenue (Brian Silver)

Transcript

AI Won’t Fix Bad Data — Identity Resolution & Enrichment Drive Revenue (Brian Silver) · The Yield Doctor with James Deaker

• 0:00
  In digital media today, almost everyone
• 0:04
  agrees on one thing. Data is the most
• 0:07
  valuable asset in the business. And yet,
• 0:10
  it's also the asset that creates the
• 0:12
  most confusion, fragmentation, and risk.
• 0:16
  Publishers are drowning in signals that
• 0:18
  they can't activate. Advertisers are
• 0:21
  demanding proof of outcomes that data
• 0:23
  can't always support. And across
• 0:25
  industries, teams are trying to prepare
• 0:27
  their data for an AIdriven future
• 0:31
  without really even agreeing on what
• 0:33
  good data is. Today, I want to unpack
• 0:35
  those challenges with someone who has
• 0:37
  had a front row seat on how data
• 0:41
  identity and measurement work in the
• 0:43
  real world. I'm joined by Brian Silva,
• 0:46
  EVP of global marketing solutions at
• 0:49
  TransUnion.
• 0:50
  Brian has spent decades in senior
• 0:53
  leadership roles across digital
• 0:55
  ecosystem including Oracle, Live Intent
• 0:59
  and IA. And we first worked together 10
• 1:02
  years ago at Yahoo. Brian, I'm so glad
• 1:05
  that we finally made this conversation
• 1:07
  happen.
• 1:07
  >> Thank you, James. It's great to see you
• 1:09
  again. And uh yeah, I can't believe
• 1:11
  Yahoo was uh 15 years ago that we first
• 1:14
  met. Crazy. Crazy.
• 1:16
  >> A lifetime in in our world. Hey, before
• 1:19
  we get into the broader industry
• 1:21
  dynamics that we're going to go deep
• 1:23
  into, can you give a quick overview of
• 1:25
  the TransUnion Marketing Solutions and
• 1:28
  your role there just so we have enough
• 1:30
  context so people understand your
• 1:32
  vantage point?
• 1:33
  >> Sure. Sure. So, so again, thanks for
• 1:35
  having me. Really uh appreciate the uh
• 1:37
  conversation. The time of the
• 1:39
  conversation is is incredible. So at uh
• 1:42
  TransUnion, I am the global head of
• 1:44
  marketing solutions, which is all of the
• 1:46
  marketing and and advertising solutions
• 1:48
  across identity, audience, and
• 1:50
  measurement. Uh all underpinned by a
• 1:53
  global technology called One True in
• 1:56
  which, you know, in order to think about
• 1:57
  it, all of our data is now in a single
• 2:00
  container on grid for the applications
• 2:02
  within each of those silos to be able to
• 2:04
  pull from. And I think what TransUnion
• 2:06
  is is solving for especially in
• 2:08
  marketing solutions is the ability for
• 2:10
  us to resolve uh to a single identity.
• 2:13
  Um it's so important today, you
• 2:15
  mentioned it in in your opening as far
• 2:17
  as fragmentation is concerned. We see
• 2:19
  that most marketers today are dealing
• 2:21
  with 16 disperate different systems in
• 2:24
  order for them to follow their customers
• 2:27
  or potential new customers across their
• 2:29
  journey, let alone even measure on them.
• 2:31
  And so, you know, we're we're solving
• 2:33
  for that and and we're, you know,
• 2:34
  excited that uh we'll we'll continue
• 2:36
  that part of the conversation.
• 2:38
  >> Excellent. Now, you you mentioned
• 2:40
  marketers and their data challenges. You
• 2:41
  also work with publishers. How different
• 2:44
  are the needs between advertisers and
• 2:46
  publishers in practice?
• 2:48
  >> You know, it's interesting. I mean, for
• 2:49
  a long time, right, there really is no
• 2:51
  difference between an advertiser and
• 2:53
  publisher. And what I mean by that is
• 2:55
  specific. So, if we go back to the
• 2:57
  equation of advertising, of course,
• 2:59
  they're different. The publisher cares
• 3:00
  about yield. The advertiser cares about
• 3:03
  spending the least amount of money for
• 3:04
  the highest ROI, right? With efficiency.
