Latest / Shiny New Object - a Marketing Podcast / Making AI change management work with Reckitt's Ishaan Kaul
Transcript
- 0:00Tom Ollerton: Hello and welcome to the Shiny New Object Podcast.
- 0:03This is a weekly or so podcast where I interview some of the
- 0:07industry's leaders about the future of data-driven marketing
- 0:11and this week is no different. Before we meet our guest, I
- 0:14should introduce myself. I'm Tom. I'm the founder of
- 0:17Automated Creative. We are the fastest way to create scale and
- 0:21test performance ads and one of our clients actually is Reckitt
- 0:25and I'm interviewing someone from Reckitt today and this is
- 0:28Ishaan Kaul, who is... Cool job title. Check this out. Director
- 0:32of A.I. Adoption and Creativity at Reckitt. So Ishaan, for
- 0:36anyone who doesn't know who you are and what you do, could you
- 0:38give us a little bit of background?
- 0:40Ishaan Kaul: Definitely. Thank you for having me. Really
- 0:42excited to be here.
- 0:43Tom Ollerton: Pleasure.
- 0:43Ishaan Kaul: So the title actually has three mandates, and
- 0:46it'll be easier possibly if I can explain each of them. So the
- 0:49first of it is around A.I. adoption. So I work with our
- 0:53brilliant product teams and ship A.I. products and features to
- 0:57our marketing function. So that's the part of deployment.
- 1:00Second, I also ensure with my team that we work with the
- 1:04marketers, almost 20 plus markets, 1,500 plus marketers
- 1:08that they can adopt it and, which essentially is a lot less
- 1:12tech, a lot more people and a lot more data. Analysing use
- 1:15cases, features and making sure people can find value out of
- 1:19what we are delivering to them. And the third and more recent,
- 1:21as you said, is creativity. This is the interesting and sort of
- 1:25exciting part for me at the minute. So I work on figuring
- 1:28out how do we leverage A.I. to deliver content and creativity
- 1:32at scale in a large enterprise. You know, as you can understand,
- 1:36CPG companies deliver and develop a lot of content at
- 1:40scale across many markets and functions. So that's in a
- 1:42nutshell what I do currently.
- 1:44Tom Ollerton: Right. That is a big job. Right, brilliant. Okay.
- 1:47Well, we'll get into that a bit later. So you've been at Reckitt
- 1:50for 13 years, intern, grad scheme. You've gone all the way
- 1:53up to this cool job. But in that time, if you made any mistakes,
- 1:59was there like one moment where you were like, really red-faced
- 2:02and made an idiot of yourself that you're glad happened.
- 2:05Ishaan Kaul: Well, many on the way, I think those sort of shape
- 2:08up your career, your learning. And I think one of this comes to
- 2:12my mind specifically when I was a... A bit of my journey in
- 2:15Reckitt has been across sales where I did account management
- 2:19and I moved to handling a few brands and innovation for a
- 2:23particular category within the Southeast Asia market. And
- 2:26within this, while I was handling the India innovation
- 2:30for our floor cleaning business, there was a big opportunity that
- 2:34I realised that was there for the taking. India is a very
- 2:37peculiar market as you would have known, and we are a floor
- 2:41cleaning brand and we realised that 40% of Indian households
- 2:45didn't even have either a tile, marble, or wood flooring. And
- 2:49for a business to understand that insight, the consumer base
- 2:53and the scale of it was quite exciting. And those people had
- 2:56cement floors and hence the whole business case and the
- 3:00whole insight was across, "Can we make something more specific,
- 3:04more customised to a consumer base that large?" There was no
- 3:08one that's in the market satisfying that need and I went
- 3:11ahead in my early days in marketing trying to prove a lot
- 3:15of my understanding of the, of the consumer and developed a
- 3:18really fantastic product with our R&D teams. And we put a lot
- 3:22of investment and, you know, money behind advertising and
- 3:26creativity. What I did realise and in the first year we did
- 3:30quite well, in fact, the product got Product of the Year. We did
- 3:34a few million households, but my entire business case was based
- 3:38on trial and trials ended after a point in time because we
- 3:41didn't get repeats. And while realising that as I, as I went
- 3:45ahead to developing that business case, I did realise
- 3:48that the big mistake that I did was I didn't go deep enough into
- 3:52understanding how people would start and then stop using
- 3:56products. So repeatability was not a big part of my business
- 4:00case and that's where we lost, not lost a lot of business, but
- 4:04I think we didn't get the business we deserved. We had to
- 4:07make sure that we did a lot of changes in the formulation over
- 4:11time. But I think that's one piece that helped me understand
- 4:15that I was looking at only data that was only possibly going to
- 4:19grow and not looking at measures for myself that possibly gave me
- 4:23the overall loop of the story. And I think that's one big area
- 4:27that I understood from my early part of my marketing career when
- 4:31I was more sales oriented that I need to look at longer term
- 4:35pictures and repeats and try to understand where I need to
- 4:38create loops of marketing and repeatability.
