Welcome to the Publisher Newsletter by Media Voices: profiling the people and products powering publishing.

How AI is reshaping four B2B data and information businesses

Leaders at RELX, Lions Intelligence, William Reed and Global Water Intelligence share how they’re using AI to enhance their B2B offerings, as well as balancing opportunities with the need to protect IP and content value.

On a panel at our recent Definitive AI Forum – moderated by Natasha Christie-Miller – were Dean Curtis, CEO at LexisNexis Risk Solutions, Nicola Tillin, EVP of Lions Intelligence, Christopher Gasson, founder at Global Water Intelligence, and John Barnes, Chief Digital Officer at William Reed.

The panel covered a wide range of topics around AI and B2B, including licensing content and protecting IP. The full session is available to listen to on this week’s episode of The Publisher Podcast (below, or wherever you find podcasts). 

However, one of the more striking points of the discussion was the vast range of use cases for AI internally in these businesses. Here are some of the initiatives and tools our B2B leaders highlighted as making a real difference to their teams.

Marketing and sales

One of William Reed’s top internal AI initiatives is a tool called MarBot – a marketing robot which helps its marketing teams produce targeted emails more quickly. “It’s resulted in 50% less generation time, better click-through rates, and better open rates,” Barnes said.

Lions Intelligence have also seen success internally with a similar tool. Tillin explained that they have a brand tone-of-voice agent, which their sales teams use to keep a consistent voice across their intelligence and insights businesses. “Our sales teams are selling one of five products at any point,” she noted. “So it allows them to write their emails and put their presentations in the style and tone of the brand that they’re leading with.”

It’s not just marketing teams who are benefitting from AI. Barnes’ final highlight from William Reed’s internal AI initiatives came from some training sessions their sales teams had undertaken. “Some of the sales people have actually built – in Copilot – a sales ranking tool,” he explained. “It can look at a sales call… and rate it against seven criteria they’ve been trained on.” 

LexisNexis Risk Solutions are also using AI to enhance their sales strategies. They use platforms like Gong, a specialised AI OS for revenue teams, which automatically captures and analyses customer interactions.

Reducing some of this capture work has saved the sales teams an average of five hours a week per user. “That means they aren’t [needing] to put anything in the CRM system. It means that feedback about products is going straight to products,” Curtis explained. 

He also highlighted that where Meta are using similar tools in Europe, their AI forecast is 11% more accurate than their sales forecast based on AI analysis of the language and sentiment used in customer meetings.

Data and analysis

Tillins has seen the transformative potential of AI in the work WARC’s media team does. They have 100 years’ worth of advertising spend data from over 100 countries around the world, all in different formats. “It used to take the media team one to two hours to run some relatively basic [data],” she said. “Our [AI-powered] media analyst tool reduced that to about 20 or 30 seconds, which has been really transformative for what else they can do with the two hours.”

LexisNexis Risk Solutions also use AI in parts of their content aggregation and creation, like running analytical models. “How do we put our analysts in place of most potential, doing what they love, and where they add value?” Curtis said, emphasising that they want to free humans up from churning spreadsheets.

William Reed have also built a tool for their content platform which is optimised for EEAT – Google’s algorithm for Experience, Expertise, Authority and Trustworthiness. Barnes explained that this scans articles before publishing to ensure they hit those markers, which in turn helps maintain search traffic when the pieces go live.

For Gasson at Global Water Intelligence, AI has transformed their ability to package and sell tracking data. He explained that fifteen years ago, they had people employed to track projects taken from the Capital Improvement Programmes (CIPs) of water utility companies across the US. “It was a complete disaster,” Gasson shared. “It cost far too much to pay people to download these things, search for the information, pull it out and database it so we could sell it.”

That all changed last year when their US correspondent started to create lists of projects by using Python to scrape the CIPs. “By April, he had a huge spreadsheet, which he converted into a database,” said Gasson. “We then started to add the council decisions from council minutes, so it became more of a tracking project – you could see the project and how far along it was.”