• 3:07
  You have to remember that piece. Um, but
• 3:09
  I think as time has gone on, you know,
• 3:11
  if you're on somebody's site, yes,
• 3:13
  they're a publisher. Let's say it's
• 3:14
  Nordstrom's, right? But all of a sudden,
• 3:16
  if you are, you know, out in the
• 3:17
  universe, Nordstrom wants to understand
• 3:19
  me as a consumer sitting on their site,
• 3:21
  but they also want to understand other
• 3:23
  attributes or other, you know, habits
• 3:25
  that I may have. And so the difference
• 3:27
  between the collected data or the
• 3:29
  persona of an individual that Nordstrom
• 3:31
  understands whether or not they're
• 3:33
  publisher or advertiser really has come
• 3:35
  together. I think where it gets very
• 3:37
  interesting is now in the world of AI
• 3:40
  where the utilization of somebody's
• 3:42
  first party data or the utilization of
• 3:45
  the understanding of a consumer as they
• 3:48
  travel around the internet. the ability
• 3:50
  for somebody on the publisher side or
• 3:52
  the advertiser to then reconcile to that
• 3:54
  to be able to perform uh against uh that
• 3:57
  consumer has become even more
• 3:59
  interesting. So I I want to get into
• 4:01
  that in particular into sort of identity
• 4:03
  mapping, but let's let's step back and
• 4:05
  just talk about the fragmentation
• 4:07
  because I still see a lot of
• 4:09
  fragmentation across the value chain,
• 4:11
  especially in CTV and streaming and
• 4:14
  especially when you layer in different
• 4:15
  regulatory environments in different
• 4:17
  countries. From your perspective, how do
• 4:20
  companies think about data consistency
• 4:23
  and measurement when the ecosystem is
• 4:26
  still so fragmented? Well, I mean, I
• 4:28
  think that's where they're looking for a
• 4:29
  partner like TransUnion, right? Where
• 4:32
  they they understand that the ability
• 4:33
  for us to resolve that identity into a
• 4:36
  single uh persona is incredibly
• 4:38
  important. And then to be able to then
• 4:40
  understand how to amalgamate more and
• 4:42
  more of those personas together, even in
• 4:45
  an anonymized way for them to be able
• 4:47
  to, you know, find their new customer,
• 4:50
  measure against those customers, etc. I
• 4:52
  think that's that's that's important. I
• 4:54
  think that you know when you talk about
• 4:56
  CTV channels if you talk about you know
• 4:58
  every other channel that's out there yes
• 5:00
  there are fragmented point solutions
• 5:02
  that are there when I talk to our
• 5:04
  customers the most important thing that
• 5:06
  they tell me is that the value
• 5:07
  proposition that TransUnion or others
• 5:10
  like us can provide is a foundational
• 5:12
  element to their business. a way for
• 5:15
  them to understand the identity, a way
• 5:17
  for them to understand how to measure
• 5:19
  against their campaigns and
• 5:21
  continuously, you know, pull that into
• 5:23
  almost like a flywheel into a loop
• 5:25
  where, you know, the outcomes, right,
• 5:28
  are as critical as the inputs. And
• 5:31
  today, that fragmentation makes that
• 5:33
  very difficult unless you're dealing
• 5:34
  with or working with a partner like
• 5:36
  TransUnion. So one way that people try
• 5:40
  to deal with fragmentation is to ensure
• 5:42
  that everybody's at least speaking the
• 5:44
  common language. But as an industry we
• 5:47
  don't have a single common identifier
• 5:49
  and obviously cookie deprecation is you
• 5:52
  know slowly uh deteriorated in value.
• 5:56
  You work with all of the major identity
• 5:58
  solutions ramp ID ID5 and others. How
• 6:02
  big a painoint is identity today and how
• 6:05
  do you see that landscape evolving?
• 6:08
  >> Yeah, look, I think it's still to play
• 6:09
  out. I think AI has a big piece to play
• 6:12
  in this uh as far as activation is
• 6:14
  concerned, but I think you're right. I
• 6:16
  think the the story is still being
• 6:18
  written today. I think that just having
• 6:20
  interoperability with all of the
• 6:22
  alternate IDs is an important part of
• 6:24
  the story, but I'm not sure that it's
• 6:26
  the end of the story. I don't not sure
• 6:27
  that that that actually's been written
• 6:29
  yet. And I think that people are
• 6:31
  utilizing AI to to see where um and how
• 6:34
  this could become more efficient as time
• 6:36
  goes on.
