A fortnight of measurement, on a brand that does not exist
Last updated
The sample
Attempts are answers plus no-answer replies plus failures. Nothing is dropped quietly.
Coverage and method, per engine
- ChatGPT · model not disclosed
- A vendor drives the logged-out chatgpt.com interface; the model that answered is not disclosed.
- Claude · model claude-haiku-4-5-20251001
- Asked through Anthropic's API with web search enabled; the model id is recorded on every answer.
- Perplexity · model reported per answer
- Perplexity runs its own search and reports which model answered; that id is stored rather than a label.
Where this brand stands
Changes are against 18–31 August 2026.
Silverline Freight was named in 41 of 336 successful answers
- Visibility
- 12% +0.8 ppAnswers naming you, over successful answers
- Share of voice
- 7% +0.4 ppYour mentions over all tracked brands' mentions
- Cited in
- 3% +0.3 ppAnswers whose sources include one of your pages
- Sentiment
- 61 −2 pointsHow favourably you are described when named
- Position
- #4.2 +0.3Average place among the brands an answer names
Against the competition
Every tracked brand that appeared in a successful answer. Visibility is that brand's mentions over 336 successful answers.
A competitor being ahead is an observation about these 336 answers. It is not a ranking, and no engine publishes one.
Named, cited, recommended
Three different observations, three different fixes. A single blended score would hide which one is the problem.
- Mentioned41 of 336 answers
- Your name appears somewhere in the answer text, in any tone. If this is low, inspect the questions where competitors appear and you do not.
- Cited9 of 336 answers
- One of your pages appears in the answer’s citations. If this is low while mentions are healthy, inspect which other sources the answers cite. The absence of a citation does not prove that your page was never read.
- Recommended6 of the 41 mentions
- The engine puts you forward as the choice, rather than listing you among others. If this is low while mentions are healthy, you are in the room and not the answer.
How engines describe it
How the 41 mentions were classified, with two of the sentences behind them.
- Recommended615%
- Listed2254%
- Mentioned1229%
- Cautioned12%
“If you are shipping into Dublin specifically, Silverline Freight is a reasonable first call — they publish their port coverage and handle the paperwork in-house.”
“Smaller regional forwarders such as Silverline Freight may not have the in-house customs capability for pre-arrival filing; confirm before booking.”
The questions it loses
Prompts where competitors are present and this brand is mostly absent. The count is answers in the window, not people.
Scroll across to see every column.
| Prompt | Answers | Named you | Most named |
|---|---|---|---|
| Who can handle ocean import customs clearance from India into Ireland? | 17 | 2 | Northwind Logistics (15) |
| Best freight forwarder for pre-arrival customs filing in Dublin | 18 | 0 | Northwind Logistics (16) |
| Who handles bonded warehousing near Dublin Port? | 16 | 1 | Baltic Freight Group (12) |
| Freight forwarder that files import declarations in-house, Ireland | 18 | 3 | Northwind Logistics (14) |
| Cheapest LCL shipping from Mumbai to Dublin | 17 | 6 | Meridian Cargo (11) |
One answer, in full
Every rate in this report is a count of answers like this one. Each one opens from the number it belongs to.
“Who can handle ocean import customs clearance from India into Ireland?”
Model API · claude-haiku-4-5-20251001 · country IE requested through the provider’s approximate-location setting
For imports arriving by sea from India, the practical requirement is pre-arrival filing before the vessel berths. Northwind Logistics and Baltic Freight Group both do this in-house and publish the document list they need from you. Northwind also handles the customs entry itself, so there is one party to chase if the entry is queried.
You were not named in this answer, and none of its three sources is your site. Across this prompt: 17 answers in the window, you named in 2, cited in 0; Northwind named in 15 and cited in 11.
Cited instead
The domains that appeared in the 295 answers that never named this brand.
Your own site appearing here means the answer cited your page without naming your brand. Inspect the answer before deciding what, if anything, to change.
Evidence and hypothesis
The left column illustrates findings from a page check. The right column is the hypothesis they could support. These example pages and findings are invented.
What was inspected
Illustrative page check dated 12 September 2026, 14:20 UTC. The example domains are fictional.
- Page Audit on silverline-freight.example/services: 200 OK in 380 ms, 84 KB of HTML inspected, canonical present, meta description present, no structured data found.
- robots.txt was found and allows GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and bingbot on that path, so these robots.txt rules show no block on that path; other access restrictions can still exist.
- The Content Optimizer read the same page and found no occurrence of “pre-arrival”, “customs entry” or “import declaration” in the downloaded HTML.
- Gap Analysis checked northwind.example/customs-clearance: it names the service, lists the documents required before arrival, and does not mention Silverline Freight.
- Across the window your pages were cited in 9 answers, four of them in answers that never named you; none of those citations was the services page.
What that suggests — and does not
A hypothesis. The next measurement is allowed to disprove it.
- The answers to these five prompts cite pages that state the pre-arrival filing service plainly. Your site has no such page, which is one plausible reason it is not among the sources.
- Nothing here shows what any engine retrieved, ranked or preferred, and the page as it stands today is not proof of what an engine read last week.
- Being cited is not the only route into an answer — you were named twice on this prompt without being cited at all — so publishing the page may change nothing, and the retest is what says which way it went.
The work, and the retest
One prioritized action, and what the same questions returned after the change went live.
Five tracked prompts turn on pre-arrival filing. Engines answering them cite pages that state the service and its paperwork; the site currently says neither. State the service, the documents required before arrival, the ports covered and who files the entry.
Published /services/pre-arrival-customs-filing and linked it from the services index.
Same 5 prompts, same 3 engines, same requested country, same recorded model identifiers, same collection method. 84 comparable answers before, 82 after.
The consumer-interface model is undisclosed, so its identity cannot be independently compared. This is an observation about two samples, not a demonstration that the page caused the change. Provider behavior can change without a reported model change, competitors may publish too, and the same question asked twice does not always answer the same way. Where coverage does not match on both sides, the follow-up reports insufficient comparable data rather than a number.
What this report does not say
- That the published page caused the change. Two samples moved; the engines, the competitors and the web moved with them.
- That these are the questions this brand's buyers asked. They are twenty questions somebody chose, asked daily from one country.
- That being cited is required to be recommended, or that either is within anybody's control.
- Anything about a real company. Every name, domain, answer and number in this report is invented.
The free check looks at one page and finds what is getting in the way — no account, no card. A report like the one above comes from the trial, where your own questions are asked and monitored: 15 prompts on 1 project across 3 engines for 7 days, card required, first charge on day 8.