Orphan-Page Reachability Sweep
A page in your sitemap that nothing links to, nobody shared and no index lists is not published. It is stored. This sweeps a library page by page — and refuses to make a claim about the whole thing when the way you chose the pages could never have found the missing ones.
The worked example ends in CANNOT SAY. Seven pages were swept from a sitemap, five of them are not findable, and the library verdict is still a refusal — because a sitemap cannot list the pages missing from your sitemap. The five findings stand. The extrapolation does not.
instant download · .xlsx · yours to keep
deterministic · offline · grades pages, never people
The problem
A page nothing leads to is not published. It is stored.
In the sitemap, externally linked, distributed, findable by search — and nothing on your own site links to it. The link floor caps it, and 78 is the highest score that can happen at.
Seven pages swept from a sitemap. Five are not findable. The library verdict is still a refusal, because a sitemap cannot list the pages missing from your sitemap.
Nothing is fetched or sent. You look, then you mark — which is also its honest limit: it will not find a page you never listed.
Every library over about fifty pages has a tail nobody has looked at in two years, and the usual way to check it is to crawl the sitemap. That method cannot work, and the reason is not subtle: the pages you are hunting are the ones missing from the list you are hunting with. Analytics has the same problem in a different direction — a page with no traffic does not appear, and no traffic is the symptom. This kit makes you name the method first, and tells you plainly when the method cannot answer the question.
See it work
Change the frame. Watch the library line move and the pages stay put.
Seven pages, swept from a sitemap. Switch the frame to every page, from a complete list — the library line resolves and not one page verdict moves.
How were these pages chosen?
| Page | Score | Band | Verdict |
|---|---|---|---|
| P-01 Pillar guide: choosing a channel | 100 | FINDABLE | FINDABLE |
| P-02 Product page: Channel Continuation Assay | 88 | FINDABLE | FINDABLE |
| P-03 Case study: how we cut three channels | 78 | FINDABLE▼ | ORPHANED |
| P-04 Blog post from March 2024 | 58 | ORPHANED | ORPHANED |
| P-05 Campaign landing page | 35 | STRANDED | STRANDED |
| P-06 Category page nobody links to | 34 | STRANDED | STRANDED |
| P-07 Pricing FAQ | 67 | ORPHANED | ORPHANED |
Editing P-03 — Case study: how we cut three channels
In a navigable indexweight 26
A nav, category page or hub a visitor would plausibly reach.
Inbound links from a reachable pageweight 22
A link from another orphan is not a path.
In the sitemap and indexableweight 16
Present, and not excluded by robots, canonical or a redirect.
Inbound external link or citationweight 14
Something you did not control points here.
Ever actually distributedweight 12
Sent, posted or shared — more than once, deliberately.
Findable by its own obvious queryweight 10
Your own site search returns it for the thing it is about.
LINK FLOOR — nothing on your site links here, so it cannot read FINDABLE at any score. It scored 78.
Page verdict
ORPHANEDscore 78
Fix first: Inbound links from a reachable page
Not findable
5/7
71% of the pages swept
Stranded
2
3 orphaned
Fix first
P-06 - Inbound links from a reachable page
FRAME-VALIDITY GATE — a sitemap cannot list the pages missing from your sitemap. The page verdicts above still stand; every stranded page found here is genuinely stranded. What cannot be claimed is anything about the pages this frame could never have shown you.
Library verdict
CANNOT SAY
A refusal is a finding. It tells you the next move is a complete list.
Your observations only — nothing is crawled, fetched or sent. It grades pages, never people. Not financial advice.
Switch the frame from “taken from the sitemap” to “every page, from a complete list”. The library verdict resolves from CANNOT SAY to ORPHANS FOUND — and not one page verdict moves. That is the separation the whole product rests on, and the verifier asserts it rather than trusting it. Then set coverage to 40 with a 60 minimum and watch it refuse for a different reason.
The model
Six signals, a floor on the page, and a gate on the claim.