• 6:38
  >> So um you've said AI and I've I've got
• 6:41
  to let you get there now. So we we can't
• 6:43
  talk about the
• 6:44
  >> we can't have a conversation today
• 6:45
  without you know the utilization of uh
• 6:47
  of this technology. It's it's sort of
• 6:49
  all consuming at this point. So do you
• 6:52
  first of all do you think that companies
• 6:54
  have really come to terms with the sort
• 6:56
  of quality structure and governance of
• 7:00
  the data given that AI systems are now
• 7:02
  depending on it to be effective. So
• 7:05
  quick answer is no. I think everyone is
• 7:08
  is starting to realize that this is not
• 7:10
  just another industry uh cliche uh you
• 7:15
  know like the year of mobile was many
• 7:17
  decades ago or measurement is a real
• 7:20
  thing. you should go after it, right? AI
• 7:22
  is not um just another feature or
• 7:25
  functionality. It it's a way of life
• 7:27
  that everybody has to um you know step
• 7:30
  up and recognize. So that's number one.
• 7:32
  Number two, right, I have already
• 7:34
  started to put out thought leadership
• 7:36
  piece around the utilization of AI. The
• 7:39
  cliche of garbage in equals garbage out
• 7:42
  is as real with AI as it's ever been
• 7:44
  with any other uh movement and and
• 7:46
  evolution. And so we believe very
• 7:48
  strongly that you have to have the
• 7:51
  highest quality and scale of data. Um
• 7:53
  that's important. We just did a um a um
• 7:57
  a use case with Actible and one of the
• 7:59
  things that came out of the Actible
• 8:01
  study was is that at least 60% of all AI
• 8:04
  initiatives today will fail because of
• 8:06
  the lack of quality of their data or
• 8:08
  lack of signal within their data itself.
• 8:11
  And so you have to have the right types
• 8:13
  of data. you have to have the right
• 8:15
  ability to then consolidate that data so
• 8:18
  that you when you're utilizing AI and
• 8:20
  we'll just talk about it in two ways
• 8:21
  either efficiency or propensity. So on
• 8:24
  the efficiency side, I think a lot of
• 8:25
  people have really figured out search
• 8:26
  and discovery, right? What's the ability
• 8:28
  for me to add some form of, you know,
• 8:31
  agent that's sitting on top of my UI so
• 8:33
  that I can discover data or my users can
• 8:36
  discover data much quicker and much
• 8:37
  faster. Great. I think that's a perfect
• 8:39
  start. But then to be able to then use
• 8:42
  that information for propensity
• 8:44
  purposes, right, solve real problems. I
• 8:47
  think people are just starting to
• 8:48
  understand that that is something that
• 8:50
  needs to happen and needs to happen
• 8:51
  rather quickly. So, so let me follow up
• 8:53
  on that. If 60% of and was it the
• 8:56
  projects that fail or 60% of companies
• 8:59
  fail in their AI initiatives?
• 9:00
  >> Companies/ projects, same same thing.
• 9:03
  >> Of the 40% that presumably are not
• 9:06
  failing, what are they doing differently
• 9:08
  that the 60% is not of deliberately
• 9:11
  open-ended questions.
• 9:12
  >> No, no, I get it. So, I think that um
• 9:15
  again quality of data and the data
• 9:17
  signal that they get is incredibly
• 9:18
  important. Right. So I notice I didn't
• 9:21
  say the size of the 40% of each of those
• 9:23
  opportunities. A lot of them are
• 9:25
  relatively small. People are just
• 9:26
  starting to figure this out in little
• 9:28
  microcosms, right? It takes it takes a
• 9:31
  while for it to scale to the size and
• 9:34
  the scope of what is needed in order for
• 9:37
  these to become more industry-wide and
• 9:40
  accepted applications or uses of AI. And
• 9:42
  so again where where we said it's very
• 9:45
  strong in the fact that we believe that
• 9:47
  our data being you know um being brought
• 9:51
  to the table has been you know developed
• 9:54
  over the 40 plus years of TU's history
• 9:56
  and then the ability for us to apply the
• 9:59
  right type of utilization of AI for our
• 10:02
  customers in our in our in our sweet
• 10:04
  spot right in identity resolution in the
• 10:06
  ability for us to provide audiences and
• 10:09
  solve problems and I'll give an example.
