Mel: [00:00:00] Have you ever asked ChatGPT, Claude, or Gemini what it thinks of your organisation? Whether you've checked or not, these AI tools are already telling a story about your brand to anyone who asks, and it's a story you didn't write and you didn't approve.

My guest today calls this your brand's synthetic reputation. Amrita Sidhu is the managing director of MediaNet, one of Australia and New Zealand's leading media intelligence and communication providers. She's also a board director of AMEC, which is the Global Body for Measurement and Evaluation of Communication, and she leads Asianet, a consortium of Newswire and PR agencies across the Asia Pacific.

I heard Amrita speak at the IABC APAC Fusion conference recently, and I knew I just had to get her on the show. Now, in this episode, we talk about what synthetic reputation is and why it's different from showing up in a Google search. What MediaNet's research found about Australian brands in AI answers, including why they skew so [00:01:00] positive and how far back these models dig for content.

We also talked about how to start measuring your reputation in AI without creating yet another vanity metric, and practical steps communicators can take from structuring content to becoming the organisation's content archivist. And if you wanna find out what that means, well, stay tuned. Here's Amrita Amrita, welcome to Less Chatter, More Matter.

Amrita: Hi, Mel. Um, thank you for having me. I'm super excited.

Mel: I am very excited to have you here because after speaking or hearing you speak at, uh, IABC APAC Fusion a couple weeks ago, I knew I had to talk to you more.

But before we get into today's topic, can you tell our audience a little bit about you? Who are you and how did you come to develop your expertise in this space?

Amrita: Yeah. Thank you. Um, so I'm Amrita Sidhu, and I'm the managing director of MediaNet, um, which is one of Australia's leading media intelligence companies.

I [00:02:00] also have the privilege of, um, sitting on the board of AMEC, which is the Global Association for the Measurement and Evaluation of Communication, and I'm also, um, a director of AsiaNet, which is a consortium of leading PR agencies, um, across, uh, the Asia Pacific region. But I think a big part of, um, what has led me to a degree of expertise in this space, part of that has come from the roles that I hold of course, but has been this kind of sense of, I guess, curiosity and this desire to learn and then share that learning with other people, right?

And in sharing that learning, you learn along the way too as well. And I think my curiosity about all things technology really started when I got an opportunity, um, a number of years ago to move to Singapore and to run a technology agency. And my, um, you know, the velocity at which I had to learn [00:03:00] about technology at that time, it has created this curiosity and this real passion and interest, um, for me, and that's what's really developed into a bit of an expertise, um, around communications now and technology.

Mel: Mm. And I love that too because I think you probably didn't grow up thinking, "I really wanna be interested in measurement when I'm older."

Amrita: Yeah.

Mel: But it's funny how that happens, isn't it? Like, no career paths e- ever seem to be very linear, do they?

Amrita: No. No, they don't. And, and I think that's the thing and, um, you know, underpinning measurement is data, and underpinning data- Mm

is technology. And so that's the technology piece, the curiosity about that that I think has really allowed, you know, all things communication, including measurement and how we show up to, to become a strength of mine, um, is... And I'm gonna sound like a techie here. I think the root cause behind all of that has been the curiosity of the tech- Yeah

um, that drives it. Yeah.

Mel: Well, y- we're living in the best time for this, of course, because [00:04:00] we are going through so much change so rapidly, which is part of what we're here to talk about- Yes ... today, of course. Yes. And so when you talk about synthetic reputation, let's start with that as the basics. Yeah. What does that actually mean?

Amrita: Yeah. So synthetic reputation I have, um, what I hope will be an accessible way of describing it, um, you know, to, to your audience and, and to others as well. And I refer to synthetic reputation as the version of your reputation, your brand's reputation, your organisation's reputation, your spokesperson's reputation, the version of reputation that an AI model assembles and then repeats back to anyone who asks a question about your brand, your organisation, you know, your spokesperson.