In a navigable index 26, inbound links from a reachable page 22, in the sitemap and indexable 16, inbound external link 14, ever actually distributed 12, findable by its own obvious query 10. Weights sum to 100. FINDABLE at 75 or more, ORPHANED from 50 to 74, STRANDED under 50.
A page nothing on your own site links to cannot read FINDABLE at any score. Inbound links carries 22 rather than the heaviest weight precisely so the floor can bite: at 26 the best score without a single inbound link would be 74 — already below the threshold — and the floor would have been decoration. The verifier computes that maximum rather than assuming it.
The library verdict reads CANNOT SAY when the way you chose the pages could not have surfaced the missing ones. Sitemap, analytics and memory each exclude, by construction, exactly what you are searching for. A random sample below your own minimum coverage refuses too, for a different stated reason.
CANNOT SAY does not mean the sweep was wasted. Every page verdict is unaffected by the frame — a stranded page found through a bad sampling frame is still genuinely stranded, and fixing it is still worth doing. What the refusal declines is the extrapolation: any statement about the share of your library that is reachable, when the pages you looked at were chosen by a method that could not have included the missing ones. It also tells you the next move, which is to get a complete list.
Verified, not asserted
The workbook is checked against a reference engine.
You get the .xlsx — that is the product. It is built against a separate reference implementation of the same model, and every formula in the sheet is asserted equal to it before release: 65 cell-level checks on the worked example, 729 page states, and all 605 frame combinations. One of those checks rebuilds the entire workbook under a valid frame and asserts every page verdict is byte-identical. Below is that engine’s verbatim output on the same seven pages the workbook ships with, captured programmatically and diffed against this page.
==============================================================================
ORPHAN-PAGE REACHABILITY SWEEP - OPR-049 - RedHub AI LLC
==============================================================================
Grades whether a human path to a page exists at all. Your observations
only - nothing is crawled, fetched or sent. Grades pages, never people.
------------------------------------------------------------------------------
PAGE P-01 [Pillar guide: choosing a channel]
------------------------------------------------------------------------------
In a navigable index 2/2 weight 26
Inbound links from a reachable page 2/2 weight 22
In the sitemap and indexable 2/2 weight 16
Inbound external link or citation 2/2 weight 14
Ever actually distributed 2/2 weight 12
Findable by its own obvious query 2/2 weight 10
Score 100/100 -> FINDABLE
Link floor inert - something links here
VERDICT FINDABLE
Fix first nothing - every signal is at 2
------------------------------------------------------------------------------
PAGE P-02 [Product page: Channel Continuation Assay]
------------------------------------------------------------------------------
In a navigable index 2/2 weight 26
Inbound links from a reachable page 2/2 weight 22
In the sitemap and indexable 2/2 weight 16
Inbound external link or citation 1/2 weight 14
Ever actually distributed 2/2 weight 12
Findable by its own obvious query 1/2 weight 10
Score 88/100 -> FINDABLE
Link floor inert - something links here
VERDICT FINDABLE
Fix first Inbound external link or citation
------------------------------------------------------------------------------
PAGE P-03 [Case study: how we cut three channels]
------------------------------------------------------------------------------
In a navigable index 2/2 weight 26
Inbound links from a reachable page 0/2 weight 22
In the sitemap and indexable 2/2 weight 16
Inbound external link or citation 2/2 weight 14
Ever actually distributed 2/2 weight 12
Findable by its own obvious query 2/2 weight 10
Score 78/100 -> FINDABLE
LINK FLOOR FIRED nothing on your site links to this page, so it
cannot read FINDABLE at any score
VERDICT ORPHANED
Fix first Inbound links from a reachable page
------------------------------------------------------------------------------
PAGE P-04 [Blog post from March 2024]
------------------------------------------------------------------------------
In a navigable index 1/2 weight 26
Inbound links from a reachable page 1/2 weight 22
In the sitemap and indexable 2/2 weight 16
