In this Digital Insights Leadership Series video interview, Mike Vizard talks to SignalAI CEO David Benigson about how AI is being employed to track customer sentiment more accurately.
Mike Vizard: Hey, folks, welcome to the latest edition of the Digital CxO Leadership Insights videos series. I’m your host, Mike Vizard. Today we’re with David Benigson, who is the CEO of Signal AI. They just acquired an outfit called KEP. And we’ll be talking about sentiment analysis and how do you actually understand what it is that your customers are looking for? David, welcome to the show.
David Benigson: Brilliant. Thank you for having me, Michael. Great to meet you.
Mike Vizard: There’s a lot of data out there and there’s a lot of signals you can process to kind of see what people are doing, but how do you do that in a way that people will find receptive versus something that they might perceive to be a little on the creepy side where you’re kind of collecting all this data and basically showing them things that maybe they don’t even know they know yet or think they want before they know they want it?
David Benigson: Yes, the unknown unknowns, as we describe it. Well, no, I think that’s a brilliant question. And you touch upon some of the things for which we founded the business nine years ago, and I think those themes and trends are probably more relevant today even than they were when we first started the company. I think firstly, as you say, there’s never been greater volumes and availability of data than ever before, and yeah, I think businesses and business leaders have never found it harder to use that data to drive better decision making within their organizations.
I think the second big trend that we see is that the world is becoming increasingly more and more complex. The sheer number of issues that organizations are having to grapple with seems to be exploding exponentially. And of course, the last few years have demonstrated that very acutely, with everything from the pandemic to the climate crisis, race and inequality issues, Ukraine and Russia, inflation, the list goes on and on. Businesses are having to navigate through these very, very complex times but don’t have effective radar and data to be able to do that.
And the third trend we saw is the emergence of machine learning and AI that could actually turn those challenges into an opportunity. And so, our thesis has been how can we aggregate the broadest and most diverse dataset that sits outside of an organization and be ambivalent about media type and language and structure, modality, and format, and pull that all into a single platform? How can we then apply machine learning and AI that would then surface insights to help these business leaders make better decisions but more specifically get ahead of those risks, issues, and threats faster and more efficiently for their organization?
Mike Vizard: And how exactly do I do that? Because there is such a massive amount of data to analyze, so are you going to do that for me as an organization and then present me with what and how do I turn that into something that’s actionable?
David Benigson: Exactly. So, that is the sort of capabilities that we’ve been developing over these last nine years. It starts with our ability to essentially collect the world’s information that exists outside of an organization. Today, we aggregate data from about 200 markets in over 100 languages, which we translate all of that foreign language data into English in real time. And that covers everything from traditional media to social media to regulation in about 175 jurisdictions, global coverage of TV and radio, the top 20,000 most popular podcast channels, alternative datasets like product reviews and Glassdoor, and even in some cases our client’s first party primary data.
So, the first thing we do is we aggregate and collect and consume this vast amount of very unstructured information, which we pull into a single platform. The second thing we then do for our clients is we analyze that data and we transform it from unstructured information into structured insight and knowledge. We do that by applying these machine learning technologies, which enable us to extract everything from topics and concepts and themes to the sentiment of a particular document’s saliency. We’re able to look for anomalies – so, slight deviations that might be impactful or relevant for our clients. And we extract all of this knowledge and then we embed it in what we describe as our external intelligence knowledge graph, where we’ve mapped together billions and billions of relationships between companies, people, products, ingredients, diseases, risks, and issues, and we enable our clients to navigate that knowledge graph through our SaaS application or our API products and find these changing trends, these emerging patterns, these emerging risks and issues within that knowledge graph.
And what that is enabling our clients to do is sort of ask and then answer increasingly sophisticated questions of the data. Who are my emerging competitors in a particular market? Are there new regulations or policy emerging in a particular jurisdiction that might affect my business? Are there reputational issues bubbling up in my supply chain? So, these sorts of very strategic questions which historically in normalization might have gone to a consultant for – or use their own intuition to be able to try and answer, they’re now able to use this data to inform and help answer those questions for their organization.
Mike Vizard: And how do I take that data that you’re creating in those feeds and put them into my applications in a way that is accessible to the people that work for me so that they know to go execute something in a process that is tied to that data? Because, to your point, it all seems to be time-dependent, so the closer it is to the point of consumption of the data, the more analysis I can inject into my processes.