- 4:41Tom Ollerton: Yeah, it reminds me of a story I heard when we...
- 4:43Automated Creative wrote a book last year called Using
- 4:46Creativity and Data in Marketing, and I was
- 4:49interviewing a chap and he told me this really interesting story
- 4:53about, I think it might have even been in Southeast Asia. And
- 4:56then he was doing like a house visit. And they done a... I
- 4:59can't remember what it was, it was like shampoo and it had like
- 5:03little sachets to try out but it had Cristiano Ronaldo on, or
- 5:07some... I can't remember who it was. Unilever, I think. It's
- 5:10definitely Unilever and they, you know, went into the house
- 5:13and said, "You know, have you ever tried like..." whatever
- 5:17this hair product was, just a shampoo, and this kid came
- 5:20running out with these two packets of Ronaldo shampoo, and
- 5:23that was his birthday present. Do you know what I mean? And
- 5:27this guy couldn't believe it. He was just like, "Wow! That is
- 5:30amaz-" Like he had, similarly to you, oh, if we send to trial,
- 5:34they'll use it, they'll love it, they'll go to the shop and buy
- 5:37it. But actually, no, this was a birthday present. So his
- 5:41understanding was so different. So that's a, that's a great
- 5:44story. So, so moving now to like much more practical advice. So
- 5:48do you have a single silver bullet bit of advice for a
- 5:51data-driven marketer to help them become better at their jobs?
- 5:54Ishaan Kaul: I think what I've learned. I'll try to distil it
- 6:20Ishaan Kaul: But yeah, so the law is then you start making a
- 6:23measure in incentive, it stops being a good measure because
- 6:27people, our human tendency to sort of game incentives or
- 6:31measures which are related to incentives. The learning rather
- 6:35is data tells you only a few parts of the story for you to
- 6:39connect the story. And when you look at it in different types or
- 6:43the business areas, for example, when you look at e-commerce, you
- 6:43to a couple of advices if possible. I think what I've
- 6:47learned, of course, from from some of these early mistakes is
- 6:48look at customer... A lifetime value. How many people come,
- 6:52buy, how much they buy and then how much they repeat? Try
- 6:55building a story of where the consumer starts and ends for
- 6:59your business, your category or your relevant function, and then
- 7:03there is a lot of data that you can process, and you know
- 7:04try to put data in which can sort of give you the right
- 7:06there's this this good Goodhart's law that when you
- 7:07measure or possibly even the right signals because just data
- 7:08make it when you make...
- 7:09Tom Ollerton: It's in the book?
- 7:11Ishaan Kaul: Yes and data, a measure stops being a...
- 7:11or just measures does not help.
- 7:12Tom Ollerton: Yeah. Explain Goodhart's law.