Um, and it is a reputation, as I also like to say, that neither your brand nor your organisation wrote or approved, right? Mm-hmm. Um, so it has [00:05:00] literally been synthesised from all this information, and it is provided back in an answer. Um- Mm ... hence synthetic reputation I think is a very, uh, strong way of, um, referring to this rather than just AI reputation, um, in, in my opinion anyway.

Mel: Yeah. And so how does that differ from, say, you know, your, I guess, traditional, in inverted commas, Google search of a brand or a product or a service?

Amrita: Yeah, so, um, it's really, uh, when, uh, when you Google search something, it's, it's two kind of different behaviours in, um, what I have learnt as well, right? When you go to Google search something, or this is before Google search started to become quite, um, sophisticated with its AI summary- Mm

at the top now, what you did is you'd search something. So for example, you might, um, search new chocolate collaboration. I say that 'cause I love chocolate, right? New chocolate collaboration. And you'll get a series of, uh, um, you know, [00:06:00] links in Google search, and it's up to you to scroll through and make a determination on, say, which link you're going to click on, right?

Mm. With an AI-based search, I could type a question and say, "Tell me about so and so organisation and so and so organisation's la- latest chocolate collaboration," and it is giving me an answer. It's almost like- Mm ... it's done that determination of which are the links that I should be, um, you know, referencing information from, and it's providing me an answer.

I'm not making a choice about which link. I'm getting an answer provided to me. And I think- Mm ... that that nuance and what it means for the way a user or a consumer engages with your brand or learns about your brand, I think there's a, there's a real distinction there that us as communicators, marketers, corporate affairs professionals, we just need to start getting our head around.

Mel: Mm. And it sounds like, you know, the traditional way of brands [00:07:00] showing up in, say, a Google search, was very much, you know, paid adverts. It was- Mm ... um, making sure that you're right at the top of that search results list, but also the SEO- That you work on behind the scenes. But now we're working on, we've talked about this a little bit on the podcast before, the GEO- Yeah

the generative engine optimization. So essentially we're now having to write for these large language models, LLMs, uh, which is a very different thing I know from my own website, but also working with clients. It's taking us back to things like FAQs. Like we- Mm-hmm ... at one point we took all of those off the website, now we're putting them all back on.

Yeah. That's

Amrita: right.

Mel: Yeah. What are some of the differences you've noticed?

Amrita: Yeah, so, um, a couple of, couple of things here that have really come out of, um, some very strong research that Medianet, um, did in this space. So we've done a couple of phases of generative AI citation analysis for Australian-based organisations.

And really c- wanna emphasize that point because, um, our research is very, [00:08:00] um, Australian focused, so hopefully very helpful, um, to those listening to, um, these podcasts who are, who are based in the Australian market or work for, um, Australian organisations. Now, a couple of areas where we found some really strong data points for listeners to be, um, aware of, and that's firstly, um, our research found that, um, the answers that you get out of these LLM models, they skew very positively in sentiment.

Mm. Right? When we looked at all the answers for retail banks in Australia, for example, um, for two of the leading, um, retail banks in Australia, 97% and 100% of the responses that these models gave back to them, thousands of prompts we kind of put to it, were positive. 97% to 100%- Mm ... were positive. So the brands are being painted in a very positive light.

Mel: Interesting.

Amrita: Why? Because these models are sourcing a lot of their answers [00:09:00] from owned content that brands and organisations are putting out. Now, media, mainstream media, it is absolutely a really important source. It's a really important communication stakeholder. But in Australia, because some of that really valuable media content is locked behind paywalls- Mm

or, you know, there's scraping blockers or there's different copyright restrictions, the models are going a lot to owned content sources. So that is corporate websites, PR newsrooms, blog content, authoritative, you know, articles as well, and it's finding information from there. And if this content is being produced by the organisation itself, you know, it's not surprising that these LLM models are using that, um, and in a way almost repurposing that into their answers.

So, um, th- they are positively, um, inclined in sentiment because they're sourcing 67% of the time from owned content.

Mel: Wow.

Amrita: So we're, we're finding those two [00:10:00] as really strong data points, um, about how an organisation shows up in AI versus, you know, on a SEO where, where you kind of paid or media where it's earned.

