It’s that time of year when our teams roll up their trouser legs instead of their sleeves to dip their toes in the chilly seas of Brighton following a busy time at brightonSEO. This article contains some of their key takeaways (other than fish and chips!) and insights from the sessions attended.
With sessions focused around key areas such as: AI & Trust, Impact, Accessibility, Automation, Tech and Tooling, this year’s brightonSEO shared essential learnings and insights across the broad and growing world of SEO. Here we bring you summaries from some of our favourite talks.
AI & TRUST TALKS
LLMs vs Google: what search and engagement data tells SEOs – Baruch Toledano
In exploring searches, output and user behaviour, his big prediction is that ChatGPT may overtake Google usage by 2030, given the engagement and doubling of traffic YoY. However, whilst users are prepared to accept a poorer users journey, Large Language Models (LLM’s) still have a way to go in providing reliable results, with Google’s still being more accurate and trustworthy.
Beyond AI: How to own the SERP in 2025 with FAST content – Jon Earnshaw
Jon’s premise was that digital experiences are a conversation waiting to happen and we need to move fast to be part of it. Opening with “What Google ranked in 2023 is not what they want to rank in 2025”, Jon went on to share top tips on how to gain a voice in the conversation:
- By owning multiple features (answer cards, AIO, video etc) on SERP 1, you’re guaranteed a place in the conversation.
- Google will “Surface more content from creators through a series of improvements throughout this year”. They’ve recognised that they need to invest in a more trustworthy, dialogue rich conversation.
- Winning results now include AI overviews, quora, reddit and forum content alongside multiple video clips and images. His top tip was creating video touchpoints as people want fast information.
Everything your boss wants to know about AI search – Ryan Law
The main message from this talk was not to panic as Google still dwarfs every other source of traffic. 41% of all traffic comes from Google (in all websites in the Ahrefs study referenced in this talk). LLMs send as much traffic as reddit at the moment. The study stated that 80% of referring LLM traffic comes from ChatGPT. To prepare for their expected growth it’s worth investing in the content on core pages (homepage, articles and documents such as PDFs) which they focus on. On the question of whether to block AI crawlers, Ahrefs looked at 140M domains and their conclusion is that you probably should not (although GPTBOT was the most blocked crawler).
Google AI overviews are present in 7.26% of google searches, however this has since increased since the last core update. On the question of whether AI overviews are overtaking clicks, the research has seen them reduce clicks by 30% versus standard SEO results. It’s definitely worth optimising content for AIO queries which whilst longer and more specific, are less competitive and have a lower search volume. Work in this field has a cumulative effect, whereby you’re more likely to see an AI overview with a SERP that also contains a featured snippet (trying to solve a problem or answer a question).
On the subject of use of AI in content creation, of 900 content creators surveyed, 80% reported using AI in some capacity (but only 16% disclose its use!). Using their new “bot or not” tool Ahrefs analysed 770,000 pages establishing that 78% contained some sort of AI text, with 3% fully AI generated. The fact is that AI is increasing the frequency of publishing and therefore increasing its output. Smaller companies are seeing the most benefit from adopting the use of AI in this way, as it’s extending their capabilities beyond the limitations of manpower (5 times more expensive!). Companies that use AI content are growing organic traffic 5% faster than those that aren’t. The one caution is that publishing AI content without human editorial, is more likely to have a manual penalty imposed by Google.
Snake, meet tail: Understanding consumer concerns about AI – Laura McInley
Whilst 77% of advertisers have a positive perception of AI, only 38% of consumers agree.
In the UK, 56% are optimistic about the positive impact of AI on the economy and society, but only 35% trust companies using their data with AI .In the US there’s an even deeper divide with 51% excited about AI, but 62% not trusting that AI is capable of making ethical decisions.
People are better at identifying AI generated text than images and are more likely to lose trust in a brand that uses AI. The longer a piece of AI text is, the more likely it is to reveal itself with overuse of formal structures and repeated sentence formatting. When considering use of AI in video production, research has shown that participants have actively pointed AI generated videos in a negative way. There are generational differences in perceptions and attitudes towards the use of AI, with Gen Z most likely to be critical and Millennials being more open to it.