David Benigson: Yeah. Well, consumption, you’re absolutely right, is critical. And it’s part of our job and value proposition to make those insights as consumable and frictionless and as intuitive as possible. So, we really have three ways that we deliver that insight to our clients. We have a Web-based application, a suite of SaaS products that enable our clients to use our product to search over billions of data points, create feeds, and then get alerts to relative changes, to create dashboards to visualize insights and patterns and trends within the data, and a discovery engine that enables them to sort of surface those unknown unknowns and find the things that they didn’t even know were out there that could be relevant or impactful for their business.
The second way we deliver the insights is through a suite of API products. And that’s where our clients can access the value in our platform computationally and integrate those insights into their own products, tools, or workflow. So, you can plug us into your CRM system, like Salesforce, and then contextualize your Salesforce with these up-to-the-minute signals that we’re collecting from external intelligence data. You can integrate us into your risk and compliance product, so that as you’re looking at sort of how do you handle new regulations, new policy, we’re bringing in that flow of data to inform you of any relevant regulatory or policy changes that might impact your business.
And then, the third way we go to market is through co-creation and services. Some of our services is about analyzing, curating, and synthesizing that data and delivering it to our clients in a C-suite-ready package, a set of reports that can be consumed at the most senior levels within an organization. But some of our services is about co-creation, where we might collaborate with an organization who wants to build their own products and services on top of our APIs and then sell those products and services or deliver those products and services to their own market or customer base.
So, we’ve partnered with organizations like Deloitte, who have built out a set of tax compliance and regulatory monitoring products that they sell to their largest customers around the world, and together today we service about two-fifths of the Fortune 500. We’ve partnered with EY in audit, and audit is under huge amounts of pressure from financial regulators to come up with more nonfinancial indicators of whether a company is in good health or not. And EY built out a product on top of the Signal platform looking at all these nonfinancial indicators of trust and perception and sentiment to help inform the audit process for their clients, not specifically just looking at financial data.
So, those are the sort of three core ways that we deliver that value to our customers. And we believe it’s important to have those different channels depending on the specific use cases and needs of our specific customers.
Mike Vizard: Do you think we’ll ever get to the point where the data and the analysis that you’re collecting is injected into, say, for example, a supply chain application, using that to judge how much of a particular thing to make for a particular month based on the sentiments that you’re surfacing?
David Benigson: No, it already is. So, we integrate into some of the world’s biggest supply chain platforms. One of our clients is one of the biggest supply chain analytics vendors in the world, and what we’re producing and integrating into that platform is essentially an external data risk score related to each supplier. So, as an organization is determining which of the supplies stored within this database they should or shouldn’t work with, we’re actually providing not the financial data on that organization or the commercial data on that organization, but we’re providing insights around sentiment, relationship to specific ESG topics. Relationship to specific risk topics. And so, as an organization is assessing whether to work with that specific vendor or not, they’re actually getting this real-time enriched insight as to whether there are any issues outside of just structured information that they should be aware of.
Mike Vizard: All right. There’s also always a lot of skepticism when it comes to AI and a lot of that skepticism comes back to people don’t necessarily trust the data that’s being used to create the AI model, so they’re worried about whether the AI model is right, and then the AI model will drift over time as circumstances change. So, how do we have greater confidence in the AI models that we’re going to bet the company on?
David Benigson: Well, I think it comes down to a number of different factors. It’s always that famous line around “junk in, junk out.” And so, the training of those models, as you say, is absolutely critical. We have a team of dedicated analysts who are domain experts across a number of different industries, and they are producing that training to feed into our algorithms and ensure that the quality is there. But very critically and importantly, we then make available to our clients the ability to correct that training and train up the models themselves. So, we actually have a dedicated product which is about our clients training up the models and assessing the quality of those models and ensuring that they can input and influence the quality of those models and then also own, have exclusivity over those trained models. So, when we build a new classification model, when we build or release a new topic, if it’s been trained by one of our clients, they take exclusive ownership of that version of that model. So, that’s one piece which is absolutely critical.
The second piece is about transparency of quality. So, all of the outputs of our data are available, and all of the tags and scores that we apply to the data are available to our clients. And so, they can very transparently see what our algorithms are assessing in terms of sentiment scores or classification scores or entity scores, and they have all the ability both in our APIs as well as in our SaaS products to be able to correct and amend those scores to then feed back into the algorithm to train it over time.
Mike Vizard: Now you’ve acquired KELP. What do they add to the equation?
David Benigson: So, KELP is a fascinating business who sort of add a layer of specialization and insight on top of our platform. KELP builds these industry-specific indexes based on reputation and sitting on top of all of the external news and social and regulatory data that we have access to. So, what they really dive deep into is what drives reputation with industries like health care, pharma, life sciences, aerospace defense, and then they surface through their analytics capabilities actionable insights that executives can use to target highly influential audiences with the right message on the right channel and at the right cadence.
