566 Links, 13 Destinations: VJOURNAL Audits Its Own Related Stories

The module nobody audits

Every article page on VJOURNAL carries a small card headed "Related stories" with up to three links under it. It is the last thing a template designer thinks about and one of the first things a reader uses when a piece ends. It is also a claim, and a fairly strong one: of the two hundred and nine stories in this archive, these three follow from the one you have just read. This piece checks whether that claim survives being counted.

The method, so it can be re-run. The list is produced by a single function, selectRelatedPosts. For this audit that function was executed against every record each edition serves — 209 in English, 209 in Russian, 209 in Spanish — using the same candidate list the article route hands it, and every printed link was tallied as an edge in a graph: one edge per link, counted by destination. Nothing is sampled and nothing is estimated; the figures below describe the whole graph.

What the rule actually says

The rule has two clauses and no scoring. For a story filed under the People category, the candidates are the other People stories. For everything else, a candidate qualifies if it shares the story's region or its category. There is no keyword overlap, no recency window measured against the source story, no diversity constraint and no ranking of any kind. The function takes the first three candidates it meets.

First in what order? In the order the candidate list already has, which is the archive's global sort: featured priority descending, then publication date descending. That sort exists to put the front page in the right order, and it was never designed to answer the question "what should I read next". But because the function slices rather than scores, the front-page order silently becomes the answer on all 209 article pages.

A third clause matters more than the other two, and it lives outside the function. Both the English route and the localized route hand it the indexable archive rather than the full one. In English that pool is 29 stories out of 209; in Russian and Spanish it is 20 out of 209. Whatever the module recommends has to come from there, which means the shape of that pool decides the shape of the entire internal link graph.

Five hundred and sixty-six links, thirteen destinations

Across the 209 English article pages the module prints 566 links. One hundred and seventy-six pages show three, 14 show two, ten show one and nine show none at all. Those 566 links resolve to 13 distinct destination pages — 6.2 percent of the edition. Sixteen of the 29 eligible destinations never receive a single one.

The distribution inside those thirteen is steeper than the headline suggests. Two stories collect 143 inbound links each, which is 25.27 percent of the graph apiece. A third collects 105. The top three destinations absorb 69.08 percent of every related link the English edition prints, and the top five absorb 85.16 percent. The full ranking runs 143, 143, 105, 63, 28, 25, 16, 13, 13, 7, 6, 2 and 2.

Not one of the 180 stories the quality gate keeps out of search receives an inbound link from this module. That is not a gap in the data; it is the arithmetic of the third clause. A pool of 29 can only ever recommend those 29, so four fifths of the archive can be recommended to nobody, by anybody, ever.

The same three stories, 105 times

Count distinct triples rather than distinct destinations and the picture sharpens again. Across 209 English article pages there are 14 different related-stories lists in existence, one of which is empty. The single most common list appears on 105 pages — exactly half the edition — and consists of three People profiles published on the same day.

The mechanism can be reconstructed by hand in a minute. One hundred and thirty-seven of the 209 English stories carry the region Global. The two highest-priority indexable stories in the whole archive are People profiles that also carry Global, holding the two largest featured-priority values in the file, 140 and 139. Every Global story therefore matches them before anything else and takes them both. An audit of this newsroom's own citation habits, a guide to pricing architecture and a brief on trade policy all end with the same two profiles and the same third one.

Nine pages with a heading and nothing under it

Nine English article pages print the "Related stories" heading above an empty list. Six are filed under Top News and three under World News; by region they are two from China, two from Japan, two from Russia, two from Spain and one from India. Their emptiness is structural rather than accidental: no indexable story shares their region, and none shares their category either.

The eligible pool has a very particular shape. By region it is Global with 22, USA with six and Dubai with one, and nothing else at all. By category it is People with 13, Beauty with three, then AI, Business, Design, Marketing, Style and Travel with two each and Innovation with one. Set that against the archive it is meant to serve, where Business has 66 stories, Top News 35 and World News 11, and those three categories contribute two, zero and zero eligible destinations respectively.

The two translated editions inherit the same shape with less room to move. Their pool is 20 rather than 29, they print 535 links each, and those links reach ten destinations. One hundred and sixty-nine pages show three links, 28 show exactly one and twelve show none. The top three destinations take 73.08 percent.