- 7:13Tom Ollerton: So that's nice. So you build a story from the
- 7:16consumer journey start and end, and then you have like points,
- 7:19data along that to support or verify that story. That's
- 7:23really- That's nice, actually, because you know, one of the
- 7:26things that we discovered in the book, actually... Who was
- 7:28telling me this story? (incomprehensible name) and
- 7:30Rosie were telling me the story about in India when the, when
- 7:35the British were in India, there was too many cobras. So, you
- 7:40know this story...
- 7:40Ishaan Kaul: Yeah, that's a very-
- 7:41Tom Ollerton: Where they said they gave a bounty for every
- 7:44cobra skin, and so every time someone brought a dead cobra,
- 7:48they got money. But then people created cobra farms and then the
- 7:53British realised this wasn't working and then stopped the
- 7:56incentive and then they let all the cobras go. So the, the
- 7:58volume of cobras went up in the area and that is, it's
- 8:02optimising to Goodhart's law. Brilliant, love, love that
- 8:05story. It's so funny. Right, okay, so it's shiny new object
- 8:09time now. So your shiny new object is Reckitt System for
- 8:14Creativity and A.I. which basically sounds like your
- 8:18entire job.
- 8:19Ishaan Kaul: Well, yes.
- 8:20Tom Ollerton: We've got like 15 minutes. Why is that your shiny
- 8:22new object? So I think I know the answer to this one, but why
- 8:25have you chosen to talk about that? What is it? How does it
- 8:27work?
- 8:28Ishaan Kaul: I think why is because in any business, and
- 8:31especially in CPG business, what we realise is we create a lot of
- 8:35content. There's a lot of creativity goes in different
- 8:38parts of the business and processes, the attempt is to
- 8:42help and most businesses try to do it as a pipeline. We start
- 8:45from upstream, we do midstream and we do downstream. So the
- 8:49challenge to solve right now for me and for, for many businesses
- 8:53is to not just do content at scale but to create or rather
- 8:57change this pipeline into a loop. And that is how you also
- 9:00leverage not just A.I. but learning. So how are we trying
- 9:04to solve it is we divide the whole journey into three parts.
- 9:08There's upstream, which is all of your insights, briefing, the
- 9:12idea of what we want to do. Here, there is less of
- 9:15technology at the minute or solutions at the minute that can
- 9:19help. But there still are a few then there's midstream, which is
- 9:23all of your production, versioning, creation, and then
- 9:26there's downstream, which is of course implementation and then
- 9:30measurement. What we realise is the mid part has a lot of
- 9:33solutions and tools, especially A.I. led that can, you know,
- 9:37help you. Creation of content at this point in time is not a
- 9:41difficult job. But the system of understanding what, which is
- 9:45slightly less tech oriented but needs creativity and brains and
- 9:49the human in the loop, midstream to connect with it, to start
- 9:53creating this idea into, you know, different versions of that
- 9:56digital, print. And then connecting it to platforms where
- 10:00you want to show this content and then learning from it as a
- 10:04loop. So that system currently doesn't exist and that's what
- 10:08we're trying to create. And that's the shiny new podcast.
- 10:11The thing we are trying to incorporate is where do you use
- 10:15A.I. and where you don't. And I think that's, that's the, that's
- 10:19the opportunity and the challenge, where we understand
- 10:22that midstream there is enough and more. You can see most of
- 10:26the frontier model having a great, a great attempt now or a
- 10:30great improvement of the last one year on creation of content
- 10:33be that static or video or dynamic. Any types. But in the
- 10:37upstream as well as the downstream there is still a gap.
- 10:40Tom Ollerton: What's the gap? So in midstream, yeah, there's,
- 10:43there's a ton of things that can just produce stuff. How great it
- 10:47is? That's a different question. We'll get into that later but
- 10:51what, where's the gap with the insight and the implementation
- 10:55and the measurements? Is that like, you know, there's things
- 10:58like synthetic audiences, is that not working or is it too
- 11:02risky to aggregate and approximate that? What's the gap?