The latest version of ChatGPT has now enabled the Global Water Intelligence team to listen in to live council meetings, and get results and decisions as they happen, although accuracy still has some room to improve. “We then launched this as a product last month, the whole thing had cost less than £100 grand,” Gasson explained. “Selling [subscriptions] for these things takes a bit of time… but we got our first sub in at £40 grand.” 

Gasson pointed out that the most potential here comes from being able to expand and grow the product continually. One potential client wanted to get an idea of the value of water contracts and individual pumps, valves and parts going into them. Their US correspondent noted that sometimes bills and materials are published alongside the tenders. 

“Within 48 hours, our sales guy had actually scraped all of these bills and materials relevant to the kinds of projects we’re doing, and built a model for predicting the cost of each equipment line,” Gasson said. “The sales guy is actually building the product for you, vibe coding it almost overnight.”

For other publishers looking to use AI for analysis and data collection, Gasson advised upskilling staff on both the sales and the content side. Vibe coding – where language prompts are used to build applications rather than a developer – has a lot of potential, but does rely on an understanding of what can be done with Python and other code, and how a database works.

Subscriptions optimisation

AI is also being used to enhance subscriptions and renewals. At Lions Intelligence, the teams have a lot of information about their customers spread across different systems and in different formats. Tillins explained that their account development team have in the region of 150 clients to look after each, and were finding it difficult to work out which were most at risk of churn.

“Our data science team took all of the different data points that they would want to go and look at, and created a model,” she highlighted. “So you can say, tell me anybody expiring in the next six months who are paying slightly more than the average for their cohort, or where people haven’t logged in, or the seats aren’t filled.”

This gives the accounts teams a shorter list to look at and more focused on the people that have a higher likelihood of churning.

Customer service improvements

RELX are using AI to help smooth employee challenges internally. Curtis explained that they have 37,000 staff, and three years ago, they used to process approximately 450,000 queries a year on booking annual leave. “That’s something that a human doesn’t need to do,” he pointed out.

Now, they have an AI customer services assistant implemented in Teams, which acts as a first line triage for common queries. Leave can be booked directly through it, as well as getting additional direction on processing it. “It has been a game changer for eradicating non value-added work for our HR team,” Curtis emphasised.

Global Water Intelligence have also used sales bot tools to help onboard new and potential customers. “They do better at guiding people around the site and doing demonstrations than our staff,” Gasson noted.

Advice for other publishers

To conclude the session, each panellist was asked what advice they would give to publishers who are still looking to take their first steps with AI. A common theme from all four was that the most transformational changes come internally, rather than involving external tools or vendors.

Global Water Intelligence’s Gasson was particularly emphatic that the age of monolithic, multi-year IT projects is over. “You’ve got to be doing it on the fly,” he said, pointing out that internally, people are able to take advice and vibe code from various LLMs about how to solve problems. “Anything longer than two weeks, things will be so different with [AI tools] that by the time you get down to doing it, it’s all wasted.”

LexisNexis Risk Solutions’ Curtis said that machines are getting better at being machines, and we humans need to get better at being human. “But humans with machines will beat humans without machines… so we need to create the space for our teams to play and get used to it in their day-to-day work and life,” he emphasised. “We can’t do what we’ve always done… because it’s here. [We need to] create the space and curiosity to lean into that.”

Barnes at William Reed said that vectorisation of data and content – converting and structuring it in a way AI and machine learning tools can process – would make a real difference to publishers starting out with AI. “You can look at [your content] from all sorts of different angles, and in doing that, you see the real value, and sometimes the hidden value of the content you’ve got,” he explained.

Tillin added a note of caution to the ‘move fast and break things’ approach. “We had a lot of ideas when we started using [AI], and there were lots of ideas that were really rubbish,” she pointed out. “[I advise] spending a bit more time listening and thinking before actually acting.”

Listen to the full session in this week’s episode of The Publisher Podcast, available wherever you find podcasts.

More good reads: