❌

Normal view

There are new articles available, click to refresh the page.
Yesterday — 17 November 2026Main stream

Scottish Web Folk 2026: a canny, reflective and honest conference

17 November 2026 at 09:00

I’ve had the privilege of attending the Scottish Web Folk conference for the last three years and this year’s event was as informative and candid as always, tackling the topic of ‘Human vs Machine’.

At other conferences and in other industry presentations, excitement about what the machines, specifically AI, can do, often whips up a don’t-get-left-behind fervour. But at this event, I think it would be fair to say that the machines were being viewed with a canny, human eye. We are deep into our new AI era and have an increasingly solid understanding of its possibilities and parameters for safe use in web.

We’ve been here before

In his keynote speech, Andrew Millar from the University of Dundee kicked off the ‘Human vs Machine’ discussion by looking back at points in our industrial history where technological leaps had forced change in the labour market, and then made valuable the associated, human, skills that would correctly manage the new technology. The printing press, for example, made redundant scribes and copyists, created a need for typesetters, and had the knock-on effect of requiring judgement and authority to manage the wider dissemination of the printed word. When computers arrived in offices, the filing clerks that were able to shift from paper to software found their analytical and decision-making skills could usefully harness digital categorisation and reporting. Not long after inception, the internet brought new levels of digital publishing reach, and the associated friction between volume and credibility, but the human vs machine tension has taken a new turn with the introduction of AI-generated web content and AI search.

What I took from the keynote speech was not just that we’ve lived through technological change and adapted, but that we’ve been incrementally building the human muscle required to manage the current technological era where AI is ever more dominant. Judgement, authority, analysis, decision-making and credibility are going to be central to managing AI into maturity.

Highlights of the day

My thoughts were focused on advocacy and responsibility as the conference unfolded. What do we already know about wielding our new technologies well (or facing the fall-out afterwards)? A few sessions in particular stood out.

Just because we can doesn’t mean we should

Stratos Filalithis from UoE and Kris Purdy of the University of Glasgow gave an ‘unscripted’ presentation that shared their frustrations of uncovering enormous legacy web estates. Both Universities, beyond their central platforms, had experienced a two-decade boom in affiliated web publishing without fully centralised or localised content management.

Stratos and Kris shared their strategies to build an accurate picture of their Wild West estates, and then spoke of the challenges ahead of defining ownership and instilling governance to create a supported and managed herd. Not just ‘unscripted’, but ‘unfinished’, the honest, unpolished session offered no pat answers but was a resounding cheer for the humans that tackle backlogs and runaway technologies every day. Collaboration, sharing and never accepting defeat are still singularly human approaches.

We already know how to write for AI

Under-managed, Wild West web content poses a risk to reputation because it potentially serves redundant, out-of-date or incorrect information. Well-managed content, however, seems to be largely AI ready, so it’s heartening to hear that existing skillsets are still valuable here.

Paul Costello and Rob Allen from Numiko showed in their work for the University of Oxford that the findability strategies that have served SEO are still central to GEO (generative engine optimisation; optimising for AI chatbots like ChatGPT or Claude) and AEO (answer engine optimisation; optimising for AI search).

Nick Daniels, Katie Spearman and Mel Batcharj from the UoE UX team presented core approaches to writing for AI, and these are already embedded in existing content design practice. Clear, concise content, meaningful headings and links, and accessible text all create AI-friendly content. If you missed the talk at Dundee, and the session at the UoE, their blog is well worth bookmarking for actionable takeaways.

June Content Improvement Club: Is your content AI ready? – Website and Communications Blog

We have to design technologies for operators as well as end users

As technology and its associated processes mature, we have to design for its operators as well as its end users. This is high on the agenda for me as service manager for the University’s web platform, supporting our internal user community to deliver content for their external users.

Lauren Tormey’s session set out to apply design practice to the processes of an internal team creating user-centred content for UoE Degree Finder. In incremental stages, she workshopped a playbook with an already expert Prospective Student Web Team to

  • stress-test existing content guidance
  • uncover new contexts where ‘If X then Y’ had not yet been defined
  • empower the team through live editing workshops so that everyone was equally experienced in applying the styles as defined in all contexts

With major revisions overdue on our self-service resource, the EdWeb 2 Hub, I have direct takeaways from this session to recreate our existing content stylesheet and editing practices. The platform has evolved significantly in the 18 months since the Hub was initially published, but the team updating the Hub while also supporting the user community day-to-day is tiny. Workshopping our existing styles and processes to update the content might feel like a diversion under our current time pressures, but it’s also an opportunity to stress-test our content practices and invest in the team. It’s important that we are equally and consistently skilled at managing updates, whether working at volume during intense platform evolution or in a more settled pattern of rolling revisions.

