Someone asks an assistant which agents handle a particular trade into Hamburg. Three names come back. Whether yours is among them has less to do with how long you have been in the business than with whether that trade is written down anywhere a machine can read it.
This layer sits above ordinary results rather than replacing them. A site can hold a decent position for a lane term and still be absent from every generated answer on the same subject. Both are forms of presence; neither compensates for the other, and the second is not something a good relationship with a carrier can fix.
What makes this awkward in shipping is that the industry is built on relationships and reputation, neither of which leaves a trace a machine can use. Forty years of reliable service produce goodwill among the people who already know you and precisely nothing for the buyer who does not.
Why longevity does not carry into a generated answer
A results page offers ten links and lets a person choose. A generated answer performs the choice and returns two or three names with a sentence apiece. The sentence is the interesting part — it is a description of your business, assembled from whatever you have published.
Selection favours the least ambiguous rather than the most polished. Anything doing the choosing needs facts it can attach to the question: which trade, which service, which authorisation, which terminals. A page describing dependability and long experience offers none of them, however accurate both claims may be.
What has to be on the page for you to be named
Five facts every desk knows and no page carries
Each of them is asked in the first conversation. None of them is usually published.
- Trades and corridors, listed individually. Not global coverage but the routings where rates, partners and actual experience exist. Claiming everything results in being retrieved for nothing — the single most reliable rule in this whole area.
- Cargo types handled, and those refused. Groupage, part loads, full containers, project cargo — and explicitly what is turned away. The refusal does more work than the list, because it makes the list believable.
- Authorisations with their scope. Designation, extent and issuing authority written into the text. Held as a downloadable certificate, they might as well not exist.
- Response commitments with a unit attached. A quotation within four working hours can be verified and compared against a rival. Promptness cannot be either.
- Where the desk sits and in which languages it answers. In a trade where a query arrives at two in the morning local time, this decides more enquiries than any service description.
These five share one property: publishing them costs almost no effort and a certain amount of discomfort. Writing down which cargo you refuse, or where your lower limit sits, forecloses conversations the commercial side would rather keep open. That reluctance — not forgetfulness — is why they are missing. The trade is nonetheless a good one: fewer enquiries, and a markedly higher proportion of them worth answering.
Why those five statements stay unpublished
Each objection is reasonable and each has a narrower answer than it first appears.
- "We do not want to be pinned down." You are pinned down in every first conversation anyway, one enquiry at a time and under time pressure. Published once, the same statement never has to be repeated.
- "Competitors will read it." Competitors have known your corridors and cargo types for years, from carriers and from shared customers. The people who do not know are the shippers who have never dealt with you.
- "Our trades change." Then put a year beside them. A statement with a date on it is usable; one without is not, and correcting an outdated one takes five minutes.
- "We will lose enquiries." You will lose the ones that would have been declined after two calls. Volume falls, the usable share rises — which in a desk already at capacity is the better position.
Getting the material moving takes a single session with the commercial side and the operations desk present together. Split across two conversations it never concludes: each side raises half an objection, and neither half can be answered by the other side alone. The page recommendations in the workspace are useful in that meeting, since they name specific pages already published rather than proposing an abstract programme.
What can be examined without a tally
With nothing to count, all that is left is procedure: ask the same set of questions on a schedule, retain every reply, and look for what repeats. The Semalt workspace provides six screens for exactly this, showing the field as a model portrays it instead of as a results list ranks it.
- Where the Market Circle puts you. One of three bands: leading, middle or niche. It measures how far you sit from whoever gets named first, and says nothing about how well anybody actually performs.
- The model's own account of your domain. How it reads your positioning, its estimate of your traffic, the openings it identifies. Setting this beside your own company description is the single most instructive part of the exercise.
- The questions themselves, grouped by purpose. Full questions rather than terms, separated according to whether somebody is orienting, comparing, or about to award.
- Existing pages with visible leverage. Pages already published where more detail or better internal linking would repay the effort disproportionately.
- Subjects rivals have completed and you have not. Fields covered end to end elsewhere with no counterpart on your site — here, almost invariably customs and dangerous goods rather than the service overview.
- One aggregate number for the AI-assisted layer. A single value covering everything — and the exact point at which the two caveats below take hold.
For an operator the second surface repays attention first. Where a model describes the business as a general logistics provider while the margin comes from two specialised corridors, that is not a fault in the model. It is a fair summary of a site on which those two corridors are nowhere named.
"Groupage and part loads to the Far East from 1.5 loading metres, no Class 1 dangerous goods"
Four facts a selection can test against a specific question. Every one of them comes from the operations desk.
- Answers a cargo-type question
- Answers a minimum-quantity condition
"Your reliable partner for worldwide logistics solutions"
Accurate, unfalsifiable, and word for word what several hundred other sites at this port already say.
- Matches no specific question
- Distinguishes you from nobody
What the consolidated figure will not bear
The figure stays useful as long as its provenance stays in view. It is produced by repetition: questions put, replies stored, appearances counted. What that yields is a sample and not a census.
Taken together those three caveats sound like a reason to leave the subject alone. They are the opposite: they are what allows it to be discussed at all. State the limits and the number holds up in a management meeting. Leave them out and it holds until the first person asks about the method — which, among people who verify rates and documents every working day, is the next thing said.
