Media Briefs: Terrapinn’s Sharon Roessen on boosting event registration and attendance with agentic AI

Speaker A: Foreign.

Speaker B: Hello and welcome to another Media Briefs episode from Media Voices. Short Sharp sponsors episodes this time featuring a senior executive working with a vendor uh, that is making their business better. This episode I'm speaking with Sharon Roosin, Chief operating officer at global B2B events business Terrapin. She's working with Bridge Media to develop a series of AI agents designed to streamline customer experience. Sharon spoke about the development process from identifying a problem to developing a solution. And how one agentic AI experiment has led to a suite of agents that are increasing event attendance and easing registration bottlenecks. I started my conversation with Sharon by asking her to tell me a bit more about Terrapin.

Speaker A: Terrapin are a business to business exhibition conference owner operator. We've got um, about 90 events globally. Our head office is London. It's our biggest office, uh, Dubai, Singapore, uh, Australia and a small satellite office in the US as well. But we do events globally. So our events are uh, global and kind of our SU spot around the event number is probably around the 3500 to 5000 attendee mark, um for the Confex style event that we do and then we've got the large scale exhibitions and I mean they get up to 35,000 registrations and across a multitude of different industries.

Speaker B: But it's B2B.

Speaker A: It's all B2B. Yeah, it's all B2B.

Speaker B: So in terms of your role, what you're responsible for with those 40,000 attendees?

Speaker A: Well, so, well my background's marketing. I've actually been with the company 30 years. Um, but yeah, I mean my background's marketing and I rose up to be cmo. I then went off and ran our Asian business. So I went on the business side. So then got exposure to actually running the shows myself and being responsible for P and L and then came back just after Covid and came in to the chief operating officer role, came back into that and we were a much bigger company over the course of that uh, eight to ten years that I was away. And so now responsible really. I have a functional head. So I have a head of sales, a head of marketing, head of operations, HR and tech and they all report into me. So uh, we're really responsible for the running of the business, not running of the events.

Speaker B: Okay.

Speaker A: Yeah, so we've got just about uh, the 380 staff now I think we're at globally. So they're spread obviously throughout the offices.

Speaker B: Is that spread fairly even between the different offices or you got a concentration in the uk?

Speaker A: So our biggest office is the uk, uh, do the majority of the events and we have at least 50% of the staff based out of here. But each office has its own infrastructure. It has its own marketing team, its own sales team, its own production team. So we don't centralise in that way. Um, each office has its own, its own mini setup. But we are centralized from a tech perspective, so everything that we do we tend to test once or pilot once or twice and then we roll out. So if you went to work for us in Sydney, you'd walk in and get pretty much the same tech stack as if you were working for us in London or in Dubai. So, yeah, so we can share a lot of the insights and data, um, across the business.

Speaker B: Is that from an investment perspective or is it actually from that data sharing point of view?

Speaker A: Well, initially it was really from a data sharing perspective because we're really one of our kind of sweet spots as a business has been geo cloning. So, uh, like our solar and storage brands, I think we have about 19 of them now and they're run locally. So Asia does the Asian events and Australia does the Australian. Dubai do the Middle Eastern and UK do the European. Yeah. So it's really allowed us to leverage that geo our brands geographically and um, because we share the data. So the data is owned by the, by the central business, not by each business in that way and we can have access to it. We've been working like this for, I think. Well, I think we first put the first marketing database in about 2000. No, even earlier than that, 1998 or something, when we first started putting the data together. Um, and so then we've always had that foundation. So actually now with AI, that gives us quite a good step up the ladder because we've already got the data in one place or in one or two places, but we've all got access to it and everyone's feeding into it and off of it. So that gives us a bit of head start there.

Speaker B: Ah, what do you use that data for? How does that help your business?

Speaker A: Obviously marketing, we can do lots of cross sell across that from attendee perspective, but also, um, on a sales perspective. So, yeah, so our sales team have full access to all the data set. If they have a show in Singapore, they can grab that data if they work in London or in Australia and they can go and upsell or cross sell to their attendees. So the big play really has always been in sales and marketing, um, also from production because you can then go and do the same with speakers. But then it's also given us a really nice transparent layer over the entire business. So, you know, the senior team can sit anywhere in the world and can see exactly what has been sold today, what registrations have come in, what's happening. So we've got this common language, this common data structure across everything and it makes all that reporting up, uh, and down a lot easier.