- 11:06Ishaan Kaul: I think the gap is twofold. One is efficiency
- 11:09because midstream can only produce and you don't need
- 11:12essentially a photocopier because you need an
- 11:14understanding of what to produce first and the what needs to be
- 11:18connected to the down because what we sometimes do is create
- 11:23content basis, the understanding of what the consumer, the
- 11:27insight, the problem is. And to connect it to what it's working,
- 11:31it's very similar to what I was trying to explain in the
- 11:34previous, a learning, right? We understand they don't have
- 11:37cement floors, but does it really work for them? What areas
- 11:41do work? And then when you test it out and the big learning is
- 11:44as you keep small tests and learn, you will realise a lot
- 11:48more of your assumptions either were wrong or possibly weren't
- 11:51completely right and to connect this loop back to an insight
- 11:55because this whole stream has a lot of stakeholders as well. And
- 12:00the loss of possible translation is something that we also want
- 12:05to make sure that is avoided and you know, it can flow through as
- 12:09much as objectively from a data lens as possible. You're rightly
- 12:13mentioning the fact that there are some opportunities, for
- 12:15example, you can do synthetic testing. You can also understand
- 12:18at the downstream now, a video layering, which is how does each
- 12:23and every frame perform and how does the frame then give you a
- 12:29first 3 minutes hold the next three minute, sorry, 3 seconds,
- 12:333 minutes is too long now. A 20-second video, what does it
- 12:36contain layer by layer? And then you can aggregate it because
- 12:40data layering at the video level is going to be exciting and I
- 12:44think that's possibly the future of measurement as well.
- 12:47Tom Ollerton: So I used to do an innovation role when I was in an
- 12:50agency years ago. And I always think that as an innovation
- 12:54person you have to live in the future, you have to sell the
- 12:56future to your organisation or your clients, whatever it is,
- 13:00and then in your head, you're like I know we're going to get
- 13:04to this thing. This is going to be big. This is going to be
- 13:06useful. It's gonna make us money or save us money, right? But
- 13:10that's not enough because you bring the truth to the people
- 13:13and they're like, "I'm too busy." "I've got to do this."
- 13:16"No, no, that's not what we do." So usually I found that the
- 13:19issue was never the technology. It was the people. So how are
- 13:23you dealing with that? Because your second point, I think is
- 13:26working with 18,000, 1,800 how many marketers it was, and all
- 13:30of these different markets. And these people are time poor, you
- 13:34know, they're launching the new Dettol for concrete floors or
- 13:37whatever it is, and they're working super hard just to get
- 13:40one ad out. Right?
- 13:41Ishaan Kaul: Yeah.
- 13:41Tom Ollerton: and you're talking about what we're going to do is
- 13:43we're going to check every second. We're going to optimise
- 13:45all the different elements of it. Yeah, that's a huge shift.
- 13:48Ishaan Kaul: It is.
- 13:48Tom Ollerton: How are you approaching that?
- 13:49Ishaan Kaul: A big part of it is community and centre of
- 13:53excellence where we work with the few champions, we call them
- 13:56super users and try to create a community of people who can test
- 13:59and learn with us because what you can do at a centre level is
- 14:04strategy, technology, making the right workflows. But when it
- 14:07comes to actually launching it with market or region, the
- 14:11brand, you need to also understand the practical reality
- 14:14of how a brand manager, senior brand manager, an agency person
- 14:18would work. And we realise that there are certain internal
- 14:24centre of excellences and certain communities of people
- 14:28who want to test, who want to learn more. For example, the
- 14:32Southeast Asia and the Asian markets are really high on life
- 14:34commerce and social commerce and hence that is a regional centre
- 14:38of excellence. And hence if you want to test something which is
- 14:41there which is your videos, which is 20-second on either
- 14:44TikTok and Instagram, that's the place. If you want to do
- 14:47something more end to end, which is I want to understand from
- 14:52brief, the loop for one product that possibly somewhere in the
- 14:56EU, so we try to figure out, what areas can we test and it's,
- 15:01it's, it's always small test and learns with different regions
- 15:05and brand. So I think that's, that's the approach we take and
- 15:08I think that's also the right one because a big learning is as
- 15:12you also mentioned for me was you can have the best technology
- 15:16available to the people. But on a regular Tuesday morning when
- 15:21they're trying to figure out analysis, trying to ensure what
- 15:25do I need to present or what do I need to change? It needs to
- 15:28work at that moment. So a lot of the forward deployment work that
- 15:32we do, the Palantir nomenclature is to work with those people on
- 15:38those days and figure out where did you find the problem and
- 15:41it's not just about measurement because that's easy. You can see
- 15:45how many log ins and how many people downloaded X, Y, Z. But
- 15:48the real work is to find out, what exactly did you feel?