The English question and the German one
This market has a feature that shows more sharply in this layer than in ordinary search: a large share of the questions arrive in English, from shippers, agents and carrier representatives who do not know the German procedural framework and need it explained alongside the answer.
The question about the procedure
How something works at this port. The asker wants understanding first and a provider second.
- Explanatory pages get drawn upon
- Service overviews essentially never
The question about the provider
Who handles this trade out of Hamburg. The asker knows the framework and wants names.
- Attributability is what decides it
- Name the trade and the authorisation
The workable division follows from that: the English material takes the explanatory role, the German material the provider role. Running both as translations of one another gives away most of the available benefit in this layer — and produces two pages arguing identically for the same terms, which helps neither.
One organisational note before the practical part. Everything described here sits on a boundary most operators have never drawn: the facts belong to the operations desk, the website belongs to whoever handles marketing, and there is usually no established route between the two. Firms that build that route once — a standing arrangement to turn desk knowledge into published pages — solve the problem permanently. Firms that treat each page as a separate negotiation publish two pages a year, which is not enough to matter.
Saying that plainly is worth it because the analysis will report the same gap every quarter until the route exists. Reading the same absence four times is not progress; it is an expensive way of confirming what the first quarter already showed. The workspace will keep flagging it, and a flag only helps somebody who is able to act on it.
Real enquiries, verbatim
Everything starts with what actually arrives. It reads nothing like a search term — longer, hedged, conditional, frequently one sentence without a full stop. "Can you still collect on Friday if the export declaration only comes through Thursday evening?" is not something anybody types into a search box, and it is exactly what sits in the inbox.
| Source | What it contains | Effort |
|---|---|---|
| Notes from the operations desk | the questions asked before award, in the asker's own words | five minutes per call |
| The commercial inbox | conditions stated plainly: trade, date, cargo type, customs status | an hour per quarter |
| Follow-up questions after a quotation | precisely where your published material falls short | none — they already exist |
| Tenders you did not win | almost always a fact that could have been public | unpleasant reading, high yield |
The third row is free and therefore the most neglected. Where every quotation is followed by three clarifications, those three clarifications name three gaps in what you have published — phrased by the person who needed the answer rather than by the person who wrote the page.
A rhythm that fits an operational business
The evidence here is thinner than for conventional metrics, which rules out anything weekly. Three months has proved the right interval: check more often and you are watching noise, check less often and a shift becomes apparent only once it cannot be recovered.
| When | What happens | Who |
|---|---|---|
| Week 1 | Collect twenty arriving enquiries word for word | operations and commercial |
| Week 2 | Check which of them already has a page | whoever maintains the site |
| Week 3 | Read the model's account and the competitor coverage | half an hour |
| Weeks 4–8 | Write two or three trade or procedure pages | the desk, not marketing |
| Quarter end | Set placement and the consolidated figure against the previous quarter | deliberately nobody senior |
Week three carries a caveat worth repeating. A model's description of a business rarely matches what its owners would write. The reflex is to call it wrong; the more productive move is to ask what it could have been derived from, and the source is invariably the published text. Read that way it is not an error but accurate feedback. Because My SEO keeps this analysis next to the familiar figures, a text change and any subsequent movement can be read off one timeline.
What comes out at the end
Before committing anybody's time, it is worth being clear about the output. Not a figure for the management meeting. A site that states which trades are actually served and under what conditions — and therefore performs better in both layers at once.
The second effect usually exceeds the first and is rarely mentioned: every question answered publicly is a question that stops arriving by telephone. In a business whose operations desk is already at capacity, that counts for more than any placement. What has to happen first is separate — whether the pages can be indexed at all is covered under technical SEO, which German-language terms are worth pursuing under keyword research, and the split between the English and German material in our content strategy.
See how your domain is described in the dashboard
Questions from the desk
Can we measure how often we are named?
There is nothing to read off. No supplier can query a counter, because none exists. What can be done instead is to run an unchanging list of questions at fixed intervals, file every response, and work out from the accumulation which way things are moving. That conclusion is fit for its purpose and no further: it rests on a set of trials, and pushing it into the role of a visitor statistic will produce a number that cannot be defended.
Should we really publish which cargo we refuse?
Yes, and of everything on the list it has the best ratio of effort to effect. A stated refusal makes the rest of the list credible, keeps away enquiries that would have been declined anyway, and gives a selection something concrete to match. The commercial side objects to this consistently and changes position within a quarter, once the mix of incoming enquiries shifts.
We have been in this trade for forty years. Does that count for nothing?
It counts for a great deal with everybody who already knows you, and for nothing at all with a buyer who does not. Longevity becomes usable the moment it is attached to something specific: since which year you have handled this particular corridor, which authorisation you have held since when, which terminal you have worked at for how long. Stated that way it is evidence; stated as experience it is an adjective.
Will we lose traffic if assistants answer directly?
Some, and specifically on questions that only ever wanted a fact. Somebody checking how a customs procedure works in principle opens the underlying page less often than before. Where an award is in prospect it works differently: an operator that gets named then gets checked, and that requires the page. What is affected is material whose only purpose was a short answer.
How often should the consolidated figure be looked at?
Four times a year is enough. More frequent inspection shows the trials wobbling rather than the market moving. The figure means more in relation to the familiar numbers than on its own: a shifting placement against unchanged traffic and positions is worth investigating, whereas both moving together nearly always has one shared cause.