EP 286 - Using AI in Marketing Without Losing the Human Touch

Fab: hi, Rebels. ⁓ Welcome back or welcome to all marketing school. We're here to teach you how to make your marketing more impactful, human and fun. But also we want to keep up with marketing so you don't have to.

In case you don't know me, I'm Fab, your marketing fairy good mother and the head teacher at the school. And today I'm joined by the incredible Archana and together we're going to be talking about, yes, AI is one of the big things that we need to keep up and we want to help you out with that. But also in a way where we're going to look at the combination of the human touch and building it into how we think about how we use AI, how we use AI itself, and then also how we show up at the boundaries and constraints that we build.

So as well as asking you very kindly just to set the scene a bit more about obviously who you are and all the incredible work that you do, I am also going to ask you right after now to share with us. two truths and one lie, please don't tell us which one is the lie because we want to guess. But yeah, I would love to hear a bit more about your Fork in the Road moment, if we may. How did you end up doing what you do and being so passionate about the work that you do?

Archana Dhankar: you for having me here. It's been a pleasure. Well, it's been what, 20 plus years now in marketing and especially in B2B marketing as well. To be honest, I didn't think I would end up in marketing.

I've done a computer engineering degree and then ⁓ I ⁓ doing project around search engine algorithm for Google. That just me to understand how interesting this world of digital marketing is. And at that point, wasn't any like playbook, there wasn't any courses. ⁓ In fact, Google didn't even have its certifications back then.

So I was probably the first 100 people who got onto their certifications by Google when they actually went live. it's been a fun journey since then. I've worked with many startup brands. ⁓ I've always looked at what's happening at the forefront of marketing, whether that's influencer marketing, how B2B influencer marketing works, and now AI.

AI feels like a natural segue as well, also because of the technical knowledge that I have and the way I understand systems. Today, I work as a VP of marketing for Proofpoint, which is a large cybersecurity vendor. So really looking forward to talking to you all about AI. If anybody wants to later on connect with me on LinkedIn, have any questions that are not answered here, I'm happy to take that as well.

And now back to truths and one lie. first one is I enjoy both coffee and tea. Second, ⁓ I can dance anywhere even if there is no And the third is I've spoken at the UK House of Lords. Fab: So I am highly hoping that number two is true because I'm obsessed with dancing anywhere, especially bus stops, which is the best thing for anybody else, I think involved.

So I really hope that's true. Livia, live with us. Hi Livia. that one is a lie.

I think there was something about how you phrased it that got her little spider senses tingle. I love that. Livia, I'm going to jump on that bandwagon. I don't know if anybody else feels like guessing, but I'm going to go with Livia and say that one.

I enjoy both coffee and tea is a lie. Archana Dhankar: drum rolls, I think I should reveal the answer now as well, right? So first of all, you are right. I really enjoy dancing.

Like dancing is one of my passions. And as a kid, I used to tell my mom, maybe I should be like a classical dancer. So yes, I can dance anywhere. I think it's not appropriate to dance anywhere and everywhere.

So that's another one that I have to like restrict myself on in terms of like the lie. You're right. I'm a big coffee addict. I'm okay with tea and I still try and dabble in herbal teas and everything else but even though I come from India, tea is probably not my favorite drink and my family probably just doesn't like that either because tea is like a culture where I come from.

Fab: What is your favorite coffee order? As we're jumping quickly into our chat, I just had to ask you that. Favorite coffee order you go to. Archana Dhankar: Yeah, I think if I'm not counting galleries, then it would be a hazelnut cappuccino.

Fab: very, very nice. Just to make everybody a tiny bit more jealous. I am kind of enjoying sometimes a little bit of a syrup myself. So, and more coffee addicts as well live with us.

Olivia is like, hello. I I seen. So I absolutely love that. Thank you so much for sharing a bit more about yourself as well as obviously your passion and also your background.

We are talking about using AI without losing our voice. ⁓ And I think that as well using frameworks and steps, there's actually so many questions that we have around exactly that, that losing our voice, that losing ourselves in it. And this is kind what I wanted to tackle now. We talk about the human themselves, but also I'm almost kind of wondering, how do we know ⁓ when it's time for us to trust these and when should we question them?

And maybe even how should we question them as well? Archana Dhankar: Yeah, so I think there's a popular model which has been thrown around, which is called Human in the Loop. It was coined by Google and a couple other companies. It's called Hittl, actually.

