Latest / Chief in Tech Deep Dive Shortcast / Chief in Tech S1E49 - Hazal Muhtar : Moving Past the Buzz – Cultivating a Truly Data-Driven Organization

Transcript

Chief in Tech S1E49 - Hazal Muhtar : Moving Past the Buzz – Cultivating a Truly Data-Driven Organization · Chief in Tech Deep Dive Shortcast

• 0:00
  Imagine handing out, you know, your premium physical
• 0:02
  product for free to thousands of your most active
• 0:06
  customers. Yeah, you would totally expect a massive
• 0:09
  spike in engagement. Exactly. You're thinking
• 0:11
  maybe a wave of positive social media buzz, but
• 0:14
  instead you get nothing. I mean, the initiative
• 0:16
  just completely bombs. Wow. So just millions
• 0:19
  of dollars in potential revenue, plus countless
• 0:22
  hours of operational effort, all completely down
• 0:25
  the drain. Yeah, all of it. And that is exactly
• 0:28
  what happened at the fintech giant, Wise. It
• 0:30
  is a staggering scenario, honestly. And what
• 0:33
  makes it so fascinating is that Wise wasn't lacking
• 0:36
  data, right? I mean, they had mountains of it.
• 0:39
  Right. But having data and actually being a genuinely
• 0:41
  data -driven organization, those are two entirely
• 0:45
  different things. They really are. So today we
• 0:47
  are welcoming you to a deep dive into what it
• 0:50
  actually takes to build a data -driven culture
• 0:53
  from the ground up. And we are going to cut straight
• 0:55
  through the corporate jargon to do it. Absolutely.
• 0:57
  No buzzwords today. We are deconstructing exactly
• 1:01
  how WISE successfully scaled their analytics
• 1:04
  function from, well, a localized team of about
• 1:06
  40 people to this massive cross -functional powerhouse
• 1:11
  of 220 members. Which is an incredible leap.
• 1:15
  And because the reality is the gap between buying
• 1:18
  expensive data software and actually changing
• 1:20
  how a business operates, I mean, that is a graveyard
• 1:23
  for a lot of companies. Oh, totally. And scaling
• 1:26
  a team by that magnitude. It really forces you
• 1:29
  to confront every single flaw in your existing
• 1:30
  system. Every single one. Yeah. And the journey
• 1:33
  at WISE emphasizes a really surprising truth.
• 1:35
  Becoming data -driven is rarely about the technology
• 1:39
  stack itself. It's almost entirely about psychological
• 1:43
  and cultural shifts. OK. Let's unpack this, because
• 1:45
  I want to get away from the cliche that data
• 1:47
  is just, you know. the new oil. Right. Yes, please.
• 1:49
  It's not. I think a much better way to look at
• 1:51
  this is to think of an organization as a giant
• 1:54
  complex nervous system. Oh, I like that. Yeah.
• 1:56
  So your data pipelines are just the sensory nerves.
• 1:59
  They can transmit signals all day long, heat,
• 2:01
  pressure, pain, whatever. But if the brain doesn't
• 2:05
  have the architecture to interpret those signals
• 2:07
  properly, or if the reflexes are wired wrong,
• 2:12
  all those sensory inputs are entirely useless.
• 2:15
  That is a brilliant way to frame it. the raw
• 2:17
  inputs are completely meaningless without the
• 2:20
  cognitive architecture to actually process them.
• 2:23
  Right. You don't need more nerves. You need a
• 2:24
  better brain. Exactly. And what the leadership
• 2:27
  at Y has realized early on is that You have to
• 2:30
  design that cognitive architecture first. You
• 2:32
  have to build the human infrastructure. Yeah.