Final thoughts

In his keynote speech, Andrew Millar quoted an episode of Doctor Who where an AI bot was nudging the world towards Armageddon. The Doctor observes that it’s not the machine that’s the problem, but the pressures and choices around it.

As professionals working in the higher education web space, it’s our responsibility to employ that canny, human eye to embrace the technologies that benefit us when used correctly. Are we using them because we should or because we can? Are we thinking of the legacy that our actions now will leave? What knowledge have we already accumulated that can inform our use of new technologies?

And how do we ensure that everyone using our technologies, whether operators or end users, is supported and protected from pressures to do more faster so that collectively we all contribute to the successful maturation of a new technology? I came away from the conference with the sense that as long as we keep sharing our experiences of past, current and new technologies, and are honest about our challenges and approaches to managing them, our choices can be positive.

Before yesterdayMain stream

2,117 websites and counting: welcome to our Wild West Web Estate

How many websites does the University of Edinburgh have? A few years ago, the honest answer was “a lot, probably more than you think”. During recent years, we built a better idea about their number, shape and performance. This summer, we have counted them all, audited them by hand, and this is the first post in a series about what we found, what it means and what we are doing about it.

What do Edinburgh trees and our Web Estate have in common?

Last week, I had the pleasure of opening the HE Connect 26 conference here in Edinburgh and part of my introduction was, as any good host would do, about bragging about our city. As you might be aware, it is built on seven hills, two of them are extinct volcanoes, and it has more trees than residents (according to the official Council records).

Websites are not forever, and neither is fragmentation: reflections from HE Connect 26 – Blog post by Stratos Filalithis

Those weren’t random trivia about Edinburgh, at least not one out of three. I shared the information about the amount of trees by adding that approximately 75% of them are privately managed. This constitutes a hugely valued shared asset, looked after by many different people, with varying levels of care.

That is our web estate in a nutshell. A large collection of websites, hosting publicly available content, promoting the University, its study programmes, world changing research, available services and other relevant areas. This collection is following a devolved ownership model, much like every other area of the University itself, which results in a variety of quality, performance, technology usage and maintenance approaches.

What are we counting and who is accountable?

To set the scene, it’s very useful to share some definitions about what we consider a website and our web estate to be:

  • A website is a discrete collection of pages under a single web address, built and maintained as one entity, delivering content to external audiences without any requirement to login. It must have a defined owner.
  • Our web estate is all University-owned websites, web applications and publishing platforms, centrally managed and devolved. These host publicly available content for all our external audiences without any requirements to login.

Our devolved ownership to University Schools, Units and Departments is a local arrangement. For regulators, though, the University is a single legal entity. Accessibility and data protection regulations apply to all of it, and our responsibility to the Information Commissioner’s Office (ICO), which is the UK regulator, sits with the University. But, given our devolved governance model, each website owner is accountable to carry these responsibilities on behalf of the University.

How many websites we actually have and where do they live?

The latest count of websites in our web estate is 2,117. This is a growth of 30% since our last full web estate audit.

It includes all “live” websites who either use the University’s corporate domain (ed.ac.uk) or have a direct relationship or affiliation to the University.

Approximately 8% of this web estate is hosted in our University Web Platform, EdWeb 2. Even though this is a small number, it includes a lot of of our highest priority content: 20 out of 25 School websites, the Degree Finder, our corporate homepage and website, and the majority of our student and staff services.

The rest of the 92% websites are using various technologies, hosted across several platforms, managed by roles and teams having a variety of skills.

Through our close collaboration with the University’s Digital Preservation team, we have uncovered a lot more detail about the history, legacy, structure and lifecycles of our websites. David Mahoney, Digital Archive Assistant, has conducted thorough research on the history of our web estate, and published a blog post about his approach and findings.

Strata: Uncovering the Layers of the Archived Web – Blog post by David Mahoney

Key initial learnings from our manual audit

To establish the strongest possible baseline and benchmark, we took the hard decision to audit each one of the 2,117 websites manually. I would not usually recommend this approach, because it’s very time consuming, but it was essential to bring our experience and organizational knowledge to fulfil the gaps in ownership. An additional benefit of the manual audit was assessing the risk per website, based on use of technology, content freshness or other factors. It helped us, also, to identify the more common website use cases, approaches and risk triggers, and allowed us to better inform our initial key learnings.

