NewsAI and communication measurement

What changes, what holds, and where GEO fits

In April, PR Agency One was invited to the European Commission in Brussels to speak about how artificial intelligence is reshaping communication measurement and evaluation.

Our managing director, James Crawford, led a session with senior public sector communicators and analysts, focused on a question that is becoming increasingly urgent for organisations across sectors:

What actually changes when brand discovery and reputation move into AI, LLMs and the like, and what does that mean when it comes to evaluation?

The short answer is that while the environment is shifting quickly, the fundamentals of good measurement remain intact.

The fundamentals still matter

The AMEC framework continues to provide the backbone for credible evaluation, and AI introduces new data points to inform it.

• Objectives – the organisational and communications goals designed to move the needle for an organisation and its profile
• Outputs – the content, coverage and signals that organisations create or influence
  
  
• Out-takes – what audiences actually see and take away – now including AI generated answers and summaries
  
  
• Outcomes – shifts in awareness, trust, consideration or behaviour
  
  
• Impact – contribution to organisational objectives

The channel is changing; the evaluation logic is not.

What AI actually changes

Where AI does create a step change is in how information is served and consumed.

Discovery is moving into AI generated answers, summaries and assistants. That creates two immediate implications for communicators:

• The journey is no longer dependent solely on clicks; users may form an understanding without ever visiting a website
  
  
• The representation of a brand or institution is increasingly viewed via AI LLMs and summaries, drawing on a broad information environment

This means measurement must expand beyond traditional traffic and referral signals and begin to consider what audiences are actually being shown – and how they interact with it.

This also introduces new risks, as inaccurate, outdated or poorly contextualised information can travel further and faster than before.

This is why AI measurement must be treated as a standalone metric or score and must be grounded in method, evidence and governance. Importantly, this measurement needs to sit with PR and communication teams.

Where GEO fits into measurement

This is where Generative Engine Optimisation (GEO) comes into play.

GEO is often discussed as a way of improving visibility in AI. That is part of the picture, but it is not the full story.

From a measurement perspective, GEO is best understood as part of a wider system. It sits across three connected areas:

• Brand reputation – the wider public information environment that AI systems draw from
  
  
• Search and content readiness – whether key information is discoverable, structured and usable
  
  
• AI outputs – what users actually see in answer-based environments

These should be assessed together and triangulated, rather than treated as a linear sequence.

If upstream reputation is weak or inconsistent, AI outputs are more likely to reflect those weaknesses. If important content cannot be found or understood, it is unlikely to be represented accurately downstream.

For a deeper dive into how this works in practice, see our dedicated page on Generative Engine Optimisation and our GEO playbook, which sets out a practical approach to measurement and optimisation.

Separating signal from noise

One of the key themes discussed in Brussels was the need to separate credible measurement from hype.

The market is currently full of “AI visibility scores” and dashboards that promise simple answers. The problem is that simplicity often comes at the expense of validity.

AI outputs vary by prompt, user, platform and time. There is no single, stable metric that captures total visibility across AI systems.

Credible approaches treat AI observation as directional evidence:

• Measuring presence, framing, citations and accuracy across defined queries
  
  
• Recording prompts, platforms and dates
  
  
• Repeating tests over time
  
  
• Being explicit about variation and limitations

Forget about trying to create a new scoreboard; it is about building a transparent evidence base.

Why governance matters more than ever

For regulated or public sector organisations in particular, governance is critical. AI-supported measurement raises questions around:

• Data quality and bias
  
  
• Transparency of methods
  
  
• Accountability for automated outputs
  
  
• The risk of misrepresenting under-represented voices

AMEC’s guidance is clear. AI is an enabling technology, not a measurement strategy in its own right. Human oversight remains essential.

Practitioners are still responsible for how data is gathered, interpreted and presented. In short, AI should strengthen credibility, not introduce black boxes.

What good looks like now

For organisations looking to build capability, the priority is not to chase perfect metrics but to establish a disciplined approach.

That includes:

• Defining a clear set of priority queries and topics
  
  
• Auditing authoritative content and information sources
  
  
• Monitoring AI outputs for accuracy and framing
  
  
• Linking AI visibility to outcomes through analytics, research and brand tracking

Measurement remains a system, not a single number.

What comes next

The session at the European Commission was a useful opportunity to test and refine thinking with an experienced audience.

It also served as a preview of some of our thinking and the broader body of work that PR Agency One will be launching with the rest of the agency special interest group at the AMEC Summit, focused on where credible GEO and AI supported measurement is heading.

As AI continues to reshape how information is discovered and understood, the challenge for communicators is to apply measurement more rigorously than ever.

Precision over promotion.

More to come soon…