• 2:35
  And before we even get to the human element,
• 2:36
  I mean, we have to acknowledge the physical reality
• 2:39
  of scaling to 220 analysts. Oh, absolutely. If
• 2:42
  you just drop that many people into an organization
• 2:45
  without a rock -solid data infrastructure, you
• 2:48
  aren't getting insights. You are getting absolute
• 2:50
  chaos. It is a complete nightmare scenario. Right,
• 2:52
  because if the underlying data isn't clean, accessible
• 2:56
  and, you know, reliable. What happens? You just
• 2:59
  have 220 analysts arguing over whose dashboard
• 3:02
  is mathematically correct. Exactly. They spend
• 3:04
  90 percent of their week just cleaning broken
• 3:07
  tables and arguing with data engineers. So step
• 3:11
  one of this blueprint is establishing that single
• 3:14
  source of truth. The infrastructure has to be
• 3:18
  impeccable. Precisely, because you cannot build
• 3:21
  a skyscraper on a swamp. Ensuring that your data
• 3:24
  is available and consistent is just the non -negotiable
• 3:27
  foundation. Right. But once you have that infrastructure
• 3:29
  in place, you hit the real challenge, which is
• 3:31
  how do you organize the humans? Which is the
• 3:33
  hardest part. It really is. And this is where
• 3:36
  the analytics leader at WISE had a major advantage,
• 3:39
  primarily due to their very unconventional background.
• 3:42
  Unconventional how? Well, they didn't start on
• 3:44
  a predefined data science track. They actually
• 3:46
  synthesized a background in math, economics,
• 3:49
  and psychology. OK, I can easily see how math
• 3:51
  and economics apply there. You expect the person
• 3:53
  staring at the spreadsheets to understand statistical
• 3:56
  significance and market dynamics. But psychology,
• 3:59
  how does that actually play out mechanically
• 4:01
  in a data role? Well, let's look at a concrete
• 4:03
  example. Say you have a user interface change,
• 4:06
  and suddenly there is a 5 % drop in users completing
• 4:11
  a transaction. OK, pretty common scenario. Right.
• 4:14
  Now, a purely mathematical analyst calculates
• 4:17
  the drop. verifies it statistically significant,
• 4:20
  and reports the variance. Job done. Right. But
• 4:23
  an analyst trained in psychology looks at the
• 4:26
  exact same metric and evaluates the cognitive
• 4:29
  friction. Oh! Interesting. Yeah, they look at
• 4:31
  the new UI and ask, did we just add a step that
• 4:34
  creates trust anxiety? Are we asking for sensitive
• 4:38
  information way too early in the flow? I see.
• 4:40
  So the math identifies the exact location of
• 4:43
  the wound, but the psychology diagnoses what
• 4:45
  caused the injury in the first place. Exactly.
• 4:47
  That makes a ton of sense. You weren't just looking
• 4:49
  at a churn rate. You were looking at human frustration
• 4:52
  translated into numbers. Yes, exactly. And that
• 4:55
  behavioral perspective clearly influenced how
• 4:57
  they restructured the team. Because, I mean,
• 4:59
  the initial setup at Wise was pretty broken.
• 5:01
  Oh, very. They had 40 analysts, but those analysts
• 5:04
  were reporting directly to individual product
• 5:06
  managers. Which is incredibly common in early
• 5:09
  -stage tech companies. It is. A product manager
• 5:12
  wants to move fast, so they hire an analyst to
• 5:15
  sit next to them and just run the numbers for
• 5:18
  their specific feature. But the problem there
• 5:21
  is obvious. I mean, if you report directly to
• 5:23
  the person building the feature, you aren't an
• 5:26
  independent analyst. You're just a query machine.
• 5:29
  You exist just to answer whatever narrow, isolated
• 5:32
  question the PM has that morning. You have zero
• 5:35
  broader vision of how that feature impacts the
• 5:38
  rest of the ecosystem. Exactly. So they made
• 5:41
  a massive structural pivot. They defined a clear
• 5:44
  vision and mission for the analytics team as
• 5:47
  a standalone entity. OK. And they moved to what
• 5:50
  they called an embedded model. And this is a
• 5:53
  critical structural decision. They had to choose.
