Writing annual reports for the AI age

For years, annual reports have been written with two key audiences in mind - investors and stakeholders. Now there’s a third: AI engines.

Analysts, fund managers and institutional investors are increasingly using AI tools to make sense of the huge volume of reports crossing their desks. Rather than reading them cover to cover, these tools extract insights, identify trends and flag developments worth a closer look.

It’s changing how your story gets told - and who decides what’s worth telling.

Get it right, and AI becomes an amplifier, ensuring your key messages rise to the surface. Get it wrong, and the most important points in your report might never make it past the first digital filter.

Understanding the new reality

AI-driven analysis isn’t on the horizon, it’s already here. It’s being used to:

• Summarise lengthy reports and extract critical data
• Compare performance across companies and sectors
• Flag risks, opportunities and outliers
• Build initial assessments before human review

That means your report is increasingly being read by machines before it’s read by people. AI doesn’t replace human judgement, but it heavily influences what gets noticed and what doesn’t.

If your key messages aren’t easy for AI to identify, there’s a real risk that your human audience will have their perceptions and actions informed by messages other than those you intended.

Rethinking how you write and structure reports

To communicate effectively in this new environment, annual reports need to work on two levels - engaging humans while being easily interpreted by machines. That AI-first thinking requires some subtle shifts in how reports are written:

1. Structure clearly and logically
   AI relies on structure to understand what matters most. Use consistent, descriptive headings that say what’s actually in the section  “Strategic growth initiatives” will outperform “Charting new horizons” every time. Keep your hierarchy clean, with executive summaries upfront and key points called out clearly in lists or highlight boxes.
2. Prioritise clarity over cleverness
   Beautifully crafted language still matters, but when it comes to AI, clarity wins. Remember your AI doesn’t really get tone or irony so use plain, straight-up language for your core messages. Define acronyms, explain industry terms, and state your key points explicitly rather than hinting at them.
3. Give data context
   AI can crunch numbers but struggles with nuance. Always accompany tables and charts with clear labelling and commentary. Tell the story behind the numbers: what changed, why it matters, and what’s next.
4. Be consistent
   Use consistent formatting for key metrics and repeated terms. Inconsistency confuses both human readers and algorithms.

What AI is actually looking for?

I asked both Claude and ChatGPT what they were looking for when reviewing an annual report. Understanding how AI analyses reports can help you decide what to emphasise. They both indicated they looked for:

• Financial performance - revenue, profit, cash flow, ratios, and year-on-year changes.
• Strategy and direction - management commentary, market positioning and progress against stated priorities.
• Outlook and risk - guidance statements, forward-looking commentary and quantified projections.
• Operational highlights — segment results, milestones, and regional performance.

AI gives more weight to what appears early, what’s explained in detail, and what’s repeated consistently across the document. So placement and emphasis now matter more than ever.

The opportunity ahead

This shift doesn’t make annual reporting any less human - it just means you need to write for both audiences. The most effective reports of the future will:

• Tell a clear, compelling story that connects emotionally with human readers. That’s always been the case.
• Use structure, clarity and consistency to help AI (and humans) understand and elevate your key messages

By designing your reports with both in mind, you ensure your story is heard whether it’s being read by an analyst, an investor, or an algorithm scanning for insights. On the plus side, we’re often lamented about the amount of work that goes into an annual report and how few people read it. At least we now we know someone is reading it cover to cover!