The safest approach to AI adoption currently is to use it transparently, verifiably, in ‘behind the scenes’ contexts and in ways that enhance customer experience such as with chatbots and recommendations
How google is treating AI generated content – Madeleine Lambert
Lots of sites took a hit in the March 2024 Google update which targeted what it deemed as ‘unhelpful content’. This was largely where AI content was being used indiscriminately to falsely scale, where expired domains were being repurposed for ranking advantage and where 3rd party content was being hosted that abuses a site’s reputation.
The big question was, could Google actually accurately identify AI generated content? 79,000 websites were studied, of which 2% had a manual action applied to them. Common traits of these sites were: very low DR values, low quality content, keyword stuffed, with listicles and generic blog posts. 100% of sites affected had published AI content.
Of 200 sites penalised with previously high volumes of traffic, 151 had clear signs of AI content, 30% of them had content which was 100% AI generated. This leads us to conclude that AI content was to blame for the sites identified as having ‘unhelpful content’, with Google looking for quality and authenticity.
So whilst AI provides efficiency, there are associated risks. To mitigate these, use AI as a supporting tool, ensuring that your content offers real value to your audience and aligns with Google’s standards.
IMPACT TALKS
SEO gets you found , CRO gets you chosen. Why brands must do both. – Oliver Yee
The SEO landscape has changed more in the last 12 months than it has in 12 years. With ChatGPT reaching over 400 million weekly active users and AI overviews appearing on 37% of searches, we are staring down the barrel of a significant SERP change, but don’t raise the white flag just yet.
We need to make every session from the organic channel count and that takes somebody leading the charge on optimising our websites and unifying the previously siloed UX, content, CRO and SEO disciplines. The only way we will win this war is together! We need to understand how we work together and reframe the purpose to be user-centric rather than technical-discipline centric i.e.
- Instead of saying “is my metadata optimised for the terms we’re targeting” we can ask “is what’s in the metadata solving the users problem?”
- Instead of asking “is the website meeting CWV score?” ask “Can my users get to the solution of their problem more quickly?”
If we speak in the customer’s language, we’re more likely to get the right outcome, and Google’s current algorithm indicates that approach is one worthy of reward.
Winning free listings through feed optimisation – Nikolaj Mogensen
SEOs have become used to optimising for the usual “10 blue links” in Google SERPs. With Google’s new free product listing module taking up 20% more SERP real estate above the fold in the last year, is this ‘the end of blue links’?
There are 4 levers we can still pull:
1) Produce better product content
2) Make sure our technical SEO is on point with product schema and solid site speed etc.
3) Ensure that the paid search team has the Product feed – XML sewn up.
4) With Product Detail Pages (PDPs getting more content, make sure your product page templates are in line with best practice.
The product feed is essentially like a sitemap. The better the data quality, the better Google is able to surface products in results.
So make sure you’re including:
- Title – e.g. for clothes include colour, size, colour
- Description – Make sure this is not too sure and includes the main attributes of the product
- Path.
With potential further additions (where applicable) of:
- Material
- Pattern
- Colour
There are useful guidelines Google has provided for each element, including those that are rarely used which might prove to be your secret weapon:
- Product highlights – Populate this with product USPs
- Images – use Google’s preferred format for optimal results.
Local SEO for real people: 20 hard-hitting (and hilarious) audit lessons – Ehab Aboud
Ehab included an update on the amount of false competitor audits and fake profiles that are flooding Google results and revealed that 12 million reviews were removed last year! There was enlightening information like Google bots ringing business numbers for restaurants and changing their opening hours automatically if the phone wasn’t answered! He encouraged us to review Google edits within the profile and to update it little and often, by uploading 100 photos, 40 now, then 2 every week for a year to improve relevancy. His top tip was to handle bilingual naming conventions through google maps suggestions not via the GMB profile. that are flooding Google results. 12 million reviews were removed last year! There was enlightening information like Google bots ringing business numbers for restaurants and changing their opening hours automatically if the phone wasn’t answered! He encouraged us to review Google edits within the profile and to update it little and often, by uploading 100 photos, 40 now, then 2 every week for a year to improve relevancy. His top tip was to handle bilingual naming conventions through google maps suggestions not via the GMB profile.
CONTENT TALKS
Momentum – how publishing and updating frequency impact SEO – Pontus Vippelius
The most underrated part of topical authority is momentum which is achieved by publishing and updating quality articles more frequently than a competitor. Google uses frequency scores to understand search patterns, so by demonstrating a regular quality output, Google analyses that as being worthy of returning to more frequently. That said, it’s important to be organic in not being formulaic in having a set day/time to publish content, as Google perceives that negatively.