Right?

Mel: Mm.

Amrita: Yeah.

Mel: So it sounds like content, you know, we used to say content is king- It sounds like it's content is emperor now.

Amrita: Yeah. Content is king. Um, or as I also like to say, PR rising, as in PR- Yeah ... content rising. Um- Mm ... you know, marketing content, um, rising as well. Um, but, um, I think we have to be quite careful and mindful, and I'm sure we'll get to this over the course of this, um, conversation, is what we don't want as an industry, and I put my AMEC hat on as well as my Amrita hat on in saying this, what we don't want i- in, as an industry is manipulation, you know?

Mm. Going on here as well. We wanna stay away from that. Um, and so being quite mindful about how we try and understand some of this data, understand brand reputation, understand, you know, AI, understand synthetic reputation and what we can do about it. We also wanna make sure that, [00:11:00] um, the content that we continue to put out is really strong, valuable content.

And I think your point about the structure of the content- Mm ... that is really important as well. So yes, FAQs, blogs, lots of bullet points, you know, content optimised, um, that's easy for AI to ingest. Mm. Um, my marketing team probably doesn't like it as well, but un-gate your content. Mm. Um, because if it's behind some kind of form, chances are the AI model will not be able to access it as well.

So you know, there's a lot of these kinds of strategies, content strategies, that will enable AI visibility for your organisation.

Mel: Mm. And I imagine too, like we saw with the advent of search tools, like the early days of Google and Yahoo and Bing and all those- Mm ... those search algorithms will change over time too.

So the challenge is going to be keeping up with that, I imagine as well.

Amrita: Absolutely, and that's why I think, you know, we call this pa- this research series that we've [00:12:00] created our black box series, um, deliberately because so much of this is a black box. And think about these models, they're putting out new versions all the time, and the algorithms and the way they're sourcing content in those new versions of Claude, for example, or ChatGPT, um, I lose track of which number we're up to, you know-

with, with these versions as well. Because they're moving so rapidly and so fast- Mm-hmm ... um, you know, it won't necessarily be one size fits all, and I think that's why just, um, for organisations to start thinking about their synthetic reputation, putting this on the agenda of their kinda comms meetings is, um, is really important as well.

Um, on the content piece though, Mel, the other really interesting finding out of our research that I think is very relevant to your listeners as well is how far back the AI models go- Hmm ... to find this content. And I, you know, we found old blog content, we found, you know, regulatory penalties from a few years ago.

We found content that I know [00:13:00] organisations did not know was on their website- Mm ... because they told us, um, was all being used by these AI models in the way they were referencing that brand in the answers. Interesting. So it's not just about optimising content going forward, it's like, hey, let's go back and look at the content we've got, make sure that that's structured well, but is also accurate and still relevant.

And if not, you know, remove it from publicly accessible, um, you know, forums.

Mel: Yeah. It's becoming that curator as well, isn't it? Yes. And I think, uh, a friend Megan Thomas spoke about, uh, this a little bit at, uh, Fusion as well. She talked about one of our roles as communicators is really becoming that content archivist- Yes

in organisations too, which I really love that term- Yeah ... because it is. It's- It is ... going, going the archives and what's still relevant and what needs to go in the bin.

Amrita: Absolutely. And as communicators, we've always had to be so forward-looking, right? And of course we have, because we're [00:14:00] managing this, uh, brand, this reputation, you know, our own brands and reputations within organisations.

We've always had to look forward and/or react in the moment when there's been an issue or a crisis or something like that, or a positive thing, there's a launch of something. We rarely get a chance to look backwards. Um, that i- it's just like the nature, I think, of our, our kind of speciality and how fast we have to...

But that archivist thing, going back and looking back, I think it's really valuable as well.