- 15:52What's the barrier? When you did this on a regular day?
- 15:56Tom Ollerton: Yeah. And it's an adoption- One really interesting
- 15:59challenge is that we're seeing is that we get approached a lot
- 16:02by brands who, like, they have a Center of Excellence and what
- 16:05they want to do is reduce time to market, cost of producing,
- 16:10probably the middle stuff, actually. But then the designers
- 16:12who work at that organisation, they don't like doing all of the
- 16:15resizing and all that stuff, but they get paid for it. So what
- 16:21we're kind of finding is you have this, this appetite from
- 16:23marketing efficiency people, operations people, but then
- 16:27you've got real humans who are like, well if you can just press
- 16:30a button and automate all this, what do I get paid for? The
- 16:33obvious answer is creativity. You know, go and do the thing.
- 16:36But there's- That's a real tension. What- Are you seeing
- 16:39anything like that? Are you, are you seeing people think, "Well,
- 16:41if I can just automate writing a strategy or what am I going to
- 16:45do?"
- 16:45Ishaan Kaul: Well, I think that's where we have indicated
- 16:48ourselves and we try to ensure that we have an expectation
- 16:51mapping with the people that we work with, because enterprise
- 16:55A.I. is very different from consumer A.I.. Let's face it,
- 16:59Claude, ChatGPT new versions essentially I use it for, I have
- 17:03any question in the world. I can put it in a ChatGPT or Claude.
- 17:07But at an enterprise level, you need to understand that it works
- 17:11slightly differently, syndicated data and sources to navigate
- 17:15that. It's challenge to ensure where it fits. We call it
- 17:19embedment. Where does it fit in the process that you use is
- 17:22always going to be more successful than trying to change
- 17:26the process that you use and the first line of adoption for us is
- 17:30if you use seven steps, we will try to figure out which steps
- 17:34does A.I. come in rather than changing seven steps. And that's
- 17:38where it is human change management rather than
- 17:41technology related management. But as I said, there's the two
- 17:45piece to it, there's change in management, which is how do you
- 17:49ensure- So we build a very robust feedback loop from users
- 17:53to at any point in time. We know across different markets and
- 17:56users, what barriers, what areas do they need more than. You
- 18:00know, I need a feature, I need more data. Sometimes I need- I
- 18:04need to understand how this works. So can I create like a
- 18:08library of, you know, classes for you to solve X-Y-Z problems,
- 18:12use cases. And that's the piece that we take in change
- 18:15management. So a big part of my adoption role is data, which is
- 18:19solving for data. The second pillar for me is features and
- 18:23products. Whenever I want new, you know, sometimes you want a
- 18:27new way of looking at things, graph, PPT downloads, all of
- 18:31that. But the third part of it for me is something I call
- 18:34amendment and enablement, which is completely human. I need
- 18:38better comms, I need more use cases, I need a video guide. I
- 18:42need someone to work with me for 15 minutes to really show me how
- 18:46to log in or I have a challenge with this use case, can I call
- 18:50someone up and you know, help me understand how to work this out?