This is a supply problem, not a taste problem

It would be easy to read the concentration as an editorial preference for celebrity profiles. It is nothing of the kind. The module has no preferences; it has a filter and a slice. Concentration of this severity follows automatically whenever a small eligible pool meets a large archive with one dominant region tag, and it would follow just as hard if the two highest-priority stories were about tariffs.

That also means the repair is not "write better recommendation logic" in the abstract. Three specific changes would each move the number on their own: score candidates instead of slicing them, cap how many times any single destination may be recommended across the archive, and decide deliberately whether a related list is allowed to point at a page that is deliberately held out of search. The third is a policy question, not an engineering one.

What the graph cannot tell you

This audit counts link supply. It cannot tell you whether anybody clicks. There is no analytics data in this repository, so every sentence above describes what the site offers rather than what readers do with it, and a module that looks broken by these measurements might still be sending more people to a second article than no module at all.

Nor does it show that 180 stories are orphans. The related card is one internal link surface among several: the journal hub lists the whole archive with pagination, and each topic hub and each region desk lists its own members. The code that holds stories out of search says in as many words that they must stay reachable from those pages, and on that promise the archive delivers. What is missing is article-to-article flow, which is a different thing and a smaller one — but it is the surface a reader meets at the exact moment they have finished reading and are deciding whether to stay.

One last limit, and it applies to every number here. This measures the graph the code produces today. Both the eligible pool and the priority order change whenever a story is published or a gate condition is edited, so the right way to use these figures is as a method to re-run, not as a fact to quote back in a month.

The cheapest version of the fix

If only one change were available, it would be the cap. Allowing any single destination at most ten inbound recommendations would redistribute several hundred links overnight, without touching a word of copy and without changing which pages are eligible. The second cheapest is to break ties on something other than the front-page sort — shared category before shared region, then nearest publication date — which would at least stop 137 Global stories from arriving at one answer.

The wider lesson is not about this module. A recommendation surface built from a filter and a slice will always look like an editorial judgement to the reader and behave like a sort order in the code. If a newsroom is going to print the words "Related stories" above three links, it should be able to say what makes them related, and it should be able to prove it by counting.

link graph links destinations audits: Evidence map and limits

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Good search content answers the narrow question first and then earns attention with context that a result-page snippet cannot provide. The cheapest offer can become the most expensive when missing research, rights, testing or support has to be bought later. Read the publication date, update line and source trail together; any one of them alone can create a false sense of freshness. A useful brief identifies the affected audience, the time horizon and the action that could change when new evidence arrives. Compare the base case with a credible alternative and a downside case so that enthusiasm does not become the only planning assumption.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. A service comparison becomes useful when deliverables, exclusions, approval rounds and handover conditions are written in the same vocabulary. Ask what would disprove the preferred interpretation; a decision that cannot fail on paper cannot be responsibly tested in practice. Do not mix reach with demand; attention can grow while qualified enquiries, retention or purchase intent remain unchanged. Accessibility, speed and plain language are commercial quality signals because they reduce the work required from a real visitor. Estimate total cost across preparation, production, review, launch and maintenance rather than treating the initial quote as the whole budget.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. The most persuasive signal is observable behaviour: what people search, choose, keep, recommend or pay for after the announcement. A premium decision is usually won through clarity of scope, not through the largest possible list of features or talking points. Ownership must be explicit: name the person responsible for the next check, the next update and the final go-or-stop decision. A useful brief identifies the affected audience, the time horizon and the action that could change when new evidence arrives. Treat numbers as dated observations rather than permanent truths, especially when markets, platforms or public policy can move quickly.