Ownership means a named person

The majority of our web estate is directly related to our College’s activities, either teaching or research. For most websites, we can identify the College or Professional Group responsible, but we cannot name a person as the website owner. “The College is owning this website” is not a tenable position, since it does not carry any real accountability, responsibility or direct action with it.

Technology and content are our highest risk factors

The audit is looking at what our websites are made of. Most of our web estate runs using popular technologies, like Drupal or WordPress, but not necessarily on a supported version. One of our current targets is to identify these and take appropriate action. Our findings have confirmed that a portion of our web estate is not actively looked after, a percentage that grows as a website gets older, which increases our security risks.

It’s not only the technology that is left behind, though. Content, in its own right, is not reviewed, updated or removed, leaving a large number of out-of-date or inaccurate pages. This reduces the trust of our visitors against our websites, since they are not sure what to trust.

Websites must follow a clear lifecycle

As you might have guessed already from my presentation title, “Websites are not forever”, I am a true believer that everything has a beginning and an end. And websites are no exception. Some of them evolve, get a redesign, upgrade or a relaunch, which is an ending on its own, but most of them need to fulfil their lifecycle by retiring at the proper point. A large part of our websites was launched in relation to a time-limited activity: a research project, a University initiative, a conference, event or course-related requirement, but when that activity stopped, there was rarely any action to retire the website content, too.

Artificial Intelligence (AI): the great amplifier

All of the learnings and risks are amplified by the emergence of AI technologies:

  • Out of date and legacy content is scanned daily and “pollutes” the accuracy of the information AI agents reply on our behalf to their users. Someone asking information about a specific degree area in the UK might receive false information about Edinburgh, and disregard our offer based on it.
  • AI tools improve their capability in scanning digital platforms for vulnerabilities, while the frequency of cyber attacks gets higher as time goes by. One can only imagine the risks of this technology in the hands of cyber criminals.

Our best defence is accurate, current content on well-managed platforms, in a well governed, compact web estate.

Where are we going next

We are in the beginning of this journey, and we are gaining momentum. The next phase of the audit is under way, the outputs being shared with the College and Professional Group Heads of IT, covering ownership, content accuracy and age, and technology. We will trigger a decision point whether to retain, move to EdWeb 2, keep and support, or retire ensuring appropriate archiving.

In parallel, a formal web estate project with a steering group is kicking off, we are working closely to map our web estate requirements against the University’s 10-year digital plan, and our web estate governance hub is being updated to better support our website owner community, with a new Web Trainer role to turn guidance into training.

The progress of this journey will continue to be covered in future blog posts, sharing more of our learning, best practice and, hopefully, significant progress in taming our Wild West Web Estate.

 

Websites are not forever, and neither is fragmentation: reflections from HE Connect 26

The annual HE Connect came to Edinburgh this week, and I had the privilege of opening it before spending two days listening to peers wrestle with the same problems we have. Here’s what stuck with me, from large University web estates to AI recommending (or rejecting) your university before you’ve even said hello.

Coming home to HE Connect

At the end of HE Connect 25 in Manchester, it was announced that the 2026 edition would be held in Edinburgh. I wrote then that I couldn’t wait to welcome colleagues from the UK and beyond.

Reflections from a conference focussing on digital leadership in higher education: HE Connect 25 – Blog post by Stratos Filalithis

On Tuesday 6th of October, digital leaders from across higher education and agencies specialising in the sector met at the Royal Society of Edinburgh, and I welcomed everyone with presenting on “Websites are not forever: what to keep, what to fix, what to retire.”

In it, I followed the lifecycle of a website, from the excitement of launch to the less glamorous questions of maintenance and retirement, and argued that most of our estates are in trouble because we’re much better at the first stage than the last. I also made the case that AI and security have turned forgotten content from an untidy habit into a real, current risk. Furthermore, I had the opportunity to do a 5′ lightning talk going through our work in digital sustainability and testing Artificial Intelligence’s (AI) capabilities in supporting accessibility testing.

A large room hosting approximately 40 attendees sitting in rows of chairs. A man is standing presenting in front of a large screen having a photo of a landscape with sea and forests.

On Wednesday 7/10, I presented a 5′ lightning talk covering our work in Digital Sustainability and testing AI’s capabilities against accessibility testing.

I’ll share more in a separate post, because this one is about what I was most looking forward to: sitting back, listening and learning from digital leadership peers sharing their challenges, experiences and approaches, and discussing with them in an honest, collaborative and constructive way.

HE Connect 26 programme and speakers

If there was one common thread, it was this: nobody designs fragmentation. It happens to us as “a result of disconnected and, sometimes, very sensible decisions”. The interesting question is what we do about it.