Speaker B: What's interesting about that, uh, and in publishing, a lot of people talk about that single customer view, but actually you've turned that on its head there, right, because you've got a, you've got a company view. You've got a single company view on the industries that you're working on.

Speaker A: Uh, yeah, and a single customer view. Although there's dupes all over the place, but it is all in one place.

Speaker B: We won't tell anyone that.

Speaker A: Yeah, yeah, bloody hard. Particularly when you're dealing with a lot of international data, it's really hard to get that completely clean. We've been talking about single customer view for about 25 years, literally. Um, but the other advantage that it's now giving us in an age of AI, well, in any tech really, is we can roll out really quick and in theory means we can build something or buy something, test it and then roll it out really quick because it's all going off the same platform. So it's one API or it's one piece of code. It enables us to be able to scale, in theory, quite quickly because we don't have to keep building a different reg system in one office or to one event. Much easier operationally to roll out.

Speaker B: Tell me about this project that you've been working with Bridged Media on.

Speaker A: We've been working with Bridged for, I think we probably started sometime in the summer last year. What we've got working with them is three different agents and then we're scaling that over all of our events. We must have over 100 different agents through Bridgd working across the portfolio. But we started, um, by just putting an agent on the website. So we didn't have a chatbot. We had it on some websites, but it was, um, not very well adopted. So we started by putting a web agent on, just on the website to pick up web visitors questions. So relatively simple. I think we piloted that on a few events. Now we're running over all our shows. I think at the moment for this year, it's on 58, nearly 60 of our events this year just got, I think, Q4 to get loaded up. Um, but then coming out of that, then we looked at what else we could be doing. Um, and we've, we've always been, well, not always but in the last 18 months or so been very, very interested in WhatsApp as a marketing channel. Particularly I think, I guess from my side, coming From Asia where WhatsApp is like the main communication channel and it is in the Middle east as well. Looking at how we could leverage that as a channel and email becoming harder and harder and harder to get delivery rates in. So then we worked on, we tested a few different providers actually. Um, and then with Bridged we built it out as well and it's just proved very scalable. With Bridge we've got two parts to it. We've got one where we send a WhatsApp message to all of our uh, attendees that haven't yet registered from previous years, so their past customers not yet re registered, um, and try to get and ask them to register. It's all for free exhibition tickets. Um, and the agent will then it will literally say to you on WhatsApp so you don't have to download an app. Assuming we've got your right mobile number, it will then take your registration on WhatsApp. So it will say, let me confirm I've um, got your details correct. These are the details, yes, they're correct. It will register you in Salesforce, which is our registration system. Push that all out, customer gets their ticket and all done and then we'll try and engage you as a concierge and say is there anything else I can help you with? And then it goes into this concierge mode which will then go and try to keep you engaged. If you've got any questions about the event, you can go on to WhatsApp and we've got that one to one communication channel then with each of the customers. We piloted it last year, we run it on 15 shows, we've had eight. That's finished the big play for me, the bit that I'm really interested in there is obviously how you can increase registration and obviously one of the easiest ways to do that ah, is try and get as many of last year's attendees to re register and then show up which is then our third agent. So then we have a third agent which is, it's that concierge piece. Um, but what we do is we serve you your badge through WhatsApp in the run up to the show. So by the time we've emailed you a few times and you've not answered it, then you've got, we've gone into Spam and it's, you know, we've lost, we've lost that channel with you. Um, but with WhatsApp we can then send you a QR code of your badge and we can do that in the run up to the show. So we might do that two or three times to say, you know, hey Peter, remember the show's on tomorrow, here's your badge. When you get to uh, registration, scan this, get your badge really quick. And actually of all of them at the moment, this is where we're seeing the biggest success on this one.

Speaker B: Maybe hard to put numbers on this. Comparing WhatsApp response, um, with email response, what kind of uplift are you seeing?

Speaker A: Attribution is a constant problem. Typically what we're doing is we're emailing you first and we're trying to get you to register by email. So any of the really low hanging fruit, we're getting through that. But on one of our shows in Africa, um, we had, I think it's around four and a half thousand people register through that WhatsApp channel, which I think accounted for close to 20% of all registrations registered through that WhatsApp channel. That's Africa. But then we've run it in Australia and had very little response. We run it in Malaysia and we get a really great response.

Speaker B: I was going to say, I think there's probably a regional, uh, well, maybe continental rather than regional but uh, geographic specific responses. I personally can't imagine myself doing that through WhatsApp, but if that's the way the M market is working, then absolutely. So it adds another string to your bow, right?

Speaker A: Yeah. And actually from a cost perspective it's really, I'm not going to say cheap, but affordable to try is a pain in the ass to set up. Working with Meta and WhatsApp is a nightmare and it takes a while to get. So we're setting up WhatsApp business accounts by brand. So that takes a little bit of the pain. Now we don't have to do it 90 times, just 39. But it's um, it is painful. But there is that. That whole attribution piece is quite hard. But the piece that I've been looking at that I've been quite excited about is this show up rate. Our solar event in Malaysia had a 16% uplift on show up from registration to attendance. Now I don't know if it was because we sent them a badge three times on WhatsApp. No. Without asking everybody, what did you come because you got that WhatsApp message? I don't know, I just know the lift from the previous year was very big. Um, maybe something happened in the market,

Speaker B: but, you know, in some ways you don't care, right, as long as it's moving in the right direction.

Speaker A: Well, that's right. I think the point about Africa and Asia, they're used to doing business on WhatsApp, so they're, they're used to asking questions and engaging with a business through WhatsApp, whereas I think in Europe, I mean, US is a, uh, different ballgame because they don't use WhatsApp. But in, in Europe we're probably not as open to, you know, to engaging with a business on WhatsApp. Um, but I think it will start to come. I don't think people, you've got to ask, you've got to push them a lot more and prompt them, which means you've got to pay meta to send out broadcast messages to get them to interact. But we had a show in London end April and that saw a four and a half percent lift in attendance. And again, I don't know, registrations were up, um, but the, the percentage points, uh, moved well about 4 or 5%, which is quite significant on a, you know, on a 15, 000 red show, you get another 5% show up. That can have quite a big difference on your attendance. But was it the agent? I don't know.

Speaker B: What was the thinking? Was this something that you had a problem or that you thought, well, this might be an interesting way to try and increase that uplift? What was your starting point?

Speaker A: I guess the part with this show up piece really came from how can we make sure you get your badge QR code? And so really the problem was queues on site at events. So people coming and then we're saying to them, have you got your QR code? And people go in, oh, I don't know, maybe. And then people saying they're scrolling through email. Well, yeah, people printing stuff out, bringing it in. Um, people then standing at reg, holding up the reg queues, you know, scrolling through email, trying to find it. Or you have to put a human behind the desk who's then going to find you by your email account. And we wanted the queues moving really quick. Um, so that was actually the original problem that we were trying to solve was how can I get the QR code onto a device that you are going to guarantee to have with you? I'm going to be able to find it really easy because if we sent it to you in the morning of the show, you're going to have it pretty high up on your messages coming out of that. We then thought, well if actually we sent you your badge a bit earlier, maybe we could increase show up rate. And then coming out of that we said, oh well if we can do all of that, why can't I also take a registration? Can we get people that haven't yet registered that have been previously? Well we have their mobile number and their permission. Can we get them to register because we've lost them through email because we're already in there spam folders. So then that's, that's how we've led down to, down into that route. So yeah, new ideas coming up all the time on what we could do with it. And for me the power is having that, that almost one to one channel with a uh, with a customer because Once you're on WhatsApp it feels when it is one to one you are then having a conversation with them directly, not broadcasting.

Speaker B: And is that all automated?

Speaker A: M. Totally. The only bit that's not is we have to, when we send the broadcast messages out, we do that through another third uh, party provider that sits in the middle to actually broadcast it out and then you're paying on top. It's like paying for email, only more expensive. But one of the challenges that we are now facing as we scale to 90 shows is whose job is it set all this up? Someone's still got to set up the agent at the front end and you've still got to teach it about each of your 90 events. So that's, you know, someone's got to teach the LLM about your event and then teach it what questions. I think now we've answered something close to 50,000 different unique questions. So we've now got that knowledge source so we can feed that back into the teams. But you've got to be doing all of that.

Speaker B: So what's Bridge Troll been in all of this? What are they doing for you?

Speaker A: They built it, they built out the platform, the WhatsApp platform, m the web platform or um, the concierge, the language model behind it all. Um, they build out all of that and then we've integrated it with the, with a Weber account so we can send out the messaging. So yeah, so they've built all that out. They've now built a portal so that hopefully we can get some of our marketing people self serving it so it can come out of our central tech team. But yeah, they, they built all of that and continue to build on top of it. So now we're looking to what's next, looking at how we can monetize this. So we're looking at something that bridged are calling topic takeovers where we could sell topic words. So if we've got 50,000 questions being asked and says, you know, um, what solar panels have you got on display at the solar event London? People could buy those as keywords, a little bit like search. So we're going to be piloting that on uh, three shows in September.

Speaker B: Do you think that idea of opportunities coming out of the process, you thought, okay, I've got a problem, we've got people that need their badges quicker. So you solve that problem and then you see actually we could answer questions for people and then you say, well actually we could possibly monetize that as an iteration going on there.

Speaker A: Yeah. And actually that's. Your question was about bridged. And that's what's been great about them because we can sit with them and say, oh, could we? Um, is that possible? How could we do that? And because of the way that they work, they build all of that out themselves. We don't generally have to commit for three months, but it's, it's a low level. And then we buy everything on tokens. So if the customer doesn't use it, we don't use the tokens, we don't pay for it. So everyone's then uh, incentivized or we're charged on success. So if it's really successful then if we're getting loads and loads of questions being asked and there must have been a problem there in the first place, then we'll get charged for it. So that's also allowed us to keep thinking of new ideas. What else could we do on this? So we're now looking at building a calling agent with them. Um, because it'd be nice to say, okay, we've emailed you, you haven't registered, we've sent you a WhatsApp message maybe two or times, you're still not registered. Maybe we could put a calling agent on you now, like let's keep trying to get you to come back again as a visitor. So we are, we're looking at how we could take that another level into a different channel so we could have them as multi channeled and, and tie the communication together as well.

Speaker B: If I'm an event organizer or a publisher and I haven't even started on this journey that you're on, what's my starting point and what should I be looking for in a partner?

Speaker A: Um, well, my Biggest thing at the moment on what to look for in a partner is someone that's not going to tie you into a really long contract. We've been actually pulling out of long contracts, you know, where they're trying to tie you in for two, three years. Because everything's just changing so far. So that's why for us, uh, Bridged has been so great. We've just had to buy another chunk of credits because it's been successful, it's being used. Therefore, I'm only paying for stuff that's been successful rather than tying in. And one of my favorites in the event industry is everybody seems to want to charge you about 10 grand an event. Well, you've got 90 events. You can do the maths. That's a lot of commitment to go to the board and ask for, you know, or I can only ever trial it on the top events. Um, and maybe they're not the ones that would be best benefited from it. So the great thing about this flexibility in being able to do this means that we can turn it on and off. So, like now we're learning, now we've got it going on here. We might take it off the Australian events because we're going. Actually, we're seeing very little lift. Maybe it's just the events we've done it on start of the year, but as we go through true, we can be a lot more agile with that as well. So my, my first advice would be don't tie into big contracts because it's just all changing so quickly. I think that's, that's the biggest thing. And, uh, you can probably safely bet that if. Well, if we're staying with someone or using it in that way, then it's probably being successful for us because we're looking at that from a, from a macro level across all the shows rather than just on one individual event.

Speaker B: What I think is interesting about the way you've described how you start us, you start with the problem, right? You don't start with a bright new sh over in the corner. You actually start with a problem in your business.

Speaker A: Yeah, for as it was that QR code, which actually, when you look back, kind of felt like a micro problem, but it was one that, you know, when you've got a big show and a lot of reg, it was causing quite a lot of angst. And then that's just led on and on and on. My other thing is, uh, we've started our AI journey really looking at revenue and attendance rather than looking at how we could reduce headcount or make people's lives our own, um, internal lives simpler. We are looking at a lot of that as well. But really where we put the investment and the time originally is how can we move the needle, either by revenue or by attendance. And of course the two things are linked. So that's really where we've started. And if you can then prove that AI is impacting there, then, uh, I think you can go out and build a business case on how AI can help in other areas as well without spending a fortune. Because the worst thing you could do is commit to a long contract, a big amount of money, and then it fail.

Speaker B: Thanks so much to Sharon for speaking with me and to Bridged Media for sponsoring this Media Briefs episode. Bridged Media is working to democratize AI for publishers and events organizers, helping media teams scale AI beyond experimentation. They provide decision frameworks, operating systems, and measurement practices that turn AI from isolated tools into repeatable organizational capabilities. Learn more about Bridged Media's media think tank approach. On Bridged Media.