The four LLM citation patterns we see across maritime categories

We have run AI visibility audits across enough maritime categories now that the patterns repeat. Ship management, port agency, marine equipment, classification advisory, fleet management software, marine insurance, offshore services. The category specifics differ, but the citation behaviour clusters into four shapes. Knowing which pattern your category fits tells you what kind of GEO work will move the needle.

Pattern 1: The concentrated incumbent set

A small number of brands (typically three to five) dominate citations across nearly every relevant prompt. Newer or smaller players appear in maybe 5% of responses, often with caveats.

We see this in classification societies (DNV, Lloyd’s Register, ABS, BV, RINA, ClassNK), in major IACS-class topics and in container line rankings. The training data and trade press coverage is heavily concentrated on the same few brands, and the models reflect that.

What works in this pattern. If you are an incumbent, defend the position by keeping your structural data clean and your trade press cadence steady. If you are a challenger, you cannot enter the head citation set quickly. Instead, target the long-tail prompts: niche vessel types, specific regions, unusual compliance scenarios. Build authority there first.

Pattern 2: The fragmented mid-tier

Citations spread across fifteen to thirty brands, with no clear leaders. The model often hedges, offering caveats like “consider these among others” rather than confident lists.

We see this in independent ship management for mid-sized fleets, in regional port agency, in marine surveying. The category has many credible providers and the trade press does not consistently rank them.

What works in this pattern. Structural quality and brand authority signals move you up fast. Because no incumbent has locked in the citations, a year of disciplined GEO work can put a mid-sized brand into the top five citations across most prompts. This is the pattern with the highest reward for early movers.

Pattern 3: The geography-keyed cluster

Citations are essentially the same brands but reordered depending on the geography mentioned in the prompt. Ask about Singapore and you get one set; ask about Piraeus and you get another; ask about Hamburg and you get a third.

We see this in port services, bunker suppliers, ship chandlers and regional surveyors. The geography is the dominant retrieval feature.

What works in this pattern. Region-specific service pages with explicit location entities, local trade press coverage and proper LocalBusiness schema. A ship chandler with one page that says “global presence” will lose to one with five regional pages that each name the port, the local fleet types served and the specific certifications relevant to that geography.

Pattern 4: The single-source dominance

One source, often a trade publication or industry directory, dominates citations to such an extent that the model essentially recites its rankings. We have seen this in topics where Drewry’s analyses, Clarksons’ reports or specific TradeWinds rankings have been ingested heavily.

What works in this pattern. You cannot work around the dominant source through your own content. You have to be in that source. For Drewry-influenced topics, that means commissioning Drewry research or providing data they use. For TradeWinds-dominant topics, it means earning a feature. The work is harder but the payoff is enormous: a single placement in the right source can shift your citation rate across hundreds of prompts overnight.

How to identify which pattern your category fits

Run a small audit: ten prompts spanning category-defining, buyer-style and brand-name probes. Look at three things.

Concentration. How many distinct brands appear across all responses? Under ten suggests concentrated incumbent. Twenty to forty suggests fragmented mid-tier. Geography-driven variance you can spot directly. Single-source dominance shows up as identical brand lists with the same source cited repeatedly.

Stability. Do the same brands appear when you reword the prompt? Stable lists indicate concentrated patterns; shifting lists indicate fragmentation.

Source citation. Which sources does the model cite? If five different sources appear, you are in a fragmented mid-tier. If one source appears across nearly every response, you have single-source dominance.

The pattern dictates the strategy. Concentrated incumbents need defensive consistency. Fragmented mid-tier rewards aggressive structural and content work. Geography-keyed categories reward regional specificity. Single-source-dominant categories require source-level intervention. Build your roadmap around your pattern, not around someone else’s.