What We Learned At SEMrush Training | BrisTechTonic

At the end of October we traded the Bristol drizzle for Amsterdam's canals to attend Semrush Spotlight, a two-day deep dive into what SEO and brand visibility look like in an AI-first world. It was hands-on: laptops open, dashboards live, every framework applied to our own data as we went.

When someone asks an AI tool about your brand, it looks well beyond your website: review platforms and comparison sites, developer forums and support communities, support docs and help centres, Reddit threads and social spaces, news sites and PR coverage, and your own pricing pages, FAQs and product content. If your strategy stops at blog plus landing pages, you are invisible to a huge chunk of the signals AI systems pull from.

Reddit, communities and the new default

A striking graph showed Reddit's growth versus Wikipedia, with Reddit now effectively the second-largest site on the web in that data set. Another chart compared outbound clicks on SERPs with and without AI Overviews: when AI summaries appear, a noticeable portion of clicks shift towards community and video platforms. For brands, that means treating Reddit, YouTube, TikTok and niche forums as core parts of search strategy.

AI does not just ask who you are, it asks what everyone else thinks about you. Semrush's practical tips: be present on the right review sites, join community conversations on Reddit, Quora and specialist forums, and engineer social proof through case studies, testimonials and user-generated content. Listicles like 'The 10 Best X Tools for Y' dominated results and got cited inside AI responses.

You cannot fake authority. Make your site technically accessible with clean HTML, structured data and readable code. Fix public data such as Wikipedia entries and knowledge-graph profiles, because incorrect there means incorrect everywhere. Be transparent, since hidden pricing and vague FAQs push users to Reddit and LinkedIn, and that speculation becomes training data.

1. 01Research: listen to how customers actually talk, in calls, live chats, surveys and Reddit threads.
2. 02Architecture: design navigation that mirrors those mental models.
3. 03Content: enable all teams to contribute within that architecture.
4. 04Alignment: let that structure become the source of truth.
5. 05Visibility: reap the rewards as AI systems favour your depth, clarity and authority.

• /Lead with the answer, then expand.
• /Mirror the heading, for example start a 'How do I' section with 'To... you should...'.
• /Keep it human and concise.
• /Make it quotable with unique stats, expert opinions or proprietary insights.
• /Strengthen entity signals by naming products, brands and people explicitly.
• /Use question-based headings that map to what users type.
• /Cluster related prompts on the same page to show topical depth.

Semrush built this into an internal guide on two principles: freshness (content is regularly updated) and consistency (terminology and structure stay predictable).

SEO vs AEO: 90% overlap, 10% new rules

SEO focuses on crawling, SERP features, keywords, E-E-A-T, clicks and user signals. AEO focuses on semantic HTML, mentions (not just links), long prompts, personalisation, influence and consensus. The overlap is technical accessibility, content quality and authority signals. Strong SEO foundations get you most of the way there.

Good SEO KPI reports should be actionable, easy to understand, insight-led, results-oriented and contextual. One slide showed impressions soaring while meaningful outcomes stagnated: in an AI world, impressions can lie. What matters is whether you appear in AI summaries and whether that presence drives conversions.

1. 01Topic research.
2. 02Briefing and editorial review.
3. 03Creation with in-house writers or external partners.
4. 04Translation and localisation, with transcreation for key markets.
5. 05Asset sourcing and upload, including visuals, structured data and internal links.
6. 06Impact analysis and maintenance. Scale comes from process, not from shouting for more content.

One speaker framed the shift as the biggest technology revolution since Google launched in 1998. Traditional search still dominates desktop visits, but visits to AI tools are climbing month after month. Only a minority of brands that show up in AI answers are also cited as sources, a big chunk of Google's results now appear inside AI answers, and visitors arriving from AI surfaces often show significantly higher conversion rates. When AI Overviews appear, users are less likely to click a traditional organic result.

Top cited domains included Wikipedia, big review and marketplace sites, specialist content brands, and major communities. This reinforced that digital PR and partnerships should target domains LLMs already love citing. Overlap between Google AI Mode, AI Overviews, ChatGPT citations, Perplexity citations and classic top-10 results was high at the domain level, but not perfect.

1. 01Presence (citation gap): where are competitors cited and you are not?
2. 02Priority (platform priority): where do you actually need to show up?
3. 03Prompts (reasoning crawls): do you monitor the right prompts?
4. 04Perception (web footprint): are you perceived as an authority?

Prioritise platform prioritisation first (high impact, lower effort), then citation gap (high impact, higher effort), then web footprint (foundational), then reasoning crawls (advanced).

AI visibility is an outcome, not a hack. Classic SEO still matters a lot. Communities, reviews and ecosystems are now core search channels. We need new KPIs beyond impressions, tracking citations, LLM sentiment, AI-driven conversions and share of voice. And there is a clear playbook: the 4P framework, citation gap analyses and reasoning-crawl monitoring.