Common mistakes maritime marketing teams make on llms.txt

We audit a lot of llms.txt files. Most of them are not catastrophically broken; they are just half-thought-through. The marketing team set the file up after reading a blog post, listed twelve URLs, copied a corporate boilerplate as the description and never touched it again. The file does some good but leaves real value on the table.

Here are the mistakes that come up repeatedly on maritime corporate sites, and the fixes worth doing.

Mistake 1: Treating it as a sitemap dump

The wrong instinct is to list every page on the site. The right instinct is to curate a tight list of the pages that genuinely represent your business: core services, case studies, key capability pages, primary contact information.

A maritime ship manager does not need their llms.txt to point at their privacy policy, their cookie banner page or their generic “About us” landing. They need it to point at their tanker management page, their crew management page, their two strongest case studies and their certifications page. Eight to fifteen URLs is plenty for most maritime corporate sites. Forty is too many.

Mistake 2: Marketing prose where structured facts belong

The first paragraph of llms.txt is the highest-value real estate on the file. Many maritime sites waste it on a sentence like “We are a leading global provider of ship management services committed to operational excellence.” That sentence contributes nothing to the parser.

The fix is to replace it with a dense factual paragraph. Founding year, fleet size, vessel types, certifications, office locations. Six concrete facts in two sentences. The parser walks away knowing what your company is. The marketing prose can stay on your homepage where humans read it.

Mistake 3: Outdated facts

A surprising fraction of llms.txt files we see contain numbers that were true two years ago. The fleet size has grown, an office has opened or closed, a certification has been added and nobody updated the file. The discrepancy between llms.txt and your live website is exactly the kind of inconsistency LLMs are sensitive to. The model hedges, downgrades the source or picks a competitor.

The fix is to schedule a quarterly review of llms.txt alongside your other AI SEO maintenance. Five minutes once a quarter. A surprising number of teams have never edited the file since they first published it.

Mistake 4: Aspirational claims

We have seen llms.txt files claiming “global presence” for companies with one office. We have seen “tanker, bulker, gas carrier, container, offshore and specialised vessel management” from companies that genuinely manage two of those segments. We have seen “AI-powered fleet analytics” listed as a service when no such product exists.

The aspirational version of your company is the one your marketing team likes to write. The version the LLM cites is the one that matches your real capabilities and the third-party coverage of your real capabilities. Where they diverge, the model trusts the third parties.

Write llms.txt the way you would draft a regulatory filing. Specific, accurate, defensible.

Mistake 5: Linking to outdated or missing pages

Several llms.txt files we have audited point at pages that have been moved, redirected or deleted. The crawler hits a 404 or a 301 chain and either gives up or follows to a page that does not match the description in llms.txt.

Run a link-checker against your llms.txt monthly. Fix or remove dead links. Update descriptions when a page has been substantially rewritten.

Mistake 6: No structured factual block

The best llms.txt files include a short factual block, often as a quoted blockquote near the top:

> Acme Ship Management. Founded 1994 in Limassol.
> 84 vessels under technical management as of Q1 2026.
> Tanker, bulker and gas carrier vessels above 5,000 GT.
> ISM DOC certified. Offices in Limassol, Singapore and Hamburg.

This block alone, parsed by an LLM, lets the model answer most basic questions about your company without needing to retrieve any other pages. We see clear evidence in retrieval logs that this block gets quoted directly in citations more than any other section.

Mistake 7: Not updating after corporate changes

A merger, a new office, a divestment, a rebrand. The llms.txt is rarely on the change-management checklist for these events. Three months later the LLM is still citing your old fleet size or your old office addresses.

Add llms.txt to the checklist for any material corporate change. The cost is five minutes; the cost of leaving it stale is a quarter of misleading citations.

Mistake 8: Including the file but not linking to it from robots.txt

Some implementations of AI crawler discovery look at robots.txt for hints about where llms.txt lives. The convention is loose, but at minimum your robots.txt should explicitly allow the path to llms.txt and llms-full.txt rather than relying on default behaviour.

What good looks like

Short. Factual. Updated quarterly. Eight to fifteen curated URLs. Description block dense with verifiable facts. Aligned with what the rest of your website and your trade press coverage say. Maintained as part of routine marketing operations rather than as a one-time setup.

The brands that do this consistently are easier to cite, easier to verify and easier to trust. The model picks up on the consistency and rewards it. The brands that treat llms.txt as a tick-box are leaving the marginal advantage to whoever takes it more seriously.