Startseite Keyword-Recherche OnPage-Optimierung Linkaufbau Lokales SEO Content-Strategie Technisches SEO Über uns Kontakt Blog Tools

Eleven thousand parts, none of them findable: catalogue discovery for suppliers

12.09.2026 5 Min. Lesezeit
Zurück zum Blog

A supplier with eleven thousand part numbers usually has eleven thousand pages nobody can reach. The catalogue sits behind a search form: type a designation, get a result. For a customer who already knows the designation this is perfect. For everyone else — and for anything trying to read the site — the catalogue does not exist.

This is the single most expensive structural problem in industrial B2B, and it is almost never described as a problem. The catalogue works. Customers use it daily. Sales refers to it constantly. What nobody notices is that the entire product range is invisible to anyone who has not already been told what to type, which excludes exactly the buyers a company would most like to reach.

Diagnosis · What is actually reachable

The gap between published and reachable

Nobody spots this through falling visitor numbers. It shows up as a discrepancy instead: internally everyone is certain a given product page is live, yet it never surfaces against any query in the reporting. No investigation follows, since nothing raises a flag for material that was never taken in to begin with. The page is merely absent, and being absent generates nothing to notice.

What the company assumesWhat is actually the caseHow long it goes unnoticed
The catalogue is onlineReachable only through a formIndefinitely
Datasheets are publishedPublished as files nothing links toUntil somebody searches for one
Every variant has a pageVariants are generated on requestUntil a variant query is checked
The English catalogue matches the GermanHalf the English entries were never builtUsually years

The first row is the structural one. A form is an instruction to a human, not a route. If nothing on the site links to a product page, that page has no path leading to it, and the number of parts in the database is irrelevant — what counts is the number of parts something can walk to.

The test that settles it in five minutes. Take a product designation, search for it externally with the company name attached, and see whether your own page comes back. Repeat for five parts chosen at random from the catalogue. If three of the five produce nothing, the catalogue is not reachable, and every other measure is premature.

The fourth row deserves separate attention because it is specific to exporters. English catalogue entries are frequently built only for the products the sales team was actively pushing when the English site was commissioned. Everything added since exists in German alone. Nobody notices because the English site looks complete — the categories are all there, the navigation works, and only a direct comparison of entry counts reveals that one catalogue holds four thousand items and the other eleven hundred.

The commercial consequence is straightforward and rarely quantified. Products absent from the English catalogue cannot be found by export buyers, which means the products a company most wants to sell abroad may be precisely the ones invisible there. Producing the count is a database query taking minutes; acting on it is a translation project taking months. Most companies never run the query.

Routes · How pages get found

Two ways in, and only one you control

Route one

Links from pages already known

The normal mechanism. Anything one or two steps from a well-known page tends to be picked up; anything reachable only through a form is not reachable at all.

  • Depends entirely on structure
  • A form is not a link
Route two

Explicit submission

A list is handed over. Useful when the structure is deep, newly built, or when a catalogue has just been made linkable for the first time.

  • Works regardless of depth
  • Bound by fixed daily limits

For a catalogue the order is fixed: first make the pages linkable, then submit them. Submitting an address that nothing on the site links to produces a page that exists in isolation and carries no weight from anywhere. The fix is unglamorous — category listings that link down to individual products, and product pages that link across to compatible parts — and it is worth more than any submission.

Dropping route one in favour of route two would be a mistake. Work on structure keeps paying out; a filing pays out once. Where every product page carries links across to alternatives and compatible items, the whole catalogue stays easier to walk through from then on — an individual filing helps that one address on that one occasion and no further. The sensible arrangement is to repair the structure until route one carries the load, then hold route two in reserve for what structure cannot solve: a first publication, a change of naming, a move that touches several thousand addresses in one go.

Limits · Numbers to plan against

The submission ceilings, stated plainly

Throughput

How much moves per day and per batch

Whether eleven thousand parts clear in a fortnight or drag on for a third of a year comes down to these.

1 000 URLs / day / account
  • One thousand addresses a day, counted per account. Not per site. A company with a German site and an English one shares one allowance between them.
  • Ten thousand addresses in one job. Accepted at once and processed against the daily ceiling, which makes a full catalogue submission a multi-week operation by arithmetic alone.
  • Two jobs running, twenty waiting. Order matters. A large low-priority job filed first will hold a slot while the urgent one waits behind it.
  • Sitemaps nested three levels deep, one thousand per job. Ample for a catalogue split by product family and variant, provided the index files are genuinely linked.
10 000
addresses per job
2 / 20
running / queued
1 000
sitemaps per job
Consequence

What that means for eleven thousand parts

The planning error is nearly always identical.