• 10:10
  So, I was talking to a large OEM not
• 10:13
  very long ago, and one of the things
• 10:14
  they told me was is, hey, 50% of all of
• 10:18
  my customers today I already know will
• 10:21
  not come back in the future to buy this
• 10:24
  car brand again. So, I said, well, what
• 10:27
  if I have the ability to not only tell
• 10:28
  you which 50% of your customers will
• 10:31
  come back, but more importantly, what if
• 10:33
  I have the ability to provide you with
• 10:34
  data, audience segments, etc. to make
• 10:37
  sure that the other 50% come back you
• 10:39
  know are become new customers for you
• 10:42
  and their mind was blown they said you
• 10:43
  can do that I said we can do that and we
• 10:45
  we can do that in real time and so it's
• 10:47
  not just understanding like the
• 10:49
  resolution of the foundational elements
• 10:51
  that I think is important James I think
• 10:53
  we still have to be curious of what
• 10:55
  problems are we trying to solve for the
• 10:57
  customers for the marketers for the
• 10:58
  brands and that I think is critical
• 11:00
  while everybody is running to say hey I
• 11:03
  have this AI agent or I'm working with
• 11:05
  this application or whatever is it
• 11:07
  really get at the point of scale to
• 11:09
  solve customer problems and that's where
• 11:10
  that's where we're heading today and
• 11:12
  we're really excited about that.
• 11:13
  >> What are the themes in those that are
• 11:16
  using AI effectively? Uh is it that
• 11:20
  they're more focused that they're
• 11:22
  focused on smaller projects that they're
• 11:24
  experimenting quicker that they know
• 11:26
  their data better? I'm obviously
• 11:28
  throwing lots of stuff at you, but there
• 11:30
  people would have lots of different
• 11:31
  hypotheses for the companies that are
• 11:34
  succeeding in their AI projects. What
• 11:37
  are you seeing in in terms of
• 11:39
  commonalities across the companies that
• 11:41
  are being effective right now?
• 11:42
  >> Yeah, look, I think it's three, right?
• 11:44
  I've mentioned them before, but I'll
• 11:45
  categorize them explicitly, right?
• 11:47
  Identity resolution is key, right? The
• 11:50
  ability for you to be able to resolve
• 11:52
  into that identity is an incredibly
• 11:54
  important part. And I think if you're a
• 11:55
  company today and you're listening to
• 11:56
  this and you don't feel like you have
• 11:58
  the highest fidelity of resolution to
• 12:00
  your own customer data set or to future
• 12:02
  data sets, put a little effort in there
• 12:05
  because it's going to be worth the
• 12:06
  return on investment. The second is just
• 12:09
  data enrichment, right? The ability for
• 12:11
  the lack of depth or the lack of, you
• 12:13
  know, uh breath of the amount of data
• 12:16
  that you can then utilize, call it coal
• 12:18
  in the engine and then what energy
• 12:19
  output you're going to be getting. like
• 12:21
  that's where you know people need to
• 12:23
  understand that that data you know
• 12:25
  enrichment is is truly important and
• 12:27
  then I think last is connectivity right
• 12:29
  and so you know today there's a lot of
• 12:31
  companies out there that may not have
• 12:33
  the integrations deep into the walled
• 12:35
  gardens there's companies that don't
• 12:37
  have the breath of you know different
• 12:39
  publishers or marketers outside of the
• 12:41
  US and I think that globalization is key
• 12:44
  I think that connectivity is key and so
• 12:46
  that last piece is not just the ability
• 12:48
  to go and find your users there. But
• 12:50
  then how do you measure and bring that
• 12:52
  information back? Again, that flywheel.
• 12:55
  So that way you utilize your identity
• 12:57
  resolution. You're doing the highest
• 12:58
  quality of data enrichment and then
• 13:00
  you're measuring off of that so that
• 13:01
  you're continuously optimizing and
• 13:03
  making sure that not only are you
• 13:04
  solving your problem, but you're doing
• 13:06
  so at the highest level of efficiency.
• 13:08
  And I think that people are again just
• 13:10
  starting to get into this notion, but to
• 13:12
  me categorically like those are the
• 13:14
  three things that are critical in
• 13:15
  today's market
• 13:17
  >> more broadly. So, we we've been focused
• 13:19
  on AI, but what is the biggest mistake
• 13:22
  that you see companies make with their
• 13:24
  data and identity assets? Is it
• 13:26
  governance? Is it the fact they're not
• 13:28
  structuring it correctly? Like them,
• 13:31
  again, deliberately open-ended question.