• 5:56
  Right. Do you build a centralized model where
• 5:58
  all the analysts sit together in a separate department
• 6:00
  and just take support tickets from the rest of
• 6:02
  the company? Right. Or do you embed them? Right.
• 6:05
  Let's define the mechanics of that embedded model
• 6:07
  because it can mean a few different things depending
• 6:08
  on the company. At WISE, this meant the analysts
• 6:12
  formally reported up through the data leadership
• 6:14
  structure. So their performance reviews, their
• 6:17
  standards, their career paths were all managed
• 6:19
  by data experts. Yes. But their physical day
• 6:22
  -to -day work was embedded within specific product
• 6:25
  teams. They sat with the product managers. They
• 6:27
  went to their daily standups. They focused entirely
• 6:30
  on that specific domain. And they chose this
• 6:33
  specifically to enhance contextual work. By being
• 6:37
  embedded, the analyst deeply understands the
• 6:40
  nuance of the product they're analyzing. Right,
• 6:42
  they aren't just looking at cold numbers from
• 6:44
  afar. Exactly, they understand the strategy behind
• 6:46
  the numbers. I'm tracking with you, but I have
• 6:48
  to play devil's advocate here. Okay, go for it.
• 6:50
  Human nature is to fit in with your immediate
• 6:52
  peers, right? So if I'm an analyst and I sit
• 6:55
  next to the lead product engineer every single
• 6:57
  day, we get coffee, we are friends, don't I lose
• 7:01
  my objectivity? It's a valid concern. Right.
• 7:04
  If she spends three months building a new feature,
• 7:06
  I don't want to be the guy who tells her it's
• 7:08
  a total failure. Don't embedded analysts just
• 7:11
  become an echo chamber for the product team,
• 7:14
  maybe unconsciously hunting for data that proves
• 7:16
  their friends are succeeding. What's fascinating
• 7:19
  here is how the dual reporting structure perfectly
• 7:22
  neutralizes that exact risk. Oh, really? Yeah.
• 7:25
  Yes. If the analyst reported to the product manager,
• 7:29
  the pressure to sugarcoat the data would be immense.
• 7:32
  But because their formal reporting line, you
• 7:34
  know, their boss, their promotion trajectory
• 7:36
  is tied strictly to the central data organization,
• 7:40
  they have the psychological safety, to be brutally
• 7:43
  honest. Ah, I see the mechanism now. The central
• 7:46
  reporting line protects their objectivity, maintaining
• 7:49
  the high standards of the data organization.
• 7:51
  But the physical embedding gives them the context
• 7:54
  they need to know what the data actually means.
• 7:56
  Right, because a centralized, isolated data team
• 7:59
  might be mathematically objective, but they are
• 8:01
  often completely detached from reality. Exactly.
• 8:04
  They might see a dip in engagement and spend
• 8:06
  three days running complex regression models,
• 8:10
  completely unaware that the product team simply
• 8:12
  moved a button on the homepage yesterday. Yeah,
• 8:15
  that makes sense. The embedded model bridges
• 8:17
  that gap. It's basically structural engineering
• 8:20
  for human behavior. That's a great way to put
• 8:22
  it. You are designing the org chart to force
• 8:24
  both honesty and context, but an org chart is
• 8:27
  just empty chairs. If you drop traditional isolated
• 8:31
  number crunchers into those embedded seats, the
• 8:33
  system still fails. Oh, completely? You need
• 8:36
  a completely different psychological profile
• 8:38
  to make that structure work. Which brings us
• 8:41
  to a massive hurdle, hiring. Scaling from 40
• 8:45
  to 220 people means you are constantly recruiting.
• 8:48
  Yeah, and the natural instinct in tech is to
• 8:51
  just filter for the smartest coders, the people
• 8:54
  who can write the most elegant Python scripts.