Coherence, not centralisation

That line came from Samantha Fanning, Director of Digital Experience at University College London (UCL), and it might be my favourite of the conference. UCL is a similar size to the University of Edinburgh, having around 50,000 students and 17-18,000 staff, so it was hard not to see our own estate in her slides.

Her advice was simple: don’t focus in superficial approaches to fix “the front door”, like a homepage. The issue is not the door, it’s the house itself. Part of UCL’s approach is a 3-space model, which covers all levels from trusted, owned and governed centrally, institutional knowledge and services, owned and governed centrally; to local knowledge and innovation. The principle to take away is to create shared foundations before seeking control, make ownership visible, and aim for coherence, not centralisation. UCL backs this with their own design system and experience framework. A well researched and supported design system is fundamental to implementing effective digital governance, and establishing one is one of our own objectives here in Edinburgh, too.

Johannes Nygaard, Chief Digital Advisor at the University of Copenhagen, showed what web estate transformation looks like. In 2023 they had one site serving everything from recruitment to science communication: 130,000 pages, 900 editors and 1,200 subsites. Two stepping stones, a new CMS and a new governance structure, signed off by the university’s senior management, included simple core principles: Every site targets specific external audiences, and who the most important are, resources should be prioritized based on importance of interaction and traffic, usage of templates for sites and pages. Perhaps the most controversial one in a University setting is that pages and sites with very little traffic would be closed, but it’s also the difference between a managed web estate and a digital museum.

Both of these sessions struck a chord with our work in Edinburgh, especially since the contain some of the possible answers to our own fragmented web estate challenges, and I have been encouraged to see what takes place in UCL and Copenhagen.

When AI is the front door

Piero Tintori, CEO of Terminalfour, brought numbers justifying the rise of AI usage in selecting a University programme: About 7% of prospective students go to an AI chatbot first for a specific institution request, with 18% of then having removed one from consideration based on an AI-generated result only. Still, 73% of parents name the college website as their top source, and 43% of applicants still start on a university website compared to 13% who start with AI. Most people still visit and trust our websites more than AI, but the latter covers the ground faster, and we will need to keep checking these numbers to keep track of if, and how fast, the prospective students and parents decisions change.

A very interesting sharing was that old content pollutes AI answers. Archived and outdated pages get used as sources, results vary from session to session, and the parts most likely to be wrong are the ones that matter most, like entry requirements, fees or key programme information. Piero’s answer was a move to content engineering: modelling, metadata, schemas and separating content from design, with AI helping to reduce the effort involved by ensuring the published content is accurate across all the web estate. My action will be to expedite action to archive and retire heritage content.

It’s clear that, especially in the age of AI, effective web governance and coordination is the key to address the evolving challenges and risks that carries a devolved web publishing environment.

Why is digital so hard?

Kris Purdy, Head of Digital Experience at the University of Glasgow, asked exactly that. His answer was even though digital sits in the middle of business, user and technology: “we built the digital, but we didn’t build the organisation to support it.”. Glasgow’s challenges are very similar to our own: many websites without a named owner, and uncertain plan for decision making, consolidation and retirement. Our own estate is larger, but the problem is exactly the same one, and I am very happy working closely with Kris to share how best to address these.

I really liked the mention of the “missing link” between strategy, which sits at board level, and delivery, which sits at team level. The defined step between approval and development, which translates strategy into “how”, is often the most critical part to unlock effective change, yet is missing either due to lack of funding, availability or prioritisation within our organizations.

This workshop triggered my interest a lot since I wanted to hear what the delegates had to say, especially those representing agencies as there could be insights directly related to our own web estate journey.

Governance is about people (again)

One unsurprising common theme across all sessions and informal discussions was that people is the catalyst and enabler to any change. This is not a technology challenge and has never been. Its fast evolution, though, like AI currently, amplifies the risks of not having a coherent and effective governance structure, alongside the appropriate technology.

This was a key takeaway from June Shannon’s and Paul Leigh’s sessions, both from the University of Limerick, too. In a very similar note, Helen Robertson and David Jones from GAIN Higher Education in their session, “What does your frontline know before your numbers do?”, also made a very good point: the person closest to the problem is the one closest to the solution. Finally, I thoroughly enjoyed Lucía Tornero González’s, who travelled from Madrid, about how Universidad Pontificia Comillas is approaching the graduate experience, creating everlasting moments for them and their families.

A big thank you to Janus Boye, who organised and facilitated the conference, and to every speaker for being so open about what works and what doesn’t. If any of this sparks a thought, a challenge or a question, I’d love to hear from you. Use the comments below, or get in touch directly at stratos.filalithis@ed.ac.uk.

❌
❌