Ever since I started in PR several hundred years ago, the premise of the tier one and two media approach has always held strong — albeit with the odd beef about how these were defined.

The halo effect of the masthead would often dominate tier one (I’d have no mortgage if I had £1,000 for every RFP request to be in the FT), and, if you were lucky, key business, tech, and IT titles. The seemingly less glamourous titles sitting pretty (or otherwise) in tier two, were made up of verticals and niche trade.

Understanding influence

This approach has merit, but could often overlook the model of influence. Writers at the nationals get information and news into their beat from trade media — these are sources of insight and opinion that inform the bigger outlets. But, this isn’t always appreciated.

The traditional tiering model also ignores the concentration of readers within that title that are potential buyers. For example, The Times is a peerless publishing institution, but a fraction of its readers are decision makers when it comes to buying semiconductor equipment. The EETimes on the other hand — while less shiny — has a concentration of readers along this purchase funnel far higher than any national.

As such, a combination of both tier one and tier two has established itself as the standard. And yes, I realise I am preaching to the choir on this.

Changing news dynamics

Large language models (LLMs) such as ChatGPT have changed the world of search. After amassing 100 million within two months of launch (by comparison, it took Instagram almost three years), the platform now counts 800 million active users, both at a consumer (free) model and at paid enterprise level.

Arguably, the disruption goes beyond just search — LLMs have taken it closer to discovery, through the premise of asking questions, rather than trying to think of related keywords and scanning through links.

In asking a question, the door is opened to nuance and consideration, shared back as prose and explanation, with sources (and yes, with bias and hallucination but we’re getting better at analysing information). Google was, in this regard, merely ‘find the thing which matches these words’.

This initial limitation is undoubtedly a big factor in its inclusion of AI Overviews in results.

Indeed, I’ve found myself changing how I search. I type prose into Google as I do on AI models and I’ve found myself saying thank you to LLMs.

Understanding influence all over again

As comms professionals, the seismic shift in how people are now using LLMs in their search for information must be reflected in how we in turn behave. We need to understand ‘influence’ through a whole new lens — namely those sources which inform the results answering LLM queries.

This is where the need for a ‘tier LLM’ exists — a targeted approach based around the media sources that most directly inform how LLMs understand and present information about a brand, product, or sector. Building visibility within this tier requires understanding the questions buyers are asking, how they’re phrased, and which outlets most frequently appear in AI-generated results.

Ignore tier LLM at your peril

Tools can illuminate this landscape, mapping share of search, source weighting, and content influence.

However, it demands continual refinement to reflect evolving algorithms, competitive shifts, and changing buyer behaviour.

Incorporating tier LLM analysis into every campaign is no longer optional. It’s an essential evolution of media strategy and one that ensures brands remain discoverable, credible, and influential in both human and machine-mediated ecosystems. Without it, media strategy will remain static amid a landscape that is constantly evolving, more so than it ever has and at exceptional pace.

Hoffman has developed its own proprietary solution — GEDI — which is being integrated into client QBRs. It aims to identify the sources involved in answering queries and how we need to react to these shifting sands. More of that here.