What changed in search
More and more people ask ChatGPT or see Google's AI answer above the results. The model assembles its answer from websites and cites sources — a new shelf of visibility exists: being the AI's source. It doesn't replace SEO; it extends it. AI answers mostly draw on sites search already treats as authoritative.
What language models love
- Direct answers to questions. A "How much does X cost?" block with a concrete number in the first sentence is perfect quoting material. That's why FAQs with FAQPage markup are the best investment.
- Tables and structure. Comparisons, tiers, steps — LLMs happily retell tables because the facts are already laid out.
- Numbers and dates. "In 2026 a landing page costs from £800" gets cited better than "prices are affordable". Freshness counts too: the year in the title works.
- Consistency. If one page says £800 and another £1,000, the model either picks at random or doesn't cite you at all. One fact — one version everywhere.
- Authors and expertise. Bylined articles with real names and roles are the E-E-A-T both Google and the models read.
What to do on your site this month
- Add FAQ blocks with schema markup to key pages — with direct answers
- Open articles with a "60-second" summary block carrying the key numbers
- Reconcile facts across pages: prices, timelines, case numbers
- Byline articles with real authors and roles
- Don't block AI bots in robots.txt if you want answer visibility
How to tell it's working
There's no "Search Console for ChatGPT" yet. What you can do: watch referral traffic from chatgpt.com and perplexity.ai in GA4, ask new clients "how did you hear about us", and periodically quiz the models about your niche to see if you come up. It's an early market: whoever structures content now collects the citations for years.
We build this into every site by default: FAQ markup, structured data, consistent facts — part of the standard described on the SEO services page.