eleven days at full rate
  • Eleven thousand parts is eleven days of allowance. At best, and only if nothing else is competing for it. In practice three to four weeks is the realistic figure.
  • Assuming the allowance multiplies per site. A company with two sites plans against two thousand a day, receives one thousand, and concludes the process is slow rather than the plan wrong.
  • Submitting the whole catalogue repeatedly. A monthly full resubmission spends the entire allowance on parts that were already fine.
  • Submitting before the catalogue is linkable. The pages get picked up, carry nothing, and rank for nothing. The work is not wasted but it is badly sequenced.
1 000
addresses per day per account
11
days for eleven thousand parts
1
hop a redirect should take

None of these ceilings is an obstacle in itself. They become obstacles when a plan assumes unlimited capacity, which is the default assumption until somebody reads the numbers. Ten minutes dividing part count by daily allowance produces a realistic schedule and removes the commonest reason a catalogue rollout appears to stall for no reason.

A trap in automated catalogue exports. A nightly job that files a submission whenever the product database changes will fill the twenty-slot queue within a day in any company with active product management. Every job is valid, every one is a duplicate, and all of them occupy capacity. Trigger the automation on changed page content, not on a database write.

It is also worth being clear about what these ceilings are not. They are not a quality signal, and staying below them earns nothing. They are a scheduling constraint of the same kind as a machine capacity or a delivery window, and they deserve the same treatment: know the throughput, sequence the work by importance, and do not queue anything that does not need to run.

Four states between submitting and the indexSubmittedsitemap or nudgeDiscoveredthe URL is knownCrawledcontent was fetchedIndexedthe page can rankIndexing Hub: 1,000 URLs per day per account · 10,000 per batch · 2 jobs at a time
Eleven thousand parts enter the first state as a single sitemap and leave the last one as a fraction of that number.
Selection · Not every part

Which parts deserve a page at all

Eleven thousand entries do not mean eleven thousand pages worth finding. A large share of any industrial catalogue consists of size variants, colour options and packaging units whose text differs by one figure. Published individually they compete with one another and dilute the family they belong to.

CategoryTreatmentReasoning
Product familiesA full page eachThis is what buyers search for
Distinct types within a familyA page each, genuinely different textDifferent application, different buyer
Size and material variantsListed on the type pageOne figure of difference is not a page
Packaging and order unitsNot published separatelyNobody searches for these

Applied to a typical catalogue this reduces eleven thousand entries to something between three and six hundred pages worth building properly — which changes the submission arithmetic from four weeks to a few days, and changes the writing problem from impossible to merely large.

One question settles nearly every case. Where the whole body of a page is reproduced at some other address on the same site, that page needs no separate route in. Asking it disposes of most catalogue arguments before they start, and the check can be wired into the export itself.
Build properly

Pages a buyer would search for

Product families and distinct types, each with specifications in the text, an application description and links to compatible parts.

  • Three to six hundred of them
  • Submit deliberately
List instead

Variants that differ by one figure

Sizes, materials, packaging units. A table row on the type page, not an address of their own.

  • Ten thousand of them
  • Never submit

Product management usually resists this, and the objection is understandable: every variant is a real product with a real order number, and leaving it without a page feels like hiding inventory. The counter-argument is measurable rather than aesthetic. Ten near-identical pages compete with one another and none of them ranks; one type page carrying a complete variant table ranks and answers the question in full. Nothing is hidden — the variant is on the page, in the table, with its order number.

Files · Datasheets and drawings

The documents nothing links to

Technical documentation is where industrial suppliers are unusually strong and unusually careless. Datasheets, dimensional drawings, certificates and installation instructions exist as files, are genuinely useful, and are frequently uploaded to a directory that nothing on the site references. They can be opened by anyone with the address and found by nobody.

Keeping the decision about which documents belong where in a shared overview stops the next person from quietly reinstating a category that looked like an oversight.

The second point is where most of the value sits. A buyer comparing two suppliers wants a torque figure, a tolerance and an operating range, and they want them without downloading anything. Putting those three numbers into the page text is an afternoon of work per product family and changes what the page can be used for entirely. Tracking which families have been done in one workspace keeps a job of that size from stalling halfway.

Certificates deserve a separate word. Conformity declarations, material certificates and approval documents are searched for by designation surprisingly often, usually by an engineer verifying a specification during a procurement process. A supplier whose certificates are linked, named intelligibly and current will be found at exactly the moment somebody is checking whether they qualify — which is a narrow window and a decisive one.

Filters · A specific trap

When the catalogue generates its own problems

Catalogue software with filters produces addresses for every filter combination. Material, diameter, thread type, availability — four filters with five options each generate hundreds of addresses, every one of them a legitimate page and almost none of them worth finding. Submitted wholesale they consume the entire allowance and fill the site with near-identical listings.