So you trust AI for patterns, and then you trust AI, human, to actually decide. Now, AI can help you look at a lot of high-volume data. You can look at all the measurable results, and you can look at all the repetitive patterns. But you then have to have the rule, is AI drafts, human decides.

Yeah, I can suggest, but it would be human who would sign off. So everything is actually based on a human steer at that point. Fab: I love that. It's that kind of almost like, yeah, steering and taking that next step forward.

sure I think it's really, really important as well. Now gets into again, talking about customers again, and talking about customers data. How do we go back to the ethical piece? I think, and if we are using some support and some help.

But we also want to get the benefits of obviously AI without compromising the fact that some of the data will come in. I think this is a more of a technical question, but I had a variation of this question about 15 different times. And so I want the experts to help us out on that because it's very important to me. And I think it's such a big ethical boundary in Gardeley that not many people talk about or know how to talk about.

Archana Dhankar: Yeah, I think this is very important. And I was talking about this yesterday as well, which is like, how do you make sure that your employees are treating data in the right way, right? With AI tools and so I think training the team itself and just having them understand the risk of data being shared with AI the not so approved platforms essentially, because all businesses today have their approved AI tools. So we should all be looking at those, but we have to also treat data like trust.

Once gone, destroyed, you cannot have it back from your customers. And if you as a company, if you as a person cannot openly talk about it on stage and think like, this is how I've used the data, then don't do it. Right? Like that's where it is.

Like if you're using data in your approved platforms and your IT team, or if you're a small company, have you looked at all the guardrails? It's very important. Think about consent. Think about transparency.

And what is the minimum data needed? Do you need to put all of the data? Do you need the PII information to go into the tool? Those are the things I would think about.

Fab: One quick follow up on that, thinking about my little rebels and thinking about a lot of them being either like one person band, sometimes even consultants and working with clients. What would you say is, is there a place that we could go? there any research that we could do to understand and to get a bit more familiar with which tools? Which tool should we use and which limitations should we have?

You for example, some tools like the use the training that use the training, where should we go? I know it's a big question, but even if it's just like one thing they can do, because I think in that extent, knowledge is power. Just understanding what to choose for what is also powerful. And I wanted to touch on that quickly.

Archana Dhankar: I think this is thank you for further asking that question. So what I would say is number one, you are a smaller organization, think about what, as I said, minimum data that is required. So start to de-anonymize and only share as ⁓ as much as needed. The other one is to also update your own policies, right?

Like your DNCs, how you're using, just be more transparent about it. Are there certain tools that you're going to use? and be specific of what those tools will take as data point and just have that list with you. That would really help.

So here's the data that I might put into Charge JVD versus these are the data points that are going into my email software would be clarity that you have to have. Fab: I love that. That's a great starting point because I generally know it's such a big, I think it's a bit overwhelming for a lot of people that are like when the realization comes in, right? Oh, actually am I keeping that data safe?

And then it's like, fine. How do I actually even know whether I should change anything before I go back to doing it as well? I think this question is another big one that comes up a lot. And I think it's more potentially it's about things to do with and just things to think about in the bigger scheme of posting and sharing online.

And then as we say, we scroll and it all sounds the same. If we still want to use AI for creativity in some ways, but we don't want to lose our brand voice. Is it still a case of air drafts and we review? Is there anything more that we can do at the different ends of this journey to actually help with not losing our voice, especially when it comes to our brand voice?

Archana Dhankar: OK, yeah, I think this is a great one. If you look at what AI produces, AI is just a mirror of what you feed into it at this point. So probably AI is not at fault in losing the brand voice, but a weak brief probably is. So we have to train AI.

That's where the context layer is so very important. We have to train the AI with our brand voice, with our brand context. ⁓ what the foundation would look like, the golden rules and the nuggets before we actually expect AI to produce something that would be more personalized to us. And a lot of marketers, a lot of people, it's okay to just brainstorm, but if you are creating your own brand content, whether it's a small organization or a large organization, we should make sure that at some level, the AI has those guardrails and understanding of what the brand is.

Fab: And the fact that you mentioned that it mirrors what we put in, which we know, but it's kind of that little tough love that we need sometimes. It's just that understanding that if the input is just, I mean, this is the most basic thing, announced that I have been promoted, then, you know, it's going to just say exactly what you want it to say. But as you say, it's just that, and this is a very simplistic example, but it's just to give you an idea. Sometimes that's what we ask as well.