Mel: And it's interesting- Yeah ... I was listening to a podcast yesterday actually about crisis communication, and they talked about this as well, how far back these LM- LLMs go for individuals. So people who are, like, famous celebrities or- Yeah

you know, prominent politicians or whatever. You know, it might be something that they haven't been in the spotlight for years, um, and then suddenly something happens and they are, and all this old stuff is still out there- Yeah ... and it's still being dragged through. So even from a personal brand perspective, it's about making sure that what's [00:15:00] coming up on these is, is what you want to see.

Um, but to your point- Yeah ... not manipulating that either because If you've done something wrong in the past, somebody'll find out about it.

Amrita: Yeah, they absolutely will. They absolutely will. Um, yeah, that's a good share, for sure.

Mel: So how... This is putting your measurement hat on- Yeah ... because measurement for comms people has been one of those things that I know a lot of people have struggled with, and I've struggled with.

Mm. You know, how do you make the correlation between, you know, what you did in the internal comms campaign and the business result?

Amrita: Mm.

Mel: I'm hoping that how we measure our reputation in an LLM, uh, result might be a little easier.

Amrita: Yeah.

Mel: What can we do?

Amrita: I'm hoping too. I'm hoping too. Um, look, um, my best tips on this piece around measuring reputation in LLMs is, um, firstly I'd encour- I'd really, really encourage everyone, um, to go to the amecorg.website, and I'll [00:16:00] tell- Mm

I'll tell you exactly why, right? This piece, you called it AEO at the start. It's also called GEO, LLM visibility. I'm calling it synthetic reputation. Wh- whatever buzzword you wanna kinda put on it, right? Before we end up with a whole bunch of dashboards and another vanity metric, um, AVEs, anyone, that, you know- Yeah

that takes years to rewind from. Before we end up there, we need to make sure how we measure in this space has discipline and rigor around it. Mm. And I'm very proud that earlier this year Amec launched, um, a GEO principles and a framework for how organisations could start measuring synthetic reputation and AI-led discovery.

There's seven very clear principles. They're not like, "You must do this," you know? Mm. They're really great thought-starters, and it's a framework, um, with which, um, an organisation can look at how to, um, start, um, a me- their measurement journey for, um, for, uh, AEO, GEO, [00:17:00] like I said, however you wanna call it, right?

Mm-hmm. Ultimately, though, it does come back down to what you just said earlier there. Mel, any measurement, it's like, what are you measuring? Why are we measuring it? And how does it fit with our organisational objectives or the- Mm ... communication outcomes that we want to achieve? So my tip to anyone is, who wants to get started here, is start small and start scalable because this is a black box space.

Mm. So establish some kind of measurement prototype, just a small one in-house, right? Which LLM model do you wanna test against? You don't have to test against them all. Mm. You know, which model do you wanna test against? Establish a prompt framework that's really relevant to that- outcome, that organisational objective.

Um, and then be really clear which aspect of reputation or measurement. There's a lot that could, we could be doing. There is a lot of data points, sh- Mm ... sentiment, share of model, spokesperson, themes, competitors, you know, all of this. So [00:18:00] which bit do you wanna get started measuring, right? And use some tools or some experts in this field, you know?

Shameless plug for MediaNet here. Yeah. Of course, we, we are there, um, to assist as well, but there are a lot of others that have developed a really nice expertise in this space. So you also don't have to do it alone. Mm. But start small, and start scalable, as you should do with any measurement journey, and start aligned with your organisational objectives, as you should also do in any measurement journey.

Mel: I love that. And I think just, too, with your AMEC hat on, because I do get this question a lot about how do I measure the impact that my communications is making, whether it's internally or externally. So let's say, you know, somebody's done a whole lot of social media content. Has it actually resulted in sales?

Or internal comms campaign, has it actually made people behave more safely? From an AMEC perspective, where could people start? Because I know that is something, and particularly now when we're really having to [00:19:00] prove our value as communicators, having that data and that measurement is going to be so much more important than ever before.

Amrita: Yeah. So AMEC has this, um, great tool. It's called the Integrated Evaluation Framework, IEF, also accessible, um, from their website, and it really makes you think and break down that piece that you just said. So it's like, what is just an activity? What is simply an output versus what is an outcome, right? And the outcome relates to the impact.