- 18:54Tom Ollerton: But how are you going like that's- As I said at
- 18:57the start, it's a massive job, you're your data and you're
- 19:00building features and products. I mean vibe coding, fine, but
- 19:04it's still work and then embedment and enablement and
- 19:08comms, that could be like a 15 people like team so how, how do
- 19:12you- And you got all the markets and you got all the timezones,
- 19:16how are you even doing that?
- 19:18Ishaan Kaul: So initially we started at the very global
- 19:21approach but then we moved into a lot more of a hub and spoke
- 19:24model. So as I said some of our champions are super users now
- 19:28have been trained to be the voice and to be the change
- 19:31managers at that local level.
- 19:32Tom Ollerton: And how does that work? Do they have like is it
- 19:35like a lunch and learn or is it like email or is it a events or
- 19:38is it all of those things?
- 19:40Ishaan Kaul: It's, it's a, it's a combination between in-person
- 19:44physical events as well as then coffee and learn sort of things.
- 19:48And then it's always focussed on a particular problem to solve.
- 19:53So we run a governance with our regional leads to synthesise
- 20:00what their users want to help prioritise what could make sense
- 20:06and then through them educate.
- 20:08Tom Ollerton: If you could summarise like the main thing
- 20:10that's coming back because you know, you talked about a e-com
- 20:14in Southeast Asia, but it's more about insight and learning in
- 20:17Europe.
- 20:18Yeah.
- 20:18Very interesting. Very similar mapping to our FMCG clients
- 20:21actually. What's the thing that's coming back? People just
- 20:24like, I just wanna do this quicker or you know, digital
- 20:27performance, like what's the main without giving anything
- 20:30away, like what is the, what's the main thing that's coming back?
- 20:34Ishaan Kaul: I think the best and the main thing that is
- 20:37coming back is people want to solve or rather use A.I. in many
- 20:41different forums, but there is so much that there is a big need
- 20:45to help understand where A.I. is useful and where it's just
- 20:49automation. And I think that's one of the big pieces we're
- 20:53trying to solve but from a human lens, what's coming forward is
- 20:56there is a lot of change and how do I navigate this change?
- 21:01Especially it's not just internal, right? Because it's
- 21:04all over. The signals are all over, not just internally, but
- 21:07externally as well. So people have a, I won't say far more
- 21:11anxiety, but a big will to really try to understand how do
- 21:15I work and move with this change? How can I use it in my
- 21:17work, which works both ways. It works both ways because it helps
- 21:22us get the right use cases to try to build on, but it also
- 21:30puts the problem of plenty because there's so much to
- 21:32solve.
- 21:33Tom Ollerton: So we have to leave it there. And I think
- 21:35that's quite a profound point is just navigating change, right?
- 21:39You got the FOMO people, you've got the leaning into it people.
- 21:43We're all navigating change, right? This stuff wasn't around
- 21:45three years ago, four years ago. And it is scary, it's
- 21:49challenging, it's interesting. And it's been very good to get
- 21:53like an under the hood exploration of how Reckitt are
- 21:58approaching it and, and your, your ambition and your process
- 22:02so Ishaan, thank you so much.
- 22:04Ishaan Kaul: Thank you.
- 22:04Tom Ollerton: Very quickly, if someone wants to get in touch
- 22:06with you, where and what, where should they do that and what
- 22:09makes a message you will actually reply to?
- 22:12Ishaan Kaul: I think LinkedIn is the best place. That's where I
- 22:15typically reply as well, what usually works is either a
- 22:22problem you're trying to solve that is in the area that I'm
- 22:25trying to work in. Learnings that you might have had.
- 22:28Anything generic is not advisable. In this space, if
- 22:31you're trying to solve a problem or if you have any learnings in
- 22:34around adoption, creativity, always happy to hear.
- 22:37Tom Ollerton: Brilliant. Thank you so much.
- 22:38Ishaan Kaul: Thank you. Thank you for having me.