Inbound external link or citation 1/2 weight 14
Ever actually distributed 1/2 weight 12
Findable by its own obvious query 1/2 weight 10
Score 58/100 -> ORPHANED
Link floor inert - something links here
VERDICT ORPHANED
Fix first In a navigable index
------------------------------------------------------------------------------
PAGE P-05 [Campaign landing page]
------------------------------------------------------------------------------
In a navigable index 0/2 weight 26
Inbound links from a reachable page 0/2 weight 22
In the sitemap and indexable 2/2 weight 16
Inbound external link or citation 1/2 weight 14
Ever actually distributed 2/2 weight 12
Findable by its own obvious query 0/2 weight 10
Score 35/100 -> STRANDED
Link floor inert - already below FINDABLE
VERDICT STRANDED
Fix first In a navigable index
------------------------------------------------------------------------------
PAGE P-06 [Category page nobody links to]
------------------------------------------------------------------------------
In a navigable index 2/2 weight 26
Inbound links from a reachable page 0/2 weight 22
In the sitemap and indexable 1/2 weight 16
Inbound external link or citation 0/2 weight 14
Ever actually distributed 0/2 weight 12
Findable by its own obvious query 0/2 weight 10
Score 34/100 -> STRANDED
Link floor inert - already below FINDABLE
VERDICT STRANDED
Fix first Inbound links from a reachable page
------------------------------------------------------------------------------
PAGE P-07 [Pricing FAQ]
------------------------------------------------------------------------------
In a navigable index 2/2 weight 26
Inbound links from a reachable page 2/2 weight 22
In the sitemap and indexable 1/2 weight 16
Inbound external link or citation 0/2 weight 14
Ever actually distributed 1/2 weight 12
Findable by its own obvious query 1/2 weight 10
Score 67/100 -> ORPHANED
Link floor inert - something links here
VERDICT ORPHANED
Fix first Inbound external link or citation
==============================================================================
THE LIBRARY
==============================================================================
Pages swept 7
Findable 2
Orphaned 3
Stranded 2
Not findable 5 (71% of the pages swept)
How these pages were selected sitemap_only
Coverage of the inventory 100% (your minimum: 60%)
FRAME-VALIDITY GATE FIRED
a sitemap cannot list the pages missing from your sitemap
The per-page verdicts above still stand. Every stranded page found
here is genuinely stranded. What cannot be claimed is anything
about the pages this frame could never have shown you.
LIBRARY VERDICT CANNOT SAY
Fix first P-06 - Inbound links from a reachable page
==============================================================================What it is — and isn't
It grades pages you look at. It does not go and look at them for you.
- A page-by-page reachability sweep of a library you already have.
- Deterministic and offline — same inputs, same verdicts, every time.
- A named model you can argue with: the weights, both bands, the floor and the gate are all on this page.
- Dateless. Re-run it after any restructure or migration.
- Not a crawler. It fetches nothing and will not find a page you never listed.
- Not an SEO audit. Reachable is not the same as ranking.
- Not a promise. A page reading FINDABLE is a page someone could arrive at, not one they will.
- Not financial advice, and it grades pages rather than people.
Scope note: every mark is your own observation, so the output is only as honest as the marking — and the sweep can only speak about pages you listed, which is precisely why the frame gate exists. No verdict guarantees a traffic, indexing or revenue outcome. This is not financial advice.
Pairs well with
The Sweep finds the pages. These three decide what happens next.
This sweep asks whether a human path exists. That one asks whether a machine can fetch and render the page at the end of it.
ViewOnce a stranded page is reachable, the next question is whether it was worth reaching. This is the quality guardrail.
ViewHuman reachability first, then machine citation. Running this before that one saves you optimising pages nothing links to.
ViewCommon questions
The mechanics, stated plainly.
One spreadsheet
Find the tail.
- Three tabs, 50 rows, live formulas, a worked seven-page example.
- Three consistency guards, so the library-level inputs cannot silently disagree.
- Opens in Excel, Google Sheets or Numbers. Nothing to install.
- Deterministic and offline. It grades pages, never people.
Sold by RedHub AI LLC · Secured by Stripe · redhub.ai