• 13:33
  >> Yeah, I know. And uh unfortunately, I'm
• 13:35
  going to keep bringing it back to the
• 13:36
  same point, which is, you know,
• 13:40
  I think that there's been a very
• 13:42
  interesting conversation in market for a
• 13:43
  while now. We'll call it the data war,
• 13:46
  right? between demand side and supply
• 13:49
  side. You mentioned it earlier. I think
• 13:51
  that people are starting to wake up and
• 13:53
  recognize that the supply side has a lot
• 13:56
  more power than it was giving itself
• 13:57
  credit for for a while. When people go
• 14:00
  to sites and they become customers,
• 14:02
  right, or they even become just known to
• 14:05
  uh the publisher side, the sell side of
• 14:07
  the universe, right, they have the
• 14:09
  ability to then have a relationship with
• 14:11
  that with that consumer.
• 14:13
  And I think that that's a critically
• 14:15
  important part to right how to solve the
• 14:17
  the the problem that you're mentioning.
• 14:19
  I think on the demand side, right, where
• 14:22
  they basically just said, "Hey, look, I
• 14:23
  know you want to do these things, but
• 14:25
  without my demand, you're not going to
• 14:26
  have the return that you need from a
• 14:28
  business perspective in order for for
• 14:30
  you to go out there and win." I think
• 14:32
  that we're reshaping what the
• 14:33
  relationship is between the demand side
• 14:35
  and the supply side. So, that's number
• 14:36
  one. So number two is
• 14:41
  there aren't that many publishers in
• 14:43
  this world. And even when we were at
• 14:44
  Yahoo, right, as one of the largest
• 14:46
  publishers that existed on the planet,
• 14:48
  we just didn't have enough regulatory
• 14:51
  regulated data for us to be able to, you
• 14:54
  know, continuously grow. You had to go
• 14:57
  have partnerships and you had to go, you
• 14:59
  know, increase your data size and your
• 15:01
  fidelity and resolution of that data in
• 15:03
  order for you to continue to drive your
• 15:05
  business forward. And so it's not a
• 15:07
  mistake that anyone is making, but
• 15:09
  everyone, you know, understands that,
• 15:11
  you know, you can't just do it alone.
• 15:13
  You're not going to have the amount of
• 15:15
  scale that you think you're going to
• 15:16
  need in order for you to continuously
• 15:18
  grow your business. You're not going to
• 15:19
  have the feature functionality set
• 15:21
  that's sitting in your technology stack
• 15:23
  in order to provide the right, you know,
• 15:25
  return against that audience or even
• 15:27
  provide the best customer experience for
• 15:28
  that audience. And so I think that the
• 15:30
  problems haven't changed, James. So I
• 15:32
  think we've been solving these types of
• 15:33
  problems for decades like you said. Um
• 15:35
  but I think it's really critical that
• 15:37
  people recognize that even though the
• 15:39
  problem hasn't changed, right? It's the
• 15:42
  partnerships that matter. It's the
• 15:43
  people that are thinking now about what
• 15:46
  the future of our industry is going to
• 15:48
  look like and how best to then capture
• 15:50
  the most amount of dollars in order to
• 15:52
  solve our customers problems.
• 15:54
  >> I want to come back to something that
• 15:55
  you said because it really piqu my
• 15:57
  interest. the the phrase that publishers
• 16:00
  maybe haven't been getting the credit
• 16:01
  for their data because I you know as you
• 16:03
  know I've been doing this a really long
• 16:05
  time and I've had a lot of publishers
• 16:06
  say my data is oh so valuable why does
• 16:09
  nobody recognize that and so I I want to
• 16:13
  understand where you're coming from
• 16:14
  there is it that the economics is
• 16:16
  starting to change that new business
• 16:18
  models are starting to evolve maybe that
• 16:21
  publishers instead of uh just using
• 16:23
  their data for targeting are actually
• 16:25
  starting to syndicate and use it in
• 16:27
  different ways Where were you going with
• 16:28
  that comment? Because it's it's so
• 16:30
  interesting to me.
