BrightonSEO, our go-to SEO conference, was back on the seafront this October, and so were we. With a packed schedule of talks and panels, the Hookflash team split up to cover as much ground as possible, then spent the journey home comparing notes (and caricatures).

What we found was a conference with a slightly different feel to the spring. Below, we've pulled together the sessions that sparked the most conversation in the office afterwards, from ecommerce and product feeds to what SEO teams could look like by 2030.

Ecommerce SEO, AI agents and getting back to basics

If April was all about AI visibility, October felt like the industry taking a breath and asking what that actually means in practice. AI was still everywhere, but the conversations were far more grounded. Speaker after speaker came back to the same point: AI platforms and agents can only recommend what they can read, understand and trust.

That put ecommerce firmly in the spotlight this time around, with product pages, product feeds and structured data getting a lot of attention. Alongside that, there was a refreshing focus on the fundamentals, and a look ahead at what SEO teams might look like by 2030.

Technical SEO: Going Back to Basics

With so much noise around AI search, it was good to see a talk that stripped things right back. The message was simple: the fundamentals haven't gone anywhere, and they matter more than ever.

Technical SEO is about removing friction between being found and being indexed
Content does the heavy lifting when it comes to being ranked and cited

The common mistakes we still see

Marking up content users can't see. Schema should only describe what's actually visible on the page.
Confusing your metrics. Core Web Vitals and PageSpeed Insights are often used interchangeably, but they aren't the same thing.
Treating canonicals as a rule. A canonical is a hint, not a directive. Google can, and will, ignore it and pick its own.

Key takeaway: Before chasing the latest AI tactic, make sure your site can be crawled, indexed and understood. Everything else builds on that.

Ecommerce SEO

Ecommerce was one of the standout theme’s for us, and the advice was refreshingly practical.

Facets vs filters

These two terms often get lumped together, but there's an important difference:

Filters are a tool to help users refine a page and reduce friction. They don't need search demand.
Facets are categories that do have search demand and deserve their own indexable URLs.

So how do you decide which facets, and which multi-facet combinations, are worth going after? It comes down to three things:

Demand: set a minimum search volume you're willing to target
Product depth: set a minimum number of products (around 6–8). Do you have enough to justify an indexable page?
Intent: what does the user expect to find? Look at what's currently ranking on page one. Is it PDPs or PLPs?

Once a facet earns its place, the page should be indexable, have one clean URL path per facet (no matter what order the filters are applied in), a unique title, meta description and H1, a self-referencing canonical, a place in the sitemap and unique content that makes it worth ranking. Anything with three or more facets should be noindexed, and you shouldn't allow multiple facets from the same group.

URLs need careful management here. It's easy to end up with 10,000 of them, and while Google can cope with that, AI crawlers are far less forgiving.

Product pages aren't pages anymore, they're data sources

One of the more technical talks made a point that stuck with us: a product page can look perfect while the facts behind it disagree. If product data is inconsistent or incomplete, AI can't reference your products with confidence.

The answer is one source of truth, not a string of manual fixes. A single product identity, say a small green coat, needs to be consistent across five systems:

URL and canonical: is the variant URL stable? If you load the black version, switch to green and then start a new session, do the image, price, availability and variant selection stay the same?
The rendered page: turn off JavaScript. Are the product facts, like size and colour, still in the HTML?
Merchant Center feed: the link must land on the exact variant submitted, on page load
Internal linking: can your site structure actually reach it?
Product markup: valid Product and Offer schema is the bare minimum. GTINs, MPNs, brand, ProductGroup and variant schema, aggregateRating, shippingDetails and hasMerchantReturnPolicy all add extra clarity.

Canonicals signal your preferred URL, but ProductGroup schema explains the relationships, i.e. which products share the same parent.

When should a variant get its own page?

Only when it has its own search demand and genuinely unique content. Ask yourself:

Demand: do people search for this variant by name? "Nike Pegasus 41 black", yes. Sizes of a t-shirt, no.
Distinct product facts: is it actually different?
Unique assets: does it have its own photos, descriptions or reviews?
Stock and lifecycle: will it be around long enough to rank? Seasonal colours that sell out just create dead ends.
Internal-link support: can you realistically link to it from categories, filters and navigation?
Duplication risk: if the pages are near identical, you're risking thin content, cannibalisation and Google picking its own canonical.

