Optimising service pages for AI summarisation
How to structure maritime service pages so AI tools summarise them accurately and cite them with confidence, with concrete patterns and a checklist.
A service page is the workhorse of maritime AI SEO. Most citations originate from service pages, most buyer journeys go through them and most of the structural problems that suppress AI visibility live on them. Optimising for AI summarisation means writing service pages that an LLM can read, summarise accurately and cite without hedging.
The principle
LLMs that cite a service page are not reading every word. They are extracting the parts they can verify and quote, then constructing an answer around those parts. A page optimised for AI summarisation makes that extraction as easy as possible without sacrificing the human-facing quality of the page.
The constraints are mutually reinforcing. The same patterns that help an LLM extract cleanly also help a human buyer skim the page in thirty seconds and decide whether to keep reading. The marketing instinct that resists structural specificity is usually wrong on both fronts.
The pattern
Open with a factual summary
The first paragraph should be a dense factual block. What the service is, who it is for, the scale of your operation in this service, the specific certifications and capabilities that distinguish you. Six to ten facts in two or three sentences.
A typical opening for a tanker management page:
We provide third-party technical management for tanker fleets above 5,000 GT. Our 32 tanker vessels under management as of Q1 2026 include LR1, LR2, Aframax and Suezmax classes, registered under Cyprus, Marshall Islands and Liberia flags. ISM DOC certified for tanker operations. Offices in Limassol, Singapore and Hamburg.
This paragraph alone, extracted and cited, answers most category-defining queries about your tanker management service.
Use H2 for topical sections
Every distinct topic gets a heading. “What we manage”, “Certifications and class”, “How we operate”, “Reporting and transparency”, “Crewing”, “Onboarding process”, “Pricing model where applicable”, “Case examples”, “FAQ”.
The headings are signposts for both readers and parsers. They also help with internal linking and with rich-result eligibility on classical search.
Use lists for parallel facts
When you are stating five vessel types, eight ports of call or twelve certifications, use a list. Lists are easier for parsers to extract intact, easier for readers to scan and easier to update when a fact changes.
**Vessel types under management**:
- LR1 product tankers (15,000 to 45,000 DWT)
- LR2 product tankers (75,000 to 120,000 DWT)
- Aframax crude tankers (80,000 to 120,000 DWT)
- Suezmax crude tankers (120,000 to 200,000 DWT)
The list is more useful than the same information embedded in prose.
Include an FAQ block
Five buyer-realistic questions with concrete answers. The FAQ block is the single most cited section on most service pages, because LLMs prefer to quote question-shaped content. Include FAQPage schema.
Include a specific call to action
Not every page needs to push an enquiry form. But every page should make clear how a buyer would actually engage. “Request a technical management proposal”, “Speak to our Singapore office”, “Read our crew management methodology”. The CTA matters less for SEO and more for ensuring the buyer who has decided to act has a clear next step.
What to leave out
Hero copy
The big landing-page hero with a stock photograph of a vessel and a one-line aspirational statement is fine for a homepage. It hurts a service page. The hero pushes the substantive content below the fold and gives the parser nothing to extract on first scan.
Generic trust signals
“Trusted by leading shipowners” without naming any of them. “ISO certified” without specifying the standard. “Operational excellence” without supporting evidence. The parser strips these out; the human buyer reads them as filler.
Repetitive marketing claims
Saying “tailored solutions” three times does not strengthen the claim. Saying it once with the specifics that demonstrate the tailoring does.
Auto-generated CTA modules at the bottom of every page
These often interfere with structured data and add noise to the parser’s view. If you have a CTA module, make it clean and lightweight.
A pre-publish checklist
Before publishing or republishing a service page:
- Does the first paragraph contain at least six extractable facts?
- Is every claim either backed by a specific number, certification, location or named example?
- Are H2 headings present every 200 to 300 words?
- Is there an FAQ block with at least three buyer-realistic questions?
- Is Service schema present and validated?
- Is FAQPage schema present and validated?
- Are all internal links live and pointing at canonical URLs?
- Is the page free of stale facts (old fleet sizes, old office addresses, retired services)?
- Does the page match what your llms.txt says about this service?
- Does the page match what your trade press coverage says about this service?
A page that passes all ten is meaningfully more likely to be cited than a page that passes five. The work is methodical and unglamorous. It is also the highest-impact AI SEO improvement most maritime marketing teams can make this quarter.
Frequently asked questions
How long should an AI-optimised service page be?
Should every section have its own heading?
Do videos and images on service pages help or hurt AI summarisation?
More on AI SEO
-
AI SEO
The 12-month AI SEO roadmap for a mid-sized maritime company
A practical month-by-month AI SEO plan for a mid-sized maritime company, covering audit, structural fixes, content, authority building and measurement.
By Paul Clapp -
AI SEO
Auditing AI visibility: a practical methodology for maritime
How we run a structured AI visibility audit for maritime clients, what we test, how we score and how to make the results actionable for a marketing team.
By Paul Clapp -
AI SEO
AI search and the maritime regulatory question (compliance content)
How AI search treats maritime regulatory content (EEXI, CII, MARPOL, SOLAS, ISM) and how to publish compliance content that gets cited responsibly.
By Paul Clapp