• 8:56
  But Wwise flipped that on its head. They implemented
• 8:58
  a very specific values assessment during the
• 9:01
  interview process. Technical skills were just
• 9:03
  the baseline. What they really prioritized was
• 9:06
  finding individuals motivated by solving customer
• 9:09
  challenges. They were hunting for an innate,
• 9:12
  restless curiosity. Exactly. And to understand
• 9:15
  how that values assessment worked mechanically,
• 9:18
  you have to look at how they evaluated problem
• 9:20
  solving. They wouldn't just hand a candidate
• 9:23
  a clean data set and ask for a specific metric.
• 9:26
  They would present a messy, ambiguous business
• 9:29
  scenario. I love that approach. Yeah. And the
• 9:32
  candidate who immediately starts writing code
• 9:34
  on the whiteboard. They fail. Right. But the
• 9:37
  candidate who pauses and asks a dozen clarifying
• 9:40
  questions about the user's intent, the market
• 9:42
  conditions, and the business goals, that is the
• 9:45
  curious mindset they were looking for. Because
• 9:47
  curiosity is the actual engine of insight. Yes.
• 9:50
  Think about it. An uncurious analyst runs a report,
• 9:54
  sees that European transfers are down 10%, hands
• 9:57
  the dashboard to the manager, and clocks out.
• 10:00
  But a curious analyst sees that same 10 % drop.
• 10:04
  starts slicing the data by device type, realizes
• 10:07
  it's only happening on the Android app, downloads
• 10:09
  the app themselves, tries to make a transfer,
• 10:12
  and discovers the biometric login is broken.
• 10:14
  I love that example. The uncurious analyst reports
• 10:17
  the weather, but the curious analyst fixes the
• 10:19
  roof. That is the exact difference. But what
• 10:22
  is truly impressive about Y's approach is that
• 10:25
  they didn't isolate this curiosity to just the
• 10:28
  analytics team. Right. They didn't just hire
• 10:30
  220 curious data wizards and say, all right,
• 10:33
  our job is done. You guys handle all the thinking.
• 10:36
  Yeah, that would defeat the purpose. Exactly.
• 10:38
  They initiated a massive push for data literacy
• 10:41
  across the entire organization. Data literacy,
• 10:44
  meaning they wanted marketing, operations, engineering,
• 10:47
  really everyone to have the tools and the baseline
• 10:50
  knowledge to understand and utilize data themselves.
• 10:53
  Yes. It's essentially teaching everyone in the
• 10:56
  company to speak a shared language so that data
• 10:58
  isn't some mystical artifact controlled by a
• 11:00
  few gatekeepers. It reminds me of how a truly
• 11:03
  high end Michelin star restaurant operates. Oh,
• 11:06
  how so? Well, it's like He's teaching the entire
• 11:08
  waitstaff how to taste the food, understand the
• 11:11
  obscure ingredients, and describe the complex
• 11:13
  flavor profiles to the guests. Oh, I see. In
• 11:16
  a mediocre restaurant, all the palate work is
• 11:18
  left strictly to the head chef in the back. If
• 11:21
  only the chef knows how the food is supposed
• 11:23
  to taste, the customer experience suffers. Right,
• 11:26
  because the servers can't answer any questions.
• 11:28
  Exactly. But if the whole waitstaff is flavor
• 11:31
  -literate, they can guide the customer, make
• 11:33
  real -time recommendations, and catch a flawed
• 11:35
  dish before it ever leaves the kitchen. If we
• 11:38
  connect this to the bigger picture, spreading
• 11:41
  that analytical thinking. throughout all levels
• 11:43
  of a company fundamentally democratizes decision
• 11:47
  -making. Right. If only the data team can pull
• 11:49
  and interpret numbers, they become a massive
• 11:52
  agonizing bottleneck. Oh yeah. Every minor operational
• 11:56
  decision has to wait in a two -week queue for
• 12:00
  an analyst to bless it. Which means the company
• 12:02
  moves at a snail's pace. Exactly. By fostering
• 12:06
  widespread data literacy, you empower people
• 12:08
  on the front lines to make informed choices quickly.