2
days the figures run behind
4–8
weeks before movement shows
1
filing per address that changed

The resolution is a configuration decision rather than a content one: filter combinations remain available to users and are excluded from the sitemap and from submission. Available to anyone who clicks, invisible to discovery. That satisfies both requirements without compromise, and it is set once at the catalogue level rather than argued about per release.

One further note on availability filters specifically. A listing filtered to items currently in stock changes content constantly, which makes it a poor candidate for discovery under any circumstances. Keep it out of the sitemap without hesitation — nobody searching for a product category wants a snapshot of what happened to be in the warehouse on the day it was recorded.

Proof · Did anything change

Establishing that the record moved

Filing an address asks for something; it does not promise anything. Real confirmation arrives only later, when that address turns up in the reporting with impressions beside it, and the interval is measured in weeks. What lies between is a stretch during which nothing can be established either way — and it is in exactly that stretch that companies file the same thing over and over out of frustration, burning allowance for nothing.

A rule that holds up in practice: file it a single time, leave it alone for seven days, and only then look at whether coverage changed. Where a fortnight has passed and a properly filed address is still missing, the fault has almost certainly moved to the page — no internal link points at it, the same wording sits at a second address, or the body text is assembled after delivery rather than shipped with it. None of those is repaired by filing again. Watching the sequence inside a connected environment separates the possibilities without maintaining a spreadsheet alongside.

Frustration is a fair response in a business where nearly everything else reports back within hours. Not this. The remedy is to make the interval part of the procedure rather than a matter of willpower: an entry in the diary seven days out, instead of an intention to look in again at some point. That kills the reflex to re-file, and it leaves behind dates showing when each filing went in — which turns out to matter when somebody asks a quarter later why one product family took so long to surface.

Order · How to sequence the work

The order of operations

Sequence matters more here than in most work, because each step invalidates measurements taken before it. Submitting before the catalogue is linkable wastes the submission; analysing reports before confirming delivery means analysing an incomplete record.

StepTaskEffort
OneConfirm product text arrives in the delivered sourceA morning
TwoDecide which parts deserve a pageA day with product management
ThreeBuild category listings that link down to productsDevelopment work, one to two weeks
FourLink every document from its product pageLargely automatable
FiveSubmit, priority families firstDays, spread across the allowance

The first step falls to a technical review and admits no postponement, since every later step rests on its outcome. Deciding which families justify the effort is less a matter of product-management taste than of measured demand, and keyword research supplies that figure. Wherever a group of variant pages carries word-for-word the same body text, the remedy sits with on-page work rather than with further filing. Operating the three of them out of a single account is what stops the order from slipping, because whatever one step produces is on screen when the next begins.

Check what your catalogue has actually had recorded

Questions · From technical and product teams

Questions from product and technical teams

Our catalogue is behind a search form. Is that really a problem?

For customers who already know the designation, no. For everybody else, yes — a form is an instruction rather than a route, so nothing reaches those pages unless something links to them. The fix is category listings that link down to products, which is development work of one to two weeks and typically the highest-return change available on an industrial site.

Should every part number have its own page?

No. Product families and genuinely distinct types, yes. Size, material and packaging variants belong on the type page as a table. Eleven thousand entries usually reduce to three to six hundred pages worth building, which is the difference between an impossible content project and a large but finite one.

Is the daily limit per site or per account?

The account is the unit. A thousand addresses per day stretches across everything held there, meaning a German presence and an English one draw on one pool between them instead of one each. During a catalogue rollout somebody therefore has to decide the division explicitly — left alone, the earliest-filed batch simply absorbs what is available.

Our datasheets contain all the specifications. Is that enough?

Not on its own. Figures inside a file are much harder to use than the same figures in the page body, and a buyer comparing two suppliers will not download anything to make a shortlist. Put the three or four decisive numbers into the page text and keep the file linked underneath — an afternoon per product family.

What do we do about filter combinations?

Leave them working for users and keep them out of the sitemap and out of submission. They are legitimate pages with no independent reason to be found, and submitting them consumes the allowance while filling the site with near-duplicates. This is a one-off configuration decision at catalogue level.

A fortnight has gone by and the product page is still absent. Why?

By then the filing has stopped being the explanation. Three checks on the page itself: does any internal link point to it, does the identical wording exist at a second address, and does the body arrive with the delivered document rather than being built afterwards. Filing a third time repairs none of these.

Bereit für bessere Rankings?

Lassen Sie uns gemeinsam Ihre SEO-Strategie entwickeln.

Kostenlose Beratung