So something like me doing a post on LinkedIn, um, is, um, getting, um, you know, 1,000 likes and 80 comments, which I would love, don't get me wrong. Okay. That's simply, you know, um, activity I've done, right? And it's the outcome of my activity. What is the impact? Um, now if that then results in another opportunity that opens up from [00:20:00] me from someone who comments and blah blah blah, you know, that is more related to the impact of the activity that I've done.

So there's a really nice tool, and the way that it lays it out for communications professionals and really gets them thinking on that, I highly, highly, um, recommend. And, you know, happy to provide you, um, with the link to that, Mel, as well that, you know, you can link to in this podcast if you like. But that- Yeah

that I have to say, I mean, I use that even internally- Perfect ... myself at MediaNet as well. I think that's a really great starting point to re- to understand the differences between just activity-based metrics- Mm ... versus outcomes and impacts-based metrics.

Mel: I love that. Yes, and we will absolutely include a link to that in the show notes, listeners, so don't worry.

We will also include a link- Yeah ... to the Black Box research from, uh, MediaNet. Before we end the formal part of today's proceedings, what other, uh, insights can people expect to see in that research?

Amrita: Yeah, so a couple of other, um, in- insights, um, as well [00:21:00] is, um, I suppose the evolving role of social. Mm. Um, you know, a lot, I get asked this question a lot as well, how much is social a source, and that does really vary on type of social.

Um, so again, important to kinda understand. Um, a lot of social is unaccessible because accounts are private, as they should be, or community accounts. YouTube, um, was relatively prominent- Mm ... in our research as well because, you know, it's using the transcripts and the great metadata, and YouTube is, um, more open and accessible.

So I think that that's, um, a really good insight out of the research. And then we all love a little bit of competitor analysis, don't we? We all want to know how we stack up, um, against our competitors or those we benchmark ourselves against. And I would say the traditional media share of voice and the AI share of model are two very distinct things.

Mm. And we've got a couple of examples in that report of [00:22:00] organisations that are, say, very strong in traditional media share of voice against their competitors, but have pretty low profile when it comes to the AI model- Interesting ... and vice versa. So that's some really good- Mm ... um, research for your listeners to dive into as well.

Mel: Yeah, that's really interesting. Yeah. Thank you for sharing that.

Amrita: My pleasure.

Mel: Well, Amrita, we're now at the part of the episode where we ask- Yeah ... the three questions we ask every guest. Are you ready for those?

Amrita: Yes, I am ready for them. Bring it on.

Mel: All right, let's go.

Amrita: Okay.

Mel: What is an unexpected or left-of-field way you learned a valuable lesson about communication, and what was it?

Amrita: So, um, the way that-- The lesson that I learned was, um, when you need to go super tailored on your communication versus, versus broad, right? And it was my working in Asia- Mm-hmm ... and realising that this one approach does not fit all. I've always known that, but this really heightened it. Now, Asia is not one big country, obviously that [00:23:00] goes without saying.

And the nuances of the working culture in each, um, in each country, and the communication culture, how to understand that, how to respect that, how to build on that. I still remember to this day, Mel, being asked before we, um, went to Korea to do this big pitch for one of the world's largest, um, consumer electronics companies, that at the end of my team's pitch, I had to do something called a commitment speech.

And I was like, "What is this commitment speech?" And, and the advisor said to us, "They just need to hear from the most senior person in the room that you are committed to successfully delivering what your organisation has said that you're going to." Hmm. And it was just this super niche, super relevant to that culture, um, piece of communication I had to do.

Wow. You know? And that was really eye-opening. I have lots of examples of that, um, you know, through Asia, versus going broad when I have the privilege to present on stage, and you've got [00:24:00] a broad audience and, you know, you're trying to make it accessible to a very large number of people, whatever it is that you're communicating, versus this really very, you know, micro-culture niche- Mm

communication. Yeah.

