Rethinking digital strategy for a university website in an age of AI-generated answers
In my last post, I wrote about the increasing reality that more and more people will encounter our institution without ever visiting our website. AI tools are changing how people search. Chatbots are summarising content, pulling in information from across the web — and increasingly, LSTM will be part of those answers whether we design for it or not.
That post was me thinking out loud. This one is a reflection on what that means in practice, and the direction we’re starting to take as we build the next version of our website.
Here are some of the areas we’re actively exploring.
We’re shifting from designing for pages to designing for extraction.
One of the first things we’re considering is how we structure key pages. Rather than assuming someone will read from top to bottom, we’re starting to think about how our content might be quoted or summarised in isolation.
We plan to rewrite priority pages so the most important information — delivery mode, start dates, location, entry requirements — is visible in the first third of the page. This isn’t just about SEO. It’s about shaping what AI tools might pull out when users ask a question.
We’re also reviewing our approach to long-form content, and starting to test modular blocks with clearer subheadings and standalone sentences. We want to make our content more paraphrase-friendly — without compromising depth or nuance.
We’re beginning to experiment with prompts, not just keywords.
Rather than focusing solely on search engine rankings, we’re exploring the types of prompts real users might give to AI tools. Things like:
“Which UK universities offer MScs in global health with part-time options?”
“What’s the best university for research on neglected tropical diseases?”
We’re starting to feed these into tools like ChatGPT, Gemini, Perplexity and Claude to see:
- Do we show up?
- How are we framed?
- Which content is being pulled?
- What’s being left out?
This early work is helping us prioritise what needs rewriting, what needs simplifying, and where our broader digital footprint might be holding us back.
We’re rethinking what it means to “be visible.”
We’re beginning to shift our thinking away from traffic alone and towards a broader sense of presence and representation.
We want to understand:
- Are we included in the kinds of AI answers our audiences are seeing?
- Are we being described accurately and consistently?
- Do we show up even when the question isn’t directly about us?
This is pushing us to look beyond our own site: partner content, staff profiles, media interviews, and alumni mentions. Our goal is to create a digital presence that can be surfaced and trusted across a wider ecosystem.
We’re expanding how we define performance.
We still care about analytics, but we’re starting to think about other signals too.
What would it look like to track:
- How often LSTM is mentioned in AI answers?
- How we’re described or summarised?
- Whether our voice and tone are consistent across platforms?
We’re also exploring how we can capture these kinds of interactions — even when they happen without a single click.
What we want to do next
These are early ideas, but they’re shaping how we approach the next phase of our website project. Here’s what we’re hoping to explore in more depth:
- Mapping which pages are most likely to influence AI-generated answers, and treating them as high-impact content.
- Developing patterns for content structure that help maintain clarity and consistency across teams.
- Exploring tagging and metadata models to make research and academic content more accessible to machines.
Conclusion
The first blog was about the shift we’re seeing. This one is about the questions we’re starting to work through.
Because if AI is going to explain who we are, we want to help shape what it says. We may not control where users land, but we can influence how our story is told.
We’re not chasing traffic. We’re laying the groundwork to earn trust. And in a world of disappearing homepages, that’s what will matter most.
Leave a Reply