Toxic content online can kill your brand faster than COVID, but there is a solution available now.

Jim James: 0:00

Hello and welcome to this episode of the UnNoticed show. Today, I've got Ant cousins. Who's the CEO of a company called Factmata. Ant, thanks for joining us on the UnNoticed show and tell us how do you help companies to find out what facts about them do matter?

Ant Cousins: 0:14

Thank you. so we are we're in the media monitoring space overall. so I think we ended up there through trial and error and like a lot of companies heading into this space. We started off actually in the fact-checking space back in 2014 when our kind of CEO and founder Dhruv Ghulati started working on this problem academically. and scientifically like a lot of people did back then. and over the course of the last few years, we've ended up refining our vision, refining, refining our approach which has taken us into the media monitoring space, as we realized that, fact checking in and of itself was useful, especially around pinch points, likely, election debates and things like this is a really great time. but the commercially sustainable, scalable, exciting story. wasn't there just in fact checking, actually, there's a lot of other harmful content out there, not just mistruth and disinformation and fake news. We've also got racism with sexism. We've got toxicity, hate speech in general. We've got a lot of other things that are actually harming ourselves and having a society in harming brands and organizations. So that kind of broadened our scope from fact checking to basically finding lots of harmful content and that moved us. Close to towards the media monitoring brand protection type space, which is where we are now. So we are comparable to some of the players out there that you've probably heard of the bigger players like Brandwatch or the smaller players like that, the media monitoring systems like social mention or mentioned those kinds of guys.

Jim James: 1:36

So you've mentioned those other ones and obviously for anyone running a company, If you get negative reviews, for example, on Trustpilot or Google or if you're like AstraZeneca or one of these vaccine providers at the minute, you can find your brand and your business and your share price attacked pretty quickly. You've got to respond. So how is Factmata really answering that need that's different too, if you like social mention or even Cision and how is it special?

Ant Cousins: 2:05

So those companies came out of the way that PR was previously run. Another way PR was up until very recently is basically get our volume up as high as possible. Get our sentiment up as high as possible. We want, as many people are talking about it as possible in our target markets, and we want that as positive as possible if you're doing that great job. but obviously what we've found is. We need now more nuance because those narratives, those harmful narratives, those toxic racist narratives that, that sentiment being shared online has a massive potential to impact your overall sentiment of brand reputation, which ultimately hits your bottom line. So those systems that were born out of the way that PR was previously run. meant they were relatively simplistic, right? Tell me how much I'm being spoken about and the relative sentiment, but they didn't tell you why they don't. They don't do the analysis, although the qualitative analysis, that's the reason why those volumes numers are the way they are. That's simply the way it is. and that was hard to do actually really hard to do up until really just a year or so ago. When we develop the technology to allow that kind of analysis. So now what we've done is in comparison to those other guys who give you the overall numbers, we go behind the numbers we want to know, why is that volume the way it is? Why is that sentiment trending this way? why is it happening like that? That's effectively, what we're automating is the intelligence and the analysis, which is only really available now with recent advances in AI.

Jim James: 3:27

Ant so really quite path breaking stuff. And I can see, we used to, as agencies be asked for volume and actually a lot of negative sentiment wins on that because there's a lot of volume of it. So how you going behind that to understand where that's coming from, the flight, the drivers, are you looking at who is putting out that content or the context of the context? Because that is what we'd need to do from a marketing point of view is to understand the source. Isn't it. And the motivation of that content.

Ant Cousins: 3:56

Absolutely two things. So we have a topic called in the entry topic, clustering technology, so effectively in the same way that those other media monitoring sort systems will go out and find all the mentions of you. here's the tweets, there's the Facebook posts. It's the articles. And they'll give you that as a large Corpus and the best you'll get probably is like a word cloud. here's January, what is being said, but not really, but the reasoning why, what we do is the exact same thing in terms of gathering the data, gathering all dimensions. But then we run that through our topic clustering. and what we spit out is the narratives. So we find the kind of common threads we find the narratives. and identify those narratives across all the different mentions across all the different platforms so that you, as a human, aren't looking through a thousand tweets, trying to work out why is this all so negative? What is the threat here? We automate that process and identify the narratives here. So you've got a human amount of data you can look at and interact with. So that's what we do differently.