link graph links destinations audits: Search intent behind the question

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Ask what would disprove the preferred interpretation; a decision that cannot fail on paper cannot be responsibly tested in practice. Read the publication date, update line and source trail together; any one of them alone can create a false sense of freshness. The reader needs a next step proportional to the evidence: monitor, test, compare suppliers, request a brief or postpone the decision. Estimate total cost across preparation, production, review, launch and maintenance rather than treating the initial quote as the whole budget. Look for incentives behind every public statement and ask who benefits if the most optimistic reading becomes the accepted one.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Use comparable criteria for every option: outcome, proof, total cost, lead time, ownership, reversibility and the cost of delay. The strongest option is the one that can explain its method, show relevant proof and define what happens when assumptions change. Look for incentives behind every public statement and ask who benefits if the most optimistic reading becomes the accepted one. Treat numbers as dated observations rather than permanent truths, especially when markets, platforms or public policy can move quickly. Before buying, ask for the smallest verifiable milestone that demonstrates quality without committing the entire budget at once.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Good search content answers the narrow question first and then earns attention with context that a result-page snippet cannot provide. Do not mix reach with demand; attention can grow while qualified enquiries, retention or purchase intent remain unchanged. The most persuasive signal is observable behaviour: what people search, choose, keep, recommend or pay for after the announcement. Compare the base case with a credible alternative and a downside case so that enthusiasm does not become the only planning assumption. Translate a broad trend into one measurable hypothesis before changing a campaign, product, workflow or investment plan.

link graph links destinations audits: Signals that matter in practice

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. The most persuasive signal is observable behaviour: what people search, choose, keep, recommend or pay for after the announcement. Compare the base case with a credible alternative and a downside case so that enthusiasm does not become the only planning assumption. Define the decision before collecting more material, otherwise the research expands while the practical question remains unanswered. Estimate total cost across preparation, production, review, launch and maintenance rather than treating the initial quote as the whole budget. Look for incentives behind every public statement and ask who benefits if the most optimistic reading becomes the accepted one.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. A deadline is credible only when dependencies, review time and the cost of rework are visible alongside the launch date. A service comparison becomes useful when deliverables, exclusions, approval rounds and handover conditions are written in the same vocabulary. A useful brief identifies the affected audience, the time horizon and the action that could change when new evidence arrives. When sources disagree, publish the disagreement and lower the certainty instead of averaging incompatible claims into a neat sentence. Separate a confirmed fact from an interpretation, because a confident headline is not the same thing as a verified operating signal.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Ownership must be explicit: name the person responsible for the next check, the next update and the final go-or-stop decision. Good search content answers the narrow question first and then earns attention with context that a result-page snippet cannot provide. The reader needs a next step proportional to the evidence: monitor, test, compare suppliers, request a brief or postpone the decision. Define the decision before collecting more material, otherwise the research expands while the practical question remains unanswered. When sources disagree, publish the disagreement and lower the certainty instead of averaging incompatible claims into a neat sentence.

link graph links destinations audits: Options, trade-offs and alternatives

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Define the decision before collecting more material, otherwise the research expands while the practical question remains unanswered. Estimate total cost across preparation, production, review, launch and maintenance rather than treating the initial quote as the whole budget. Ask what would disprove the preferred interpretation; a decision that cannot fail on paper cannot be responsibly tested in practice. Ownership must be explicit: name the person responsible for the next check, the next update and the final go-or-stop decision. Do not mix reach with demand; attention can grow while qualified enquiries, retention or purchase intent remain unchanged.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. A service comparison becomes useful when deliverables, exclusions, approval rounds and handover conditions are written in the same vocabulary. Separate a confirmed fact from an interpretation, because a confident headline is not the same thing as a verified operating signal. Look for incentives behind every public statement and ask who benefits if the most optimistic reading becomes the accepted one. Do not mix reach with demand; attention can grow while qualified enquiries, retention or purchase intent remain unchanged. Define the decision before collecting more material, otherwise the research expands while the practical question remains unanswered.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Separate a confirmed fact from an interpretation, because a confident headline is not the same thing as a verified operating signal. The strongest option is the one that can explain its method, show relevant proof and define what happens when assumptions change. A service comparison becomes useful when deliverables, exclusions, approval rounds and handover conditions are written in the same vocabulary. A premium decision is usually won through clarity of scope, not through the largest possible list of features or talking points. Translate a broad trend into one measurable hypothesis before changing a campaign, product, workflow or investment plan.