So, it’s a product that really is about benchmarking companies against thousands of highly specific AI-trained concepts and topics from supply chains to sustainability to emerging fields in R&D, and then enabling those companies to take sort of action from the insight through the true measure of their reputation and perception. So, super synergistic with what we’re trying to build. Very, very impressive cofounders. Danny, the one cofounder, ran narrative intelligence at Weber Shandwick, and before that worked on the Obama campaign. And Shann worked for publicists as a chief strategy office and across industry as well. So, both of them bring a real wealth of experience and knowledge and expertise to the business as well, which we’re very excited about.
Mike Vizard: And it does seem there’s a lot more riding on the perception of trust these days for these organizations. And do you think that we’re seeing more C-level execs think about what is the trust measurement of the organization? I guess we used to call it brand, but at the end of the day it seems like it’s more important than ever.
David Benigson: It’s absolutely critical. And I think there’s been a huge paradigm shift in the market. I mean, we recently surveyed 1000 C-Suite executives in America and 85 percent of the respondents said they would make a decision based on reputation over a decision that helped them with the bottom line, which is a fascinating sort of response to that question, which I think if you would have asked that question 5 or 10 years ago, you wouldn’t likely have gotten the same result. And we’re seeing that play out in reality if you look at all of the companies, for example, that have pulled out of delivering services in Russia. That is a reputation-based decision, a values-and-ethics-based decision versus a decision that might help them commercially. So, that’s a really powerful and clear example that organizations are starting to think very, very consciously and explicitly about the impact that their reputation and the trust in their business has.
Now, the challenge these organizations have had is how do you measure that? How do you measure that and how do you correlate it with business performance? If we’re going to take these decisions that help us build that trust and build that reputation, how do we know it’s having the right impact, a positive impact on our business? And historically, that was almost impossible to do and was done in a very sort of intuitive and qualitative fashion, but through the sorts of technologies we’re developing at Signal and through the sort of capabilities that KELP have been developing, we can actually quantify – we measure that now. If you can tie those measures of trust and reputation back to business performance; then you start getting a real understanding of how your investment in that area is paying off, and I think that’s very powerful in today’s context and with the sort of values that consumers are expecting brands to operate with.
Mike Vizard: We see a lot of organizations building their own AI models. At what point does it just make more sense to consume something as a service rather than me going to hire all of these data scientists and building all these applications myself, and maybe just to duplicate what somebody else has already done?
David Benigson: Yeah, look, I think everyone and certainly most of our large clients have gone through the sort of infamous winding road of “Should we build, buy, or partner?” And many of our clients have gone through probably a failed consulting project or two where they’ve spent a lot of money without necessarily delivering the results. I mean, I personally encourage all large organizations to engage with this technology and understand how it can impact their business and the sort of strategic objectives they have because I think it’s without question that advanced analytics, machine learning, AI technologies are going to transform almost every sector – in fact, probably almost every sector – and almost level of an organization.
The decision between, “Should we build ourselves or should we partner or should we buy in a service?” I often think comes down to whether that service is core to the business’ fundamental value proposition and service delivery versus is it about driving better efficiency or delivering the sort of operational strategy that the organization has? And if it’s not core to that core value proposition, then actually if there’s a ready-made tool available, why not partner and bring in best-in-breed? Because often organizations are spending a ton of time trying to spin up something when actually they might not have the skills, they might not have the expertise, and they probably don’t have the culture for it. And so, often it’s, I think, a much better idea to actually figure out how they can find the best solutions in market that are focused on a particular problem and partner with them versus trying to necessarily build it themselves.
Mike Vizard: All right. So, what’s that one thing you see customers doing that just makes you shake your head and go “Why are we doing that?”
David Benigson: You would be surprised by how many organizations are still on-prem. So, the fact that many still struggle with cloud-based services is still a real head-scratcher. But I think it’s also about, as I say, organizations engaging with the technology, having an experimental mindset, being willing to try new things and not just necessarily stick with the same incumbent solutions that they’ve always used. There has to be a way, a new way of doing things because the world has changed dramatically in the last ten years, and just because you’ve used the same sort of method or process for a decade doesn’t mean it’s necessarily going to be fit for purpose in this new world context that we operate in.
Mike Vizard: All right. Hey, David, thanks for being on the show and sharing your knowledge and insights.
David Benigson: Thank you, Michael, for having me. It’s been a real pleasure.
Mike Vizard: All right. And thank you all for watching this latest episode of the Digital CxO Leadership Insights series. You can find it and others on the Digital CxO website. And once again, thank you all for watching.