• 16:31
  >> I mean, again, it's um it is interesting
• 16:34
  only because I have also spent most of
• 16:37
  my career on the supply side, right? And
• 16:39
  I think that um it was an unfair
• 16:42
  position to always feel like no matter
• 16:46
  how much fidelity and resolution I had
• 16:49
  on my customer set or even my expanded
• 16:51
  customer set, I couldn't go to a, you
• 16:55
  know, call it a PNG and say, "Hey, by
• 16:58
  the way, we really need you to spend
• 16:59
  money against this." Because they were
• 17:01
  like, "It's just not big enough." or
• 17:05
  yes, it's a nice slice, but you know
• 17:07
  what h how do you figure out the next
• 17:09
  step or the next evolution of this
• 17:11
  piece? And I think that the supply side,
• 17:13
  if you just look at the SSPs across the
• 17:15
  industry today, right, are really
• 17:17
  starting to figure out that there are
• 17:19
  alternatives to the way in which people
• 17:22
  monetize uh consumers. Uh there are
• 17:24
  alternatives to the way in which we can
• 17:26
  provide different experiences to those
• 17:28
  consumers uh in a much more intelligent
• 17:31
  way than we've been able to in the past.
• 17:33
  And I think that, you know, to me it
• 17:34
  also piques my interest, James, just
• 17:35
  like you, I don't think the answer is
• 17:37
  out there yet, but I think that, you
• 17:38
  know, the tide has shifted and it's
• 17:40
  going to be really interesting to watch
• 17:41
  how this plays out.
• 17:43
  >> Um, building on that, as someone who
• 17:46
  looks at companies from a lot of
• 17:48
  organizations from the outside, I'm
• 17:50
  often surprised by how much data gets
• 17:53
  collected but is never used. And again
• 17:56
  back I think we can talk you know about
• 17:58
  our Yahoo day since it's so long in the
• 18:00
  past we had mountains and mountains of
• 18:03
  data but only a fraction of it was used
• 18:06
  either for targeting or consumer
• 18:07
  knowledge. A lot of it was just stored.
• 18:09
  So what do you see the best companies
• 18:11
  doing when it comes to measurement
• 18:13
  experimentation or or demonstrating
• 18:15
  performance?
• 18:16
  >> Yeah I mean I'll I'll double down on
• 18:18
  what I said before. This is all about
• 18:20
  solving customer problems, real problems
• 18:23
  for them. Like the OEM example that I
• 18:25
  gave, right? We can give other, you
• 18:27
  know, retail examples all the time,
• 18:29
  which is how do I go and find my
• 18:32
  consumer or the next consumer to do
• 18:34
  certain things. Propensity models are a
• 18:37
  huge part of the future and being able
• 18:39
  to score those propensity models and be
• 18:42
  able to, you know, predict before a
• 18:44
  dollar is even spent how we think the
• 18:46
  audience is going to react to this
• 18:48
  creative or this content or these, you
• 18:51
  know, elements is is is the part of the
• 18:54
  new frontier, the AI frontier if you
• 18:56
  want if you have it. And so again,
• 19:00
  very exciting times that we're in
• 19:02
  really, you know, interested. I mean,
• 19:05
  everyone says it, but I think it's it's
• 19:06
  critical. The speed in which this is
• 19:09
  happening is astronomical.
• 19:13
  If we would have had this conversation 3
• 19:15
  months ago, I would have had half or 3/4
• 19:18
  of the knowledge that I have today. If
• 19:21
  we have this conversation in 3 months or
• 19:23
  6 months, half the answers that I gave
• 19:25
  may have been right or wrong. It's yet
• 19:28
  to be determined. That's how quickly
• 19:29
  this is going. And so if I have any
• 19:31
  advice for your audience, right, who's
• 19:32
  listening to this, it's stop standing on
• 19:35
  the sidelines fearing what you know is
• 19:39
  being said about AI and the utilization
• 19:41
  of the technologies that are out there.
• 19:43
  Start using it every day. Not just
• 19:45
  simplifying your tasks, but start
• 19:48
  applying it to the problems that you
• 19:50
  believe that you're solving for your
• 19:51
  customers today. And are you doing
• 19:53
  enough in order for it to be right the
• 19:56
  right place for your future? And that's
• 19:58
  why I'm so excited about what TransUnion
• 19:59
  Marketing Solutions is doing because
• 20:01
  we've been thinking about this. We're
• 20:03
  we're we're executing against it and our
• 20:06
  customers are are noticing and and are
• 20:07
  actually very satisfied in the way in
• 20:09
  which we're doing it.