link graph links destinations audits: Budget, timing and risk

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Treat numbers as dated observations rather than permanent truths, especially when markets, platforms or public policy can move quickly. Before buying, ask for the smallest verifiable milestone that demonstrates quality without committing the entire budget at once. Ask what would disprove the preferred interpretation; a decision that cannot fail on paper cannot be responsibly tested in practice. A service comparison becomes useful when deliverables, exclusions, approval rounds and handover conditions are written in the same vocabulary. Translate a broad trend into one measurable hypothesis before changing a campaign, product, workflow or investment plan.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Compare the base case with a credible alternative and a downside case so that enthusiasm does not become the only planning assumption. The most persuasive signal is observable behaviour: what people search, choose, keep, recommend or pay for after the announcement. Keep editorial evidence and commercial recommendations visually distinct so a reader can evaluate both without hidden persuasion. The strongest option is the one that can explain its method, show relevant proof and define what happens when assumptions change. Separate a confirmed fact from an interpretation, because a confident headline is not the same thing as a verified operating signal.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Compare the base case with a credible alternative and a downside case so that enthusiasm does not become the only planning assumption. Use comparable criteria for every option: outcome, proof, total cost, lead time, ownership, reversibility and the cost of delay. Ownership must be explicit: name the person responsible for the next check, the next update and the final go-or-stop decision. A useful brief identifies the affected audience, the time horizon and the action that could change when new evidence arrives. A premium decision is usually won through clarity of scope, not through the largest possible list of features or talking points.

link graph links destinations audits: A practical implementation sequence

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Separate a confirmed fact from an interpretation, because a confident headline is not the same thing as a verified operating signal. When sources disagree, publish the disagreement and lower the certainty instead of averaging incompatible claims into a neat sentence. Accessibility, speed and plain language are commercial quality signals because they reduce the work required from a real visitor. Good search content answers the narrow question first and then earns attention with context that a result-page snippet cannot provide. Before buying, ask for the smallest verifiable milestone that demonstrates quality without committing the entire budget at once.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Accessibility, speed and plain language are commercial quality signals because they reduce the work required from a real visitor. Before buying, ask for the smallest verifiable milestone that demonstrates quality without committing the entire budget at once. Ask what would disprove the preferred interpretation; a decision that cannot fail on paper cannot be responsibly tested in practice. Ownership must be explicit: name the person responsible for the next check, the next update and the final go-or-stop decision. Archive the evidence used for the decision so that a later reviewer can reconstruct why the team acted when it did.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. Good search content answers the narrow question first and then earns attention with context that a result-page snippet cannot provide. The cheapest offer can become the most expensive when missing research, rights, testing or support has to be bought later. Record which claim comes from a primary document, which comes from reporting and which is the newsroom's own analysis. Archive the evidence used for the decision so that a later reviewer can reconstruct why the team acted when it did. Read the publication date, update line and source trail together; any one of them alone can create a false sense of freshness.

link graph links destinations audits: How to compare the final decision

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. The reader needs a next step proportional to the evidence: monitor, test, compare suppliers, request a brief or postpone the decision. Define the decision before collecting more material, otherwise the research expands while the practical question remains unanswered. A premium decision is usually won through clarity of scope, not through the largest possible list of features or talking points. Keep editorial evidence and commercial recommendations visually distinct so a reader can evaluate both without hidden persuasion. Estimate total cost across preparation, production, review, launch and maintenance rather than treating the initial quote as the whole budget.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. A useful brief identifies the affected audience, the time horizon and the action that could change when new evidence arrives. Compare the base case with a credible alternative and a downside case so that enthusiasm does not become the only planning assumption. Archive the evidence used for the decision so that a later reviewer can reconstruct why the team acted when it did. Translate a broad trend into one measurable hypothesis before changing a campaign, product, workflow or investment plan. Accessibility, speed and plain language are commercial quality signals because they reduce the work required from a real visitor.

For the link graph links destinations audits question, the useful starting point is a decision that can be checked rather than a prediction that merely sounds complete. A premium decision is usually won through clarity of scope, not through the largest possible list of features or talking points. Accessibility, speed and plain language are commercial quality signals because they reduce the work required from a real visitor. Read the publication date, update line and source trail together; any one of them alone can create a false sense of freshness. Define the decision before collecting more material, otherwise the research expands while the practical question remains unanswered. Translate a broad trend into one measurable hypothesis before changing a campaign, product, workflow or investment plan.