And I think that's the thing, even if we start a prompt in a very good, in a very good way. by giving it the role, by giving it some of the constraints. If the context is still very little, I think the more context you can add, the more you can work with it almost. I feel like, and this is the final thing I wanted to say on this, it's a bit like if I have somebody else joining my team to do something for me, as a person, I will not just say, just go into our next email without context.

I will have, and I had that because somebody actually now does draft emails. but I gave her as much context, as many instructions, as many examples as possible so that she could do independently and self in the best possible way. And then she came to me with questions. If you think about AI as that junior team member, you will not, if you onboard somebody in a very poor way, they're gonna give you a poor output as simple as that.

So you're basically onboarding lots of people all the time when it comes to thinking about AI as that support. I don't know if it resonates with you as well, but that's how I see it. Archana Dhankar: 100%. I think that's a very great way of coaching somebody.

I like it. Fab: Final question for us, unless there's anything else also popping in the chat, I'll keep an eye out. See, I think that's kind of like, it's a similar question. So if you're using AI content, doesn't sound like me.

How do I scale my content without losing my voice? I'm going to add something else actually to that, because I think we kind of answered that question already a bit. How, context do you think is more interesting and relevant one? that we can bring when it comes to our human insights.

If we're looking to actually get our AI to better understand our brand, is it the usual common sense stuff? Is it more about examples? Is there anything else that we can add? Again, maybe data that we can share, that we think is interesting to really, you know, kind of give AI the best understanding of who we are as an organization or has people.

Archana Dhankar: Yeah, I think this is a great one. And I was thinking the same as well. So if you were an organization, there are different, and based on ⁓ large LLM that you're using, whether you're using Claude or whether you're using ChatGPD, think about how you want to train, what memory you want to save into your ChatGPD, and what commands you want it to have. So that could be ⁓ brand voice, the brand guidelines.

That could also be some previously human-related content of what what content, good content looks like for you. Here are my 10 great social media posts that I really like. Here are the best performing emails so that the AI now has a full context and a library of things to refer back to. And then once again, they have to be refreshed, right?

Like six months down the line, those things will not look the same. So you have to go back and maintain. So there's a whole maintenance and governance that needs to keep happening as well. Now for my own usage, what I've also done here is, I've created my own voice GPT and I've trained it by telling it a story about who I am, what my journey looks like.

So now it's almost like a mini version of me and it sometimes sound synthetic. again, the content out, I still review it, but it has how I talk as a person rather than just polished outcomes. Fab: I feel it almost gives you that starting point and kind of like that angle and potentially helps you realize what can you dive deeper into. And then as you say, you kind of bring yourself back in and just make sure that it sounds more like you.

you for answering that. Cause I think. There are still things that we see online, there are suggestions about what to give and what context to add, but it's always good to of dive a bit deeper into that as well. So I really, really appreciate it.

Now, if you have any specific question or not specific question, but still questions around this marriage of AI and human insight, thank you so much for taking the time to share all of said knowledge. Archana Dhankar: Thank you. I think I'll just want to add one more thing before we finish as well is ⁓ the brands or the marketers, the ones who will win will not probably not be the most automated. ⁓ It's not just, know, there's such a race right now to just do every single automation workflows and build agents AI, but it will be the brands who will still be able to retain.

⁓ the brands or the marketers, the ones who will win will not probably not be the most automated. ⁓ It's not just, know, there's such a race right now to just do every single automation workflows and build agents AI, but it will be the brands who will still be able to retain. ⁓ or marketers will still be able to retain their human voice at scale. or marketers will still be able to retain their human voice Fab: We got a couple of hearts from that.

think that's what of us needed to hear today. I think that is the of it is making that decision of where to use and how to use and how to bring ourselves back in. So you give us so many incredible examples of how to do that, ⁓ how bring ourselves back in in a very tangible way. And I'm incredibly appreciative of that.

So thank you so much again for all of your knowledge. Archana Dhankar: Thank you, Fab. It was a pleasure to be here to talk to you. Looking forward to catching up soon.

Fab: Where can people find out more about you or ask you any questions if they don't want to catch up? Archana Dhankar: I think you feel free to connect with me on LinkedIn. My profile is Archana Dhenkar. So just search, connect.

I'm fairly responsive on LinkedIn. So that would be the best place to connect and talk to me as well. Fab: We've got lots of thanks as well. little gratitude drain here, Olivia, Olga, thank you so much as well.

So sending all the thanks.