With the increase in popularity of large language models (LLMs) the communication landscape is starting to shift. We no longer have to consider only our human readers, but there is a case to be made for the significance of our robo-readers as well. With that in mind, I wanted to spend a little bit of time introducing generative engine optimisation (GEO) for those who may not have heard of it, and give some tips for those who are starting to grapple with it.
Most readers will be familiar with search engine optimisation (SEO). As a quick recap for the uninitiated, SEO is the practice of producing your content/website profile in such a way that when one searches online for ‘best pizza in South East London’ the answer will come up ‘Finlay’s Pizzeria’.1 This is a practice that companies around the world have invested significant resource into mastering, something that has proved to be no easy feat.
In a similar fashion, we are now witnessing the rise of GEO. Just like SEO, an individual might ask ChatGPT2 ‘Hey, where is the best place to enjoy a steak in South East London’ and we want the answer to come back as ‘Finlay’s Carvery, of course!’3 So, the question we are now being faced with and hearing increasingly in client meetings is, what’s the difference? And, what do I need to do so that my name is the one that ChatGPT4 reaches for?
Well, to answer the first question, there isn’t a huge difference. We still want to have solid content, keywords and lists. But, unlike SEO where there are established tools and methodologies, GEO is a shifting landscape, and the parameters of what does and doesn’t work are in a state of flux.
To answer the second question, I will outline three key tips that have been mainstays of the GEO landscape so far and which are worth considering when producing your next piece of content.
The first thing to consider is a conversational tone. LLMs are predominantly receptive to text written in a natural, human sounding way, as it is how the models are likely to reproduce the text. The second is questions. Just as LLMs are receiving and reproducing a conversational tone, they are also answering questions being asked of them. This can effectively be capitalised on by asking, and answering, questions within your own content. The third is trial and error. Everyone has access to LLMs and it is worth checking on a monthly basis how you are performing in your desired fields, and if someone else is coming up repeatedly, worth asking yourself what they are doing that you’re not.
Producing strong, optimised content yourself is only part of the puzzle. LLMs draw on a wide pool of published material, and appearing in reputable, third-party sources – trade press, news articles, or industry reports – carries significant weight with the models. For those operating in financial services, the stakes of GEO are particularly high. The sector thrives on trust and authoritative voices, precisely the qualities that LLMs are trained to source. Whether it is a compliance update, a market commentary, or a thought leadership piece on regulatory change, the content produced by financial services firms is the kind of material that individuals are increasingly turning to AI to help them navigate and understand. Getting ahead of GEO now means that when a potential client, journalist or counterparty asks an LLM to point them toward a credible voice on, say, reinsurance vehicles or placement platforms, it should be your firm’s content doing the talking.
The above tips and information are not exhaustive, but I hope they have prompted some thought. As communications professionals it is important to consider every avenue down which an audience might arrive, and facilitate accordingly. Those not considering LLMs as part of their strategies moving forward may find themselves stuck in the past.