Mel: Yeah. And it's, it's so interesting, isn't it? Because there's even nuances, not just across countries I find, but also across communities, uh, as well. Like, it can be down to those, those microcosms of people. Absolutely. Yeah. That's a great lesson learned. Thank you for sharing. Pleasure. Uh, next question: What is a book or a podcast you love and would recommend to our listeners?

Amrita: Okay, so I am a bit... I, I don't wanna-- Old school's not the, um, right way. I am a devotee of- Harvard Business Review and the stuff that they put out. I have been for years now. I, I just don't know why. Um, maybe it's because I, uh, had aspirations to study at Harvard one day. I don't know. But to this day, I love their content that they put out, especially their leadership series.

Mm. You know, some of it is available in, um, a podcast, but I still love to give it a [00:25:00] good read. The other one I wanted to make mention here, and it's not a, um, book, but it is a newsletter. Ooh. And it's called The Rundown AI. Ooh. It's free to subscribe to, also highly recommend. I said at the start a lot of my kinda expertise has been built out of curiosity.

Um, I love reading that, comes out every day. I used to understand about 10% of it. Now I understand about 50% of it, because there's still so much in this space that I have to learn and then be able to share with others. But that one in particular is a great one, The Rundown AI.

Mel: Ooh, okay. Yeah. Straight after this, I'm gonna be logging on and, um, seeing, logging onto that, so thank you.

And the last one, if you could wave your magic wand and change anything about communication at work, what would it be?

Amrita: Okay, so communication at work. Um, you know, we have, um, for the first time, I think, ever, I think the saying says, you know, four or five generations in the workforce. Um, and I would love [00:26:00] to see, uh, you know, more confidence, more assurity um, about s- their skillsets, their value from some of the younger generations, you know?

I'd love to see that. Um, I remember being that age, and I remember having the confidence to go up to senior leaders and, you know, have conversations, express my opinion. Um, so I would love to see more of that, um, through some of those younger generations. A lot of them do have it as well, but the skillset and the capabilities are just, you know, phenomenal, and there's a place for that in our organisations.

And I'd love, um, them to be able to communicate better about themselves sometimes.

Mel: Mm. And I think the flip side of that is too that older people accept their thoughts and opinions too and don't dismiss them because of their age. Yes. Yeah. Absolutely.

Amrita: Absolutely. Yeah, well said. Yes.

Mel: Well, I think that's a brilliant wish.

Uh, Amrita, if people wanna find out more about you, your work with Medianet or your work with AMEC, what's the best place for them to find out?

Amrita: I [00:27:00] would say LinkedIn. Mm. Um, check me out on LinkedIn. Connect with me on LinkedIn. I'd love that. I do put some of my thought leadership up there as well. And then, yeah, message me.

Um, like I said right at the start, I am all about helping shape the industry, changing the conversation, but sharing knowledge. So I would love nothing more than to hear from one or many of your listeners saying, "Hey, Amrita, tell me more," or, "Can we just have a chat?" That would be, um, an absolutely beautiful outcome from, um, having the privilege to be on this podcast with you.

Mel: Oh. Well, hopefully you will get a few people reach out. I know I'm following, uh, my data nerd hat on in close- closely along. And of course, you're off to Tokyo next week, you were saying, for- Yeah ... what's the event you're going to there?

Amrita: I am going to the World Media Intelligence Congress in Tokyo, um, which will be really exciting.

Um, and, um, the, uh, theme is Kaizen, which is the Japanese word for continuous improvement, right? Mm. And isn't that what we have to be [00:28:00] doing, um, as communicators as well? Just looking at all the areas where we can establish ourselves in this pattern of, um, you know, continuous improvement. And, yeah, I'm real excited.

It's gonna be three great days of programming and lots of networking, and I think hopefully just meeting, learning, absorbing like a sponge, and finding opportunities to share that.

Mel: Yeah. Well, it sounds fascinating, so I'll be looking for you on LinkedIn for all the things that you've learned- Thank you ... in Tokyo.

Yes.

Amrita: I will be.

Mel: Um, Amrita, thank you so much for joining us on Less Chatter, More Matter.

Amrita: My pleasure, Mel, and, um, thanks everyone for listening.