Common Shopify issues

Not everything here needs fixing, but it's helpful to know where these issues come from:

Redirected misspellings: Shopify redirects misspelt product and collection URLs rather than returning a 404. This only really matters where those URLs have backlinks, so review your backlink data first. Suspicious links should be 404'd, though that's rare and needs dev support.
Parameter URLs from filters: these waste crawl budget, but they're handy for bookmarking and sharing. If a filtered view has its own search demand, give it its own page. Rather than simply blocking them in robots.txt, work with your devs to remove the href links on load in the first place (a dropdown is a good alternative).
_pos and _sid URLs: these come from "pairs well with" and recommended product modules on PDPs, and Shopify uses them for attribution. If nobody on the team uses Shopify Analytics, ask your devs to remove the parameter. If they do, switch to event-based tracking in GA4.

Key takeaway: In ecommerce, consistency is everything. Your URLs, page, feed and schema all need to tell the same story.

PDPs are the new PLPs

A real team favourite was Greta Koivikko from Novos, who made a compelling case for rethinking where ecommerce brands focus their SEO efforts. Her message: product detail pages deserve far more attention than they usually get, and brands that treat them as mini storefronts will be the ones discovered across both search and AI platforms.

A shift in how people search

Category pages have long been the backbone of ecommerce SEO, ranking for the broad commercial queries that drive most traffic. That's starting to change. As people get used to talking to LLMs, they're bringing the same habits to search:

Queries are now around 1.5 times longer than they used to be
Shoppers describe what they need rather than typing a few keywords. "Black women's shoes" becomes shoes you can run in with friends at the weekend and also wear casually
Search is moving from keywords to keywords plus context

At the same time, the SERP itself is starting to behave like a category page. Shoppers research and compare without leaving the results, then land directly on a PDP ready to decide. That makes the product page the first, and often only, impression of your site.

Make sure your content can actually be seen

Great PDP content often goes unseen. Greta used size guides as an example: incredibly useful for shoppers, but often loaded with JavaScript. Google can render this, but LLM crawlers can't. She showed ChatGPT being unable to see the size guide on a UNIQLO product page. The fix is simple: if you want AI platforms to read your key content, render it on the server.

The SEED framework

To build PDPs that work for customers, search engines and AI at the same time, Greta introduced the SEED framework:

Situational relevance: be clear about who a product is for, and who it isn't for. Greta suggested turning personas into product attributes using Shopify metafields.
Example-led content: show how a product fits into everyday life. Labels like "Perfect for", "When to use" or "Do not use with" answer the questions shoppers are already asking, from whether a plant suits a north-facing room to whether a skincare product works alongside retinol.
Enriched product data: your product information feeds schema and product feeds used by Google Shopping, AI platforms, paid ads, social and marketplaces. Mismatched pack sizes, prices or missing attributes between your site, feed and retailer listings quickly undermine trust.
Distinct brand voice: shoppers can buy your products from plenty of places, and retailers are happy to earn the citations that come with telling your product story. Following E-E-A-T principles and using Digital PR to earn mentions in the publications AI already trusts helps make your PDP the source AI chooses to cite.

Greta's PDP checklist

Avoid client-side rendering for key content
Map product attributes to your personas using insights from Reddit, FAQs and reviews
Add contextual cues to your PDP template
Turn key product details into metafields that feed your schema and product feed
Pitch product-led stories to the publications AI cites most in your category

Key takeaway: Agents and AI platforms still need pages they can read and trust. Accurate product data still wins, and trusted sources still decide who gets chosen.

GEO & AI Search

Agentic ecommerce: your product feed is doing more work than you think

Henry Price from dentsu took us through agentic ecommerce, and it's a topic every ecommerce brand should have on its radar. Discovery is spreading well beyond the SERP into free listings, image search, LLMs and, soon, AI agents that buy on a user's behalf. And all of it runs off one thing: your product feed attributes.

Most brands stop at the bare minimum

Only five feed fields are mandatory (id, title, price, link and image_link), and most brands stop there. That's a missed opportunity, especially as Google-format feeds now work in ChatGPT too, so one feed can cover both.