Jim James: 4:50

Ant when you say you go across all the platforms you talk about, obviously social platforms like Facebook, but I also able to read gated media, for example, mainstream newspapers at the times, or the economist, or are you only able to see what's available in the public domain?

Ant Cousins: 5:05

It's a, it's all up for grabs. I would say right now the need from our clients is around the broader, easy to access stuff. It's the B to C the broader public opinion type information. So we are focusing largely on social media and that content, but we have one of the parts of the value of the company is we have arrangements with data aggregators. So if a company is interested in a particular data source, we can probably get it. so we tend to work most on social media and finding the online narratives of public opinion and understanding public opinion. but we can also get to that, that data.

Jim James: 5:35

That seems important because I did a research last week on the Dominic Cummings hearings and on social mention tracked the mainstream media. Had given a much more neutral view than the social media sentiment. So they'd really diverged quite significantly between what people who are not paid to write. And those people who are paid to please their publishers and their advertisers. So in terms of how a company would use it, then, because with fact matter, someone's giving you a narrative that, that. Like to have about their own company or are they doing scenario planning for crises? are, what are some of the use cases?

Ant Cousins: 6:15

Yeah, so it's generally around threats opportunities. so you know, we've done the narrative analysis. We've told you here's the top 10 narratives about you by any metric by most positive. Also by most negative, by most racists by most toxic by popularity. So we give you the ability to understand those narratives in terms of whatever's important to you. So yes, you can find the most negative. You can also, through those narratives, identify the most negative influences. so that's the second order effect is we've identified the narrative but who is it? That's actually promoting this and are those significant people that we can engage with? Can we. Can we, can we switch them? Can we educate them or can we deploy our own people? Or are the people that we think disagree with that particular stance that individuals taking and counter that narrative. we need to get behind the narratives to tell people who the influences are. We've got to give them that, but also it's really what matters to them. If they are most interesting and avoiding risks and crises, then sure you use art narrative identification, and then look for the riskiest, most threatening, most racist, most toxic content out there.

Jim James: 7:15

Ant then this content. Report coming out in real time, or is it historical? Because some of the information that comes out now trends so quickly that a company has to really deal with it. If you're in Bitcoin, for example the narratives of Elon Musk and affect the share price of Bitcoin and the price of Bitcoin within minutes. So just tell us about the timeliness. Yeah.

Ant Cousins: 7:40

So we are very much more on the insights and the longer term kind of planning and the longer-term blacks were not real time. So if, yeah, if you almost tweaks about Bitcoin, we're not going to pick that up and warn you about it within, within seconds. we are within the date. so we are our intelligence. Because of, I guess also the amount of intelligence running through all of that content. If you can imagine you've gotten, million tweets to analyze and work out what the trends are. That's actually pretty hardcore computing. so we tend to run on 24 hour cycles. So we, we updated data. We can look historically. So if you wanted to look at analyze certain period, we can do that. but generally we start the topic running. If you want some to start data, we'll give you the historic data. And then going forward every day we refresh is the narrative stuff bubbling away is the narratives that have the potential to trend. but we're not down into the kind of individual tweets.

Jim James: 8:25

Now you mentioned a couple of things in there. One is about the volume. Obviously English language is where there's a lot of volume, but what about other languages? Because many companies now have to deal across geographies. Yeah, what's the, what's their capability with other languages, for fact matter.

Ant Cousins: 8:42

Got it. So right now we do English and that was we had to solve the technical challenges in a certain language. We've done that, but actually we've we've actually won an award from innovate UK for developing multi-lingual models to deal with these same problems. and we're tackling the languages. we'll actually be in about eight weeks time. So middle of the summer. we're starting with the highest volume of kind of speakers. So we're going with Spanish first and then one through the languages. So at the moment, English only, but very shortly be Spanish and others. And if a company then wants to use fact matter to track how their own sort of social media presence is trending and being perceived.