• 12:11
  Makes total sense. When the marketing team can
• 12:13
  run their own basic cohort analysis, they can
• 12:16
  iterate on a campaign in real time. Without this
• 12:20
  widespread literacy, companies operate in the
• 12:23
  dark and they make incredibly costly reactive
• 12:27
  mistakes. Because they react to surface -level
• 12:29
  symptoms. Right, because they lack the tools
• 12:31
  to understand the underlying causes. And that
• 12:34
  perfectly sets up the case study we mentioned
• 12:36
  at the very beginning of this deep dive. Let's
• 12:38
  look at what happens when an organization jumps
• 12:41
  to conclusions without that deep contextual analysis.
• 12:45
  Yes, let's get into it. Let's break down the
• 12:47
  free card fiasco. It is one of the most illustrative
• 12:49
  examples of treating a symptom instead of a disease.
• 12:53
  So Wyze had a specific business objective. They
• 12:56
  wanted to boost the adoption of their physical
• 12:58
  debit cards. People were signing up for the core
• 13:00
  service, moving money around, but a significant
• 13:03
  portion of them were not ordering the physical
• 13:05
  card. And the organization, likely feeling the
• 13:08
  heat to get those adoption metrics up, decided
• 13:11
  on what seemed like a frictionless solution.
• 13:15
  They said, Let's just manufacture a batch of
• 13:17
  cards and mail them out for free to our active
• 13:20
  users. It sounds perfectly logical on a whiteboard.
• 13:23
  It really does. The friction point appears to
• 13:25
  be the ordering process or the fee. So if you
• 13:28
  want people to have a card, just give it to them.
• 13:30
  Problem solved. Except it wasn't. It failed completely.
• 13:34
  The cards were mailed out, the expense was incurred,
• 13:37
  and engagement barely twitched. Most of those
• 13:40
  cards probably ended up straight in the trash.
• 13:42
  And the reason it failed is because mailing the
• 13:44
  cards only addressed the symptom, which was low
• 13:47
  adoption numbers. They skipped the crucial step
• 13:49
  of diagnosing the root cause. And here is where
• 13:52
  that embedded contextual curiosity becomes totally
• 13:56
  vital. Here's where it gets really interesting.
• 13:58
  Why did they fail? What did the embedded analysts
• 14:01
  actually discover when they finally dug into
• 14:03
  the why behind those low numbers? Well, what
• 14:06
  the data ultimately revealed was a complete mismatch
• 14:09
  in the use case. Oh, really? Yeah. When analysts
• 14:12
  actually segmented the user behavior, they found
• 14:14
  that a massive chunk of these users were utilizing
• 14:17
  the platform almost exclusively for large digital
• 14:22
  cross -border remittances. Right. So they didn't
• 14:24
  even need a card. Exactly. They were expats sending
• 14:28
  money home. or parents paying international tuition.
• 14:31
  Their entire financial flow was digital. A physical
• 14:34
  piece of plastic was completely irrelevant to
• 14:37
  their needs. Wow. So it wasn't that the card
• 14:40
  was too expensive or that the ordering process
• 14:43
  was too hard It was that the product itself offered
• 14:45
  zero value for that specific customer journey,
• 14:48
  right? Mailing them a physical card didn't magically
• 14:51
  make them want to go buy a coffee with it It
• 14:53
  just confused them exactly the initial initiative
• 14:56
  looked at the metric instead of the human being
• 14:58
  behind the metric Yeah, and this highlights a
• 15:01
  huge vulnerability in smart organizations The
• 15:04
  corporate world heavily rewards velocity. You
• 15:06
  know, don't just sit there analyzing ship something.