What Google does like is when you focus on enriching content around a particular topic or entity. When determining which this should be, it may be worth researching competitor behaviour to inform your approach. Looking at lastmod in the XML sitemap to identify when content was last updated can help where this is unclear.
Another top tip, is that whilst it seems counter intuitive to maintain your publishing schedule throughout site migrations, it proves momentum, where a drop has been proven to cause site performance to suffer.
ACCESSIBILITY TALKS
What the f**k is alt-text, and how do you write it? – Liam Cumber
‘Alt’ (alternative) text is an accessibility feature. It helps partially sighted users and acts as a contingency measure where images can’t be loaded. In the UK 24% of people have a registered disability that impacts how they interact with the rest of the world. In addition to this are any number of people suffering temporary impairment who benefit from accessibility measures. A further audience is those whose internet bandwidth is limited or restricted.
Writing alt text requires an understanding of how people engage with your brand. Good alt text is rich and descriptive, breaking down the content of an image from its context. If something is inferred by the image, then describe this too, Alt text is a translation, not a transcription. If an image is purely decorative, you can simply leave the alt text unpopulated.
TECH SEO TALKS
XML + Hreflang: A practical guide for large ecommerce sites – Nick Samuel
HREFLANG is used to target geographical and language variations of sites. There is no “best” implementation. Historically, HTML has been a common implementation. Your experience tends to impact which method you prefer. Using Simlarweb analysis of a set of the worlds top ecommerce sites, the majority (61%) use HTML, on average, fashion ecommerce sites had 30 HREFLANG tags, for example:
- Lucy & Yak have 213 tags.
- Passenger have 237.
- Pull & Bear have 696 tags.
This backs up the thinking that whilst HREFLANG wont necessarily improve your rankings – it will help to ensure the right page ranks in the right market.
HTML implementation is the most intuitive method and remains the default for most common sites with 4% using HTML and XML sitemaps together (e.g Tommy Hilfiger).
There are use cases where it makes sense to use XML sitemaps over HTML implementation, these are usually determined by the business context and dev priorities such as:
- If HTML isn’t dynamically updated
- If HTML is updated, but fundamentally flawed (e.g if equivalent language pages don’t exist, thus creating 404s)
- If you operate multiple domains over multiple CMSs
- You lack development bandwidth
- Your <head> is difficult to maintain
If you wish to build a business case for HREFLANG then it likely revolves around it being a revenue generating opportunity (traffic x conversions = revenue). HREFLANG has been proven to improve the quality of traffic, which improves conversions. Local traffic converts better than non-local traffic. A strong case can be built upon a test and learn framework where you analyse, recommend and then implement.
There are 3 ways to approach making an hreflang xml sitemap:
- Automated
- Manual – best for low dev resource
- Semi-automated
Useful tools include:
- technicalseo.com (good XML sitemap tool for manual implementation)
- hreflangbuilder.com.
Handy Tools referenced throughout other conference talks included:
- coreupdates.com provides a probability of whether a keyword will trigger an AI overview.
- The SEO patent podcast
- seonotebook.com has a good LLM consensus
- Rqpredictor.streamlit.app – how Google categorises individual keywords
“The queues along the seafront tell you everything you need to know about Brighton SEO. This flagship event is super popular, attracting attendees from all over the world – and with good reason. It’s such a polished and well organised event, with great opportunities to learn regardless of your experience. You’re always sure to bump into some familiar faces as well as old friends. This year’s April event was no exception – a vast array of talks, delivered with energy – and a few beers afterwards. The wonderful Brighton weather was just the icing on the cake!”
James Chapman, Senior SEO Strategist“Sun, Fun and SEO vibes! I attended the ‘Geek fest’ that is brightonSEO 2025 last week and it did not disappoint! With talks from top SEO’s like Lily Ray and Greg Gifford, it was two days of fast paced SEO action, covering AI, tech and content best practices and upcoming trends. For me the best thing about B/SEO is how you feel so immersed in all things search. Day and night you can’t help but just be buzzing and motivated to get back to the laptop and start testing out all the new ideas and strategies you have absorbed. Totally amazing and I would recommend it to anyone regardless of your knowledge or skill levels, Brighton welcomes you all!”
Gary Billington, Technical SEO Manager.