Jim James: 9:16

That's great. But obviously we don't exist in a vacuum. What about. Our relative position to other players or the competitors, or if you're a politician, I guess you want to track, if you're Nicholas sturgeon, do you want to track how Alex Samond is being seen or Dominic Cummings versus Boris Johnson there's fact matter, helping both with the absolute and the relative narratives.

Ant Cousins: 9:39

Yeah, absolutely. if you want to track a brand or what we call it in our pallet as a topic. if you're Pfizer and you wanna attract, whatever, what are the conversations going on around Pfizer? Probably a whole bunch of, misinformation, disinformation and that space. then you can track Pfizer as a topic by Pfizer interest in what people are saying about ashes and they go well, Medina. you could track those topics as well, and it could also be individuals. for us, any topic is very simple kind of logic to, to grab them, get them mentions relevant to that topic. The one thing we do differently to other people in that space is we don't just track kind of basic sentiment like objectively is this piece of content, objectively positive, negative, neutral. We have stance. So we layer on top of that. How is this person talking about the overall topic? So in the case of Alex salmon, for a tweet, there could be negative in sentiment. It might well be positive in favor of the topic of Alexandre. we actually have two different lenses then I find that's much more interesting, especially when you're doing influencer analysis. so stance is something that we would also do that.

Jim James: 10:32

Yeah, because people's syntax mean could impact as well. If a, an engine just picks up something that's written with some irony, actually it could be interpreted as negative, but it could have just been humor. So it's got to be a little bit more. A bit more clever than that.

Ant Cousins: 10:49

Exactly. And we've got intelligence. For example, we did some analysis that was a company is protein powders and they were in the kind of that space. And you're trying to create healthy living and healthy lifestyles. They want to know what people are saying about their products. And we tracked, I think the The specific comment I think was we tracked it as a positive, towards the sorry, as a negative towards the product, even though the keywords have just looked at, basically it was saying it was sweet, which ordinarily think it's sweet. That's good. When you look at the content itself, it was sickly sweet. So we did identify that sweet wasn't positive sickly, sweet together is actually negative and sentiment towards the product.

Jim James: 11:22

Yes. And just tell us then about the house stance works because a sentence can be quite sophisticated English, not only have the noun, but the adjective attached to it. which could modify. The noun quite significantly. Are you taking that into account within Factmata?

Ant Cousins: 11:39

Yeah. So that, that's what we've been working on for the last three or more years, the models have been trained on that level of complexity. so in the case of the protein powder where people were saying, this was sickly sweet, we understand that's. Yeah, negative and sentiment, even though it contains generally a positive word, which is sweet. so if you take that and extrapolate that across the rest of the conversations and things are tracking, then yes, we will pick up kind of irony. We'll pick up. those modifiers in context is it is a hundred percent perfectly, of course not. there's no such thing, but it's it's reliable enough to give you reliable overall data and overall. statistics.

Jim James: 12:13

How is Factmata, helping someone, not just with this is the narrative, but here's some sort of prescriptions around what you could do about that. Or do you leave that at the door?

Ant Cousins: 12:24

Obviously the first step is identifying an narrative. If you've got to know what's being said, you've got to understand who is behind it and their motivations and intentions. and that's obviously you can do that within our product. You can find all the information. We also arm you with the keywords and the associated phrases that. Go along with it narratively, especially if you're trying to track hashtags or trying to find where else those conversations are happening. but then yeah, your kind of the creation of the content, that's almost next on our roadmap is a case. We know the narrative that you want it to be. We know this. Piece of toxicity or this piece of ministry or whatever. we know the narrative you're trying to beat. Is there a narrative already about your brand, which counters that narrative in terms of its kind of volume or sentiment or its influences? Can we identify that for you? So the least then you can, support and exponentially grow that narrative to counter it. That's the next step in our intelligence roadmap is identifying for you that the automatically the narratives and influences that most naturally counter and narrative, you want it to feed.