• 15:08
  Right, move fast and break things. But rushing
• 15:11
  to a solution without diagnosing the actual user
• 15:13
  context is a massive waste of capital. It's like
• 15:16
  a doctor prescribing heavy painkillers for a
• 15:18
  broken leg without ever actually setting the
• 15:20
  bone. You feel like you're doing something, but
• 15:23
  the core issue is untouched. That's a perfect
• 15:26
  analogy. Yeah. And this is why Wise realized
• 15:29
  they had to champion iterative learning. A truly
• 15:32
  data -driven organization doesn't just use data
• 15:35
  to justify launching a product. They use data
• 15:39
  to brutally evaluate their own failures. You
• 15:42
  have to create a psychological environment where
• 15:45
  a team can stand up and say, mailing the free
• 15:47
  cards was a total failure, and here's exactly
• 15:50
  why it failed based on the behavioral data. Yeah,
• 15:52
  that makes sense. That postmortem is often more
• 15:54
  valuable than a successful launch. But that realization
• 15:58
  completely shifts the timeline of when an analyst
• 16:00
  is supposed to do their job, right? If you want
• 16:03
  to avoid another free card fiasco, the analyst
• 16:06
  cannot just be the person you call after a launch
• 16:08
  to read the scoreboard. Absolutely. If you only
• 16:10
  bring the analyst in at the end to see if the
• 16:12
  metric went up or down, you've missed the entire
• 16:14
  game. The decisions that dictate success or failure
• 16:17
  were made months ago. So the operational solution
• 16:21
  they implemented was full lifecycle integration.
• 16:24
  Yes. They required analysts to be present throughout
• 16:27
  the entire product lifecycle. From day one, you
• 16:31
  know, the initial problem discovery phase where
• 16:33
  the team is just spitballing what the customer
• 16:35
  actually wants. Right. all the way through the
• 16:38
  technical development, the launch, and the post
• 16:40
  -launch iteration, the analyst is in the room
• 16:43
  shaping the strategy the whole time. And this
• 16:46
  full integration is what forces the evolution
• 16:48
  of the generalist profile. Okay, what do you
• 16:50
  mean by that? Well, when an analyst is involved
• 16:54
  from the very first brainstorming session, they
• 16:56
  can't just hide behind a SQL query. They have
• 16:59
  to actively merge their deep technical skills
• 17:02
  with highly nuanced soft skills I want to focus
• 17:05
  on that friction for a second because that is
• 17:07
  not an easy transition No, it's really hot. You
• 17:09
  are asking someone whose primary skill set is
• 17:12
  Mathematics and coding to suddenly become a master
• 17:15
  communicator They have to know how to gently
• 17:17
  tell a product manager that their thesis is flawed
• 17:21
  They have to know how to present complex data
• 17:25
  architectures to non -technical stakeholders
• 17:28
  without putting them to sleep. Right. They are
• 17:30
  basically acting as translators between the raw
• 17:33
  database and the human business strategy. This
• 17:37
  raises an important question about how we define
• 17:39
  modern data work in the first place. How so?
• 17:41
  For the last 20 years, we have treated data almost
• 17:44
  exclusively as an IT function. It was a back
• 17:47
  office, purely technical pursuit. Sure, the people
• 17:50
  in the basement running numbers. Exactly. But
• 17:53
  the leaders who are actually winning in the market
• 17:55
  today, like the architecture we're studying at
• 17:57
  Wise, they recognize that data is a behavioral
• 18:00
  discipline. A behavioral discipline. Yes. They
• 18:03
  empower their teams to act as behavioral translators.
• 18:06
  They aren't just reporting on what happened yesterday.
• 18:09
  They're predicting what will happen tomorrow.
• 18:11
  based on deep cross -functional insights. It
• 18:13
  fundamentally changes the identity of the role.
• 18:16
  You are no longer just a support ticket resolution
• 18:19
  machine. You are a co -creator of the product.