Henry's advice for getting more out of your feed:

Titles: follow Brand, Product, Category, Attributes, with key terms in the first 70 characters
Descriptions: lead with function and material
High-impact optional attributes: product_highlight, product_detail and question_and_answer
related_product: lets you tell LLMs what to recommend next
Prices: if your price only renders through client-side JavaScript, neither the Merchant Center bot nor LLMs will see it
Consistency: your feed and PDP need to match word for word

Measuring it

Use Merchant Center and Search Console together to get the full picture. Worth noting: the GMC AI report is currently US only.

Key takeaway: This is worth doing now. Going beyond the mandatory fields helps today's free listings and AI Overviews, whether or not agentic buying lands on schedule.

E-E-A-T & Trust: Lessons From Charities

Charities are held to a particularly high standard online. Between regulation and the constant fight against misinformation, trust isn't optional, and this talk broke down how E-E-A-T applies across three levels:

Brand-level E-E-A-T: the reputation and credibility of the organisation as a whole
Content-level E-E-A-T: the accuracy, depth and sourcing of each piece of content
Author-level E-E-A-T: the expertise and experience of the people behind it

Key takeaway: For high-trust sectors, E-E-A-T needs to be built at every level, not just added to an author bio.

SEO Strategy & the Future of SEO

Panel: What will SEO teams look like in 2030?

One of the most thought-provoking sessions of the day was a panel with Mark Williams-Cook, Laura Iancu, Manuel M. Madeddu and Simon Lesser, exploring how AI is reshaping SEO, from how we measure and prove value to how roles and skills will evolve.

SEO is becoming more strategic

The panel agreed that SEO is moving beyond being a standalone channel and becoming more closely tied to brand, PR, content, paid media and wider business goals. SEOs will increasingly act as strategic advisers, helping brands understand how they're discovered, represented and perceived across both traditional search and AI platforms.

Rethinking KPIs

Rankings, clicks, impressions and organic traffic still have a role, but are they enough on their own? Emerging metrics could become far more important:

Share of answers
AI citations
Brand sentiment
Brand accuracy, i.e. how accurately AI represents your products, values and expertise

That said, everyone agreed that revenue and business impact should remain the ultimate measures of success, and that AI search measurement is still maturing, with real challenges around data reliability and attribution.

What to stop doing now

The panel challenged the industry to stop working in silos, chasing vanity metrics like Domain Authority and relying on short-term tactics designed to game algorithms. There was some healthy debate around routine technical tasks like metadata optimisation. One panellist felt many of these are already becoming obsolete, while others argued technical SEO could become even more important as AI agents and new technologies add complexity.

The future of SEO talent

AI gives junior SEOs the chance to learn faster, automate repetitive tasks and explore different specialisms. But it also risks them missing out on the hands-on experience needed to really understand search. The fundamentals still matter: how search engines crawl, index and rank content, and how LLMs work, including where they get things wrong.

When it comes to hiring, the panel looked for curiosity, critical thinking, independent research, adaptability, proactivity and communication. One panellist shared how they'd hired a candidate who gave an incorrect technical answer in their interview but showed strong reasoning along the way. Technical knowledge can be taught; the ability to question, investigate and explain your thinking is much harder to find.

Key takeaway: SEO isn't disappearing, it's evolving into a more strategic, integrated and technically complex discipline.

Our Final Thoughts From BrightonSEO October 2026

This was a day filled with knowledge, fun, great food and so much more. One of my highlights was the networking. It was great to catch up with familiar faces and become familiar with plenty of new ones.

The activities throughout the day kept things engaging and fun, too. We had caricatures drawn, joined raffles, built Lego sculptures and more. It's not every conference where you leave with a sketch of yourself alongside a notebook full of takeaways!

On the SEO side, a few clear themes came through:

The fundamentals matter more than ever. If AI can't crawl, render or understand your content, it can't recommend you
Product pages and product feeds are now data sources, and consistency across every system is non-negotiable
AI agents are coming to ecommerce, but the brands that prepare now will benefit today, not just in the future
SEO is becoming a more strategic, integrated discipline, and the way we measure success needs to evolve with it

We’ve come back with fresh ideas, new perspectives and plenty of learnings to take back to the office.

BrightonSEO remains our go-to SEO conference, and we'll definitely be back next time. If you went, I hope you had as good a day as we did!