In this week’s newsletter:
The stories that investor relations professionals need to read this week:
- FTI on narrative slack: our research among investors gives us a clear idea of how key audiences are ready to punish companies when ambition and facts are out of alignment
- Do we really want to replace human bias with bots in the boardroom? The Times asks: is it good for shareholders?
- The 29th Deutscher Investor Relations Verband (DIRK) conference took place in Frankfurt last week. FTI Consulting’s Carolin Amann attended
- Bloomberg’s Matt Levine on tricking the AI investors
- Activist investors’ campaigns hit global record, with more on the way warns Barron’s
- And finally: CEOs beware – here comes the algo tone police. The Financial Times ponders the appropriate amount of arousal for an earnings call
This week’s news
FTI Consulting on narrative slack
Our latest research on narrative slack has detailed some key findings on investor behaviour if corporate claims outpace delivery. Looking specifically at investors, it is clear that the response is scepticism, with 58% losing confidence in management, 31% increasing scrutiny of future disclosures and results, and 29% pulling back from adding to their position. The pattern hardens with repetition: 74% of investors say they would likely reduce exposure if claims repeatedly run ahead of delivery, treating recurring gaps as evidence of an execution problem. For IR teams, the message is that guidance, disclosure and delivery now need to move in lockstep, or risk a de-rating that takes many quarters of consistent results to reverse.
Do we really want to replace human bias with bots in the boardroom?
AI is moving into the boardroom, with vendors offering AI tools to help non-executive directors digest board packs, flag regulatory risks and draft probing questions for management. The trend reflects a genuine burden: board packs now average 226 pages, up 30% since 2019, while directors typically spend only four hours reading them. But The Times raises pointed questions for governance. Unlike a consultancy firm, an AI tool cannot be sued for negligent advice, leaving an accountability gap if boards lean on it for judgment calls. There is also unease about vendors marketing AI tools as a way to strip out “human bias”, when diversity of perspective & challenge to management groupthink is precisely what non-executives are meant to bring. The Times suggests a blunter fix: halving the length of board packs.
The 29th DIRK conference
The 29th Deutscher Investor Relations Verband (DIRK) conference took place in Frankfurt last week, bringing together around 400 Investor Relations professionals and capital markets experts from across the DACH region for two days of discussion, Including FTI’s very own Carolin Amann. Discussions focused on this year’s theme: “Tweet or Trust – IR and the New Complexity.” With Denmark as guest country, the conference offered perspectives on one of Northern Europe’s most dynamic capital markets. In his keynote, Lars Ohnemus from CCG highlighted the Nordic corporate governance model, reminding the audience that trust is a competitive advantage built on transparency, strong institutions and long-term thinking. On day two, DHL Group’s CFO Melanie Kreis stressed the strategic role of Investor Relations as a trusted sparring partner for executive boards. AI was also high on the agenda, with participants exploring its impact on everyday IR work, KPIs and reporting. The conference closed with a Gala Dinner celebrating the latest CIRO graduates and outstanding IR teams.
Bloomberg on tricking the AI investors
Matt Levine at Bloomberg has spotted a piece of academic research that sits somewhere between genuinely alarming and darkly comic. Researchers at the University of Liechtenstein designed 10 large language model (“LLM”)- based trading systems to forecast share prices using sentiment scanning across a basket of stocks. Whilst all were in the green after 14 months, each one was tricked by hardly perceptible tweaks to financial news headlines. The manipulation technique, which involves swapping standard letters for visually identical characters from other alphabets — a Latin “A” replaced by a Cyrillic “А”, for instance — rendering the change invisible to a human reader but enough to send an automated trading system to entirely the wrong stock. Levine posits that the most obvious user of this method in this scenario is not some shadowy external actor but the company itself. If algorithmic systems increasingly drive trading decisions based on news sentiment, the incentive for a company to quietly misdirect bad headlines rather than suppress them becomes more likely. For IROs, the research raises a question: as LLM-driven trading grows, the audience for corporate disclosures is no longer purely human, and the standards of clarity and integrity that govern communications may need to account for a readership that can be fooled by a single misplaced character.
Activist investors’ campaigns hit global record, with more on the way
Activist investor campaigns hit a global record in the first half of 2025, and experts are warning that the pace shows no sign of slowing, notes Barron’s. According to data from Lazard, activists launched 184 new campaigns worldwide in the first two quarters of the year, a 20% increase on the prior year and nearly 40% above the five-year average for that period. The surge reflects several converging forces: a more permissive environment for mergers and acquisitions, a sharp rise in activity across Asia, growing interest in targeting financial institutions, and an emerging wave of campaigns centred on how companies are deploying artificial intelligence. Elliott Investment Management led the charge, launching 13 new campaigns in the first half of the year, up 86% on the same period in 2024. As companies face record campaign volume, activist preparation and proactive shareholder engagement are no longer optional extras but essential components of any credible investor relations programme.
And finally: the algo tone police
FT’s Alphaville suggests executives have another thing to worry about in presentations beyond vocal ticks and choice of words. Research from Speech Craft Analytics and the Journal of Portfolio Management has shown that through modelling “paralinguistic acoustic features”, a range of characteristics like confidence, nervousness and even arousal can be scored. So next time you’re listening to a CEO issue results that fall short of expectations, consider that if they are speaking with apparent conviction, the algorithm may reward them.