• 20:10
  >> You you've opened a door for me that
• 20:11
  I've got to jump through. Uh I gave a
• 20:14
  presentation Austin a couple of months
• 20:15
  ago on the evolution of AI and I used a
• 20:18
  framework where some activities are just
• 20:20
  going to evolve and we know exactly what
• 20:23
  the next step's going to be. Some are
• 20:25
  completely revolutionary in that the
• 20:27
  business model breaks. We have to sort
• 20:29
  of start from scratch. Yeah. And then
• 20:31
  there was some where it was too early to
• 20:33
  tell. So sort of thinking about how
• 20:36
  quickly this model, this world of AI is
• 20:39
  evolving. What do you think now you just
• 20:42
  said in the last three months that
• 20:44
  either you've changed your mind on in
• 20:46
  the last three months or you know now
• 20:49
  that you didn't know three months ago
• 20:50
  about working with AI?
• 20:52
  >> Yeah. So I I'll open it up into the area
• 20:55
  that I'm most unsure about right now. So
• 20:58
  there's a whole notion around um agents,
• 21:02
  right, becoming like synthetic traffic
• 21:05
  around the internet and then the ability
• 21:08
  to then verify those um those agents to
• 21:14
  actually reconcile to a consumer. I
• 21:16
  think that's going to be a very
• 21:17
  interesting part of the evolution of our
• 21:19
  company. So far in this conversation,
• 21:21
  we've talked about what I call real
• 21:23
  things, right? Tangible things that
• 21:25
  we're able to control, uh that we're
• 21:27
  able to put out there and solve for. I'm
• 21:29
  not sure what's going to happen, right,
• 21:32
  with synthetic traffic and how is that,
• 21:34
  you know, going to play out over the
• 21:36
  next, you know, six months or 12 months.
• 21:38
  I'll be interested to uh come back and
• 21:40
  uh discuss that with you.
• 21:41
  >> And be clear, you're actually talking
• 21:42
  about nonhuman traffic generated by
• 21:45
  agents in this context.
• 21:46
  >> So So this is where it gets really
• 21:48
  strange, right? Because some of it is,
• 21:49
  you know, so that bots, right? So
• 21:51
  non-human traffic. Yeah. Some of it are
• 21:53
  people, you know, agents that are acting
• 21:55
  on behalf of an individual.
• 21:58
  >> Ah,
• 21:58
  >> or individuals.
• 22:00
  >> Yes.
• 22:00
  >> And so it just spreads like wildfire. So
• 22:03
  now all of a sudden the size of
• 22:05
  somebody's traffic may be quadrupled.
• 22:08
  You know, is it how does that reconcile
• 22:11
  into into into ROI in our space? How
• 22:14
  does that reconcile into solving
• 22:15
  problems or getting people to actually
• 22:16
  show up in stores or back on the uh in
• 22:19
  the car dealership still to be
• 22:21
  determined?
• 22:22
  >> I I I love it. Uh Brian, this is the
• 22:25
  Yield Doctor podcast and nobody gets
• 22:27
  away without a revenue question. What is
• 22:29
  the one thing that most publishers or ad
• 22:32
  tech companies can be doing better with
• 22:33
  their identity or data strategy to make
• 22:36
  more money?
• 22:37
  >> Yeah, I mean I said it earlier, I'll say
• 22:39
  it again. If you're not investing in
• 22:41
  identity resolution, if you're not
• 22:42
  investing in really understanding how
• 22:45
  your current consumer set or future, you
• 22:47
  know, consumer set is resolved across
• 22:49
  the the web, you're making a mistake.
• 22:52
  You need to absolutely invest in that
• 22:54
  because the return that you will get
• 22:56
  from understanding that uh is uh
• 22:58
  immeasurable today.
• 23:00
  >> Thank you, Brian. This has been a great
• 23:02
  conversation. as we wrap up, uh, what is
• 23:04
  the best way for people to learn more
• 23:06
  about you and the work that your team is
• 23:08
  doing?
• 23:09
  >> Sure. So, uh, always come into, uh,
• 23:12
  transunion.comsight.
• 23:15
  Uh, of course, you can follow us on
• 23:16
  LinkedIn and I'm always available. Uh,
• 23:19
  Brian Silver, easy to find.
• 23:22
  >> Thank you, Brian. This is the Yield
• 23:24
  Doctor podcast. I am James Deco. Thanks
• 23:26
  for watching.