x
FIBEP Tokyo 2026: Who Gets Paid, Who Gets Seen, and What Media Intelligence Is Worth Now

FIBEP Tokyo 2026: Who Gets Paid, Who Gets Seen, and What Media Intelligence Is Worth Now

17th November 2026 - By Michelle Harold, VP of Global Partnerships, eMedia Monitor

From 3 to 5 November, the media intelligence industry gathered at the Hilton Shinjuku for the FIBEP World Media Intelligence Congress 2026 in Tokyo. AI was at the centre of almost every session, and FIBEP brought the whole supply chain into one room to discuss it: monitoring companies, publishers, journalists, technology providers and the corporate communicators who buy the work.

The week opened with a moment close to home for eMM. At the welcome reception, Media Track launched its new platform, MT Connect, with a film that named eMedia Monitor as its broadcast partner. It was a proud moment for the team, and a great start to three days of conversations about where our industry is heading.

One question sat underneath almost everything: when AI answers the question before anyone reaches the source, who gets paid, who gets seen, and what is media intelligence worth? Here are the themes that came through most strongly.

1. The value of content is being renegotiated

The biggest shift of the week was who was on stage. Three years ago, the industry talked about publishers. In Tokyo, the Washington Post, Dow Jones, Nine Publishing and MEI Global sat on a media intelligence panel and talked with it.

The problem is shared: AI answers draw on publishers’ content and send almost no traffic back. Supertab CTO Jesse Knight estimated zero-click search has cut traffic by 25%, and Luke Anderson of AWS noted that 69% of US news searches on Google ended without a click in May 2025. The front page has moved.

Publishers are still learning how monitoring companies use their content, especially now AI sits in the workflow. They want to protect its value, but are open to working with the industry if there is transparency. Nine Publishing’s Agostina Garola pointed to Australia as the leader on licensing. The panel’s verdict: the barrier is no longer willingness, it is visibility.

Knight brought something genuinely new: a practical way for publishers to get paid. Really Simple Licensing (RSL), soon to be renamed the Internet Licence Collective, lets publishers set licensing terms that AI companies can be required to agree to before using their content. AI companies pay based on how valuable the content is, not each time they crawl it, so the most-used publishers earn the most. Publishers can also allow some uses, such as AI answers, while refusing others, such as training. As Knight put it, robots.txt is a request, not a lock.

Publishers want to be part of the AI economy, not a free input to it. How licensing works for both sides is now an open question.

2. GEO: are you the one being recommended?

If publishers are asking who gets paid, brands are asking who gets seen. The GEO panel, moderated by Muck Rack’s Natan Edelsburg, reduced the category to one question: not where you rank or how often you are mentioned, but whether AI puts you forward.

Argus Data Insights, GlobalNews Group, Leoprd.io and pressrelations are already selling GEO services, each approaching it from a different angle. Strip those away, though, and the work is the same: building a set of prompts for each client, running them across several AI models, and advising on what to change. No single method has won yet, and that is the real story.

Carlos Alfredo Diaz of GlobalNews Group argued that SEO is part of GEO, not separate from it: AI draws on everything said about an organisation, not just its optimised pages. Celia Harding of Leoprd.io added that paid media does not influence AI answers, and only around 40% of answers come back the same across models.

There is no single global GEO either. In China, Virginia Chen of pressrelations explained, the government regulates both what goes into AI and what comes out. And asked whether AI is worth monitoring at all, the panel’s answer was crisis, not marketing: when things go wrong, what is AI saying about you?

GEO is already being sold, but it is not a switch you flip. It is a consultancy with a data product underneath, and it still needs people.

GEO Panel.

3. Buyers want fewer reports and faster answers

For anyone selling into corporate communications, the most useful session of the week was Reputation at Scale, where communications leaders from Fujitsu, Nissan and ORIX said plainly what they want from the industry. Their summary: they used to receive hundreds of reports nobody read. Now they want to know how they are trending and how they are perceived, quickly.

Their wish list was specific. Simpler platforms, because getting their own IT to integrate an external provider is hard and security teams do not want outside data in their systems. AI, but built in by the supplier rather than bolted on by them. More precise sentiment. Share of voice remains the main measure, and GEO has real strategic value if it can show a positive impact on customers. The bar for new technology is rising too: they want to try new tools, but want more proof before committing.

FIBEP’s member survey showed the same pressure from the supply side. Clients want lower costs, faster delivery and more customisation, and competition is now the top threat members name. AI has moved from existential worry to investment line.

When the buyer and the supplier describe the change in the same words, it has happened. Speed and meaning now matter more than volume.

4. Everyone can buy the same models. Nobody can buy your context.

If AI is now available to everyone, where does the value sit? Luke Anderson of AWS gave the clearest answer of the week. Language model releases went from one in 2022 to 474 by September 2026, and the price keeps falling. Every competitor can buy the same models. What separates the leaders is their context: what an organisation knows that no one else does. He quoted Bloomberg Media: “The future of media AI is archive driven, not artificially invented”.

He also gave the strongest argument of the congress for source-linked monitoring. Research by the EBU and BBC found 45% of AI assistants’ answers about the news had at least one significant issue. His example was a critical story that AI summarised as neutral. Counting mentions would miss that; measuring meaning catches it. For media intelligence, he set out three implications: clients want answers now, by chat, API or AI agent; the work is moving from counting mentions to measuring meaning; and providers should own their context and partner for the layers around it.

For eMM, this landed close to home. Years of recorded and transcribed broadcast across 94 countries is exactly the kind of context no competitor can buy with a model subscription, and clients can trust a finding when they can check it against the original clip.

Human expertise still matters. FIBEP’s survey found most members say clients prefer AI combined with human work, and accuracy is their top AI concern. Even GEO, the product that measures AI, is heavily manual. The models are a commodity. The archive, the context and the people who interpret it are not.

5. Not every market plays by the same rules

After three days of AI answers and GEO, German Saa of KyodoPR offered the counterweight. In Japan, if a story is only online, it is not treated as fact. If it is in traditional media, it is taken seriously.

National newspapers sit at the top of Japan’s hierarchy of trust, and the scale is striking: Yomiuri’s circulation of around 5.25 million is more than twenty times the New York Times’ print run. Yet circulation has halved since 2000, and most people under 35 have never read a newspaper. Access still runs through the press club system, and journalists value a fast response above all else, even though a press release can take six days to clear internal approvals.

The Japan-based journalists on the On the Record panel added that their challenge is now distribution, not writing, as they move into podcasts, video and subscriber sessions. Monitoring built around named titles and text will cover less of the story every year.

The global AI story does not translate directly to every market. Local context still decides what counts as credible.

AWS Front page.

What it means

The FIBEP Congress in Tokyo reflected an industry in continuous evolution, where implementing AI and delivering the insights clients actually want are moving forward together. AI is making the routine work faster and more affordable. The added value clients will pay more for comes from what sits around it: trusted sources, context they cannot get anywhere else, and people who can say what a story means and what to do about it.

For media intelligence providers, that means building AI into the platforms clients already use, delivering fast answers rather than reports in volume, and investing in the archives, data and expertise no model subscription can buy, while watching how the licensing conversation with publishers develops. For communications teams, it means asking more of suppliers: where did an insight come from, can it be checked against the source, and what is AI saying about you before a crisis, not during one? The models are now available to everyone. The advantage belongs to those who know what to do with them.

This piece first appeared on Michelle Harold’s LinkedIn click here.

eMediaMonitor