Jim James: 13:20

That sounds really useful. Now I do have a question for you though, about influencers because a fair amount of content, for example, around the elections in America, the view is that's been generated by foreign agents and by bots. And certainly when we look at, for example, advertising over a third of all click throughs are by automated bots. Now there's a huge amount of foreign agents of one kind or another how you're helping a company to. Safeguard themselves against that, or even to, track down the foreign agents and then take action against that.

Ant Cousins: 13:53

I think the very first thing is understanding the narratives. And we, you've got to do that for a state where it's not irrelevant where it's coming from. You do need to understand that because your actions, your countermeasures might be subtly different. I would say that the first thing you've got to is understand that the most risky narratives, once the narratives. it's not irrelevant where it started. because if the narrative has already picked up the fact is there's already going to be humans involved. because that's the ask the aim of the bot farms, Is to get a human to see that stuff and then immediately share it. so I think by the time the currently by the time brands are seeing those narratives, it's already too late because the bots have done their job. It's already being shared by humans. And that's the problem. What we do is identify those narratives much, much earlier in the cycle, down to clusters of even maybe three or four opinions when it could still be largely bot driven. So what we're trying to do is get that early warning. for those companies just much earlier in the cycle so that they can head it off and either start, take down measures or at least identify what the, or the potential potential trend or potential trajectory of that narrative is going to be. So they can head it off.

Jim James: 14:53

I can really see them the value of the Ant. So with much more depth and love your idea of the stance, because the social mention misses the context. Too much, doesn't it? And it's much more sophisticated than just, you got mentioned.

Ant Cousins: 15:05

Exactly. Context is king. and getting that context as early as possible way before a human would spot that trend. we will spot that narrative and flag it to you. This is potentially really toxic, really racist, really harmful, whatever we're going to do that much earlier than a human could possibly find in that kind of mess of mentions.

Jim James: 15:21

Yeah, or conceivably it could be positive things that there are trends happening around coming out of lockdown, for example, that people could capitalize on. Oh, absolutely.

Ant Cousins: 15:29

The opposite. The opposite absolutely applies. It's just a tool set. So we started off in the finding harm and protecting people from harm, but actually the same source that applies for finding a positive opportunities for brands too.

Jim James: 15:39

and in terms of tool sets being available for the average entrepreneur owner operator, What sort of fees are we looking at? is it accessible to SMEs or to startups?

Ant Cousins: 15:50

So we're aiming for a broad kind of market and it's not expensive at the moment. Our pricing is relatively transparent and simple. It's a thousand dollars for a topic. so if you want to track your own brand or a competitor's brand, it's just that times a thousand. and that's on a monthly basis. We'll update that every single day and the intelligence insights comes with that. but we're actually also opening up a simpler version of the product. If similar to. The Experian credit score approach. Which is, if you just want to have a broad idea, I might endanger, am I under threat or not? a simplified version of that school I think, should be available for all companies in the same way that every company has a firewall, To protect them from from harm, they need that fake news firewall. Every company needs to understand just, is there a really harmful narrative about my company? I need to investigate. so just given that sense of early warning, that threat detection that I think is a much simpler, cheaper product, which we rolling out shortly.

Jim James: 16:42

That sounds really useful because I think everyone now have a big or small can be held to ransom by negative even just negative agents or consumers that want to get their own back. Ant if people want to find out more about a Factmata sorry, I should say if you want to find out more about Factmata, how can they do that?

Ant Cousins: 17:02

easiest thing is go to the website and use the contact us page there. Or they can go onto LinkedIn and find a stair and drop me a message. I think, the PR industry in general is relatively closely connected. sorry if I'm not a first degree and probably at least a second from whoever wants to get in touch.

Jim James: 17:18

Great. Ant Cousins, thanks so much for joining us today as CEO of FactMataon the unnoticed show on how we can track how we are getting noticed and whether it's good or not.

Ant Cousins: 17:29

Thank you.