• 18:22
  And when you have 220 people operating with that
• 18:25
  level of strategic context, the compounding effect
• 18:28
  on a company's velocity and accuracy must be
• 18:31
  incredible. It creates a massive competitive
• 18:33
  edge. Companies that can quickly iterate and
• 18:36
  refine their strategies based on accurate behavioral
• 18:39
  diagnosis. They don't just survive market shifts,
• 18:42
  they dictate them. All right, we've covered a
• 18:44
  tremendous amount of ground today, looking at
• 18:46
  the mechanical and psychological realities of
• 18:48
  WISE's analytics transformation. It's a lot to
• 18:51
  take in. Let's synthesize the core blueprint
• 18:53
  here so we don't lose the forest for the trees.
• 18:55
  It boils down to a distinct sequence. OK. First,
• 18:59
  you have to establish the physical foundation.
• 19:01
  Build a rock -solid data infrastructure so your
• 19:04
  teams aren't drowning in broken pipelines. Crucial
• 19:06
  step one. Second, you structure for context,
• 19:09
  utilizing that embedded model with dual reporting
• 19:11
  lines to... balance objectivity with deep product
• 19:14
  knowledge. And then you activate that structure
• 19:16
  with the right human elements. You hire relentlessly
• 19:20
  for curiosity, testing for how people approach
• 19:23
  messy problems rather than just how fast they
• 19:26
  code. And finally, you democratize that knowledge.
• 19:29
  You push for widespread data literacy across
• 19:32
  every department, ensuring the entire restaurant
• 19:34
  knows how to taste the food as you put it. So
• 19:37
  what does this all mean for you listening right
• 19:40
  now? Yeah. The takeaway here isn't confined to
• 19:42
  data scientists or tech executives. This entire
• 19:45
  breakdown is a framework for anyone trying to
• 19:47
  make smarter, more impactful decisions in their
• 19:50
  own career. Absolutely. Before you launch your
• 19:52
  next project, before you suggest a massive new
• 19:55
  initiative to your boss, ask yourself, are you
• 19:58
  just treating a surface level symptom or have
• 20:01
  you done the tedious contextual work to diagnose
• 20:04
  the root problem? Right. Are you just looking
• 20:06
  at the numbers or are you aggressively seeking
• 20:09
  out the human behavior behind those numbers?
• 20:11
  It's about approaching every professional challenge
• 20:13
  with that same innate restless curiosity that
• 20:16
  drives the most successful analysts. Don't settle
• 20:19
  for the first answer. Dig into the friction.
• 20:22
  Which leaves us with a final slightly provocative
• 20:24
  thought to chew on. We live in an era currently
• 20:27
  obsessed with automation. The constant drumbeat
• 20:31
  in the business world right now is the desire
• 20:34
  to use AI to fully automate analytics, instantly
• 20:40
  generate the dashboards and spit out the strategy
• 20:42
  without any human intervention. Oh yeah, it's
• 20:44
  everywhere. But everything we just explored shows
• 20:47
  that building a genuinely data -driven culture
• 20:50
  relies heavily on uniquely human traits. It really
• 20:54
  does. It relies on psychological intuition, on
• 20:57
  cross -functional persuasion, and on the nuanced
• 20:59
  ability to realize that a customer doesn't want
• 21:02
  a free debit card because they are only sending
• 21:04
  digital remittances. It requires an understanding
• 21:07
  of human irrationality. Exactly. So if the current
• 21:10
  rush is to fully automate the analytics function,
• 21:13
  to strip away the human element entirely in the
• 21:15
  name of efficiency, will that actually make organizations
• 21:18
  less data -driven? That's a great question. By
• 21:21
  removing the behavioral translators, are we so
• 21:23
  eager to automate the sensory nerves that we
• 21:26
  are effectively lobotomizing the brain? It is
• 21:28
  something you definitely want to think about
• 21:30
  as you navigate the next few years of your career.
• 21:33
  Keep questioning the consensus, and until next
• 21:35
  time, keep digging deeper.