Industry Report #006
The Unlisted Network: Telecom & Media's Measurable Web Is Nearly Half Invisible
On the six measurable sites, an average of 45% of published content is link-invisible: live, in the sitemap, and pointed at by nothing. The two consumer giants that could be crawled at all are the extremes: a carrier whose commerce funnel leaves 92% of its pages unlinked, and another whose support library buries 67%. The sector's front doors are walls, and behind the walls the shelves are unlabeled. — Share of published pages with zero inbound links of any kind, averaged across the six telecom and media sites where the measure is computable, the highest rate of any cohort in this series.
Across eight Fortune 500 telecom and media sites (the survivors of the most walled sector yet), 45% of published content on the measurable sites has no inbound link at all. One carrier's commerce funnel leaves 92% of its pages unlinked; another's support library buries two-thirds. One media site routes 98% of its structure through 1% of its pages.
TL;DR The fast read. Each line is measured below.
- 45% of published content is link-invisible on average across the six measurable sites, the highest rate of any industry in this series (tech: 32%, industrial: 20%).
- 92.5% of one consumer carrier's pages have no inbound link at all: a commerce funnel of hundreds of buy-flow pages that exist only in the sitemap.
- 6+ of the sector's biggest consumer names were unmeasurable outright, blocked by hard anti-bot walls at carrier, cable, and studio domains. What we measured is the sector's back office: corporate media and tower infrastructure.
- 98.4% of one media conglomerate's structural load runs through its top 1% of pages: a 243-page site that is functionally a star around one hub.
- 1.7% of the cohort's 14,578 schema annotations are answer-shaped, the sector's saving grace next to retail's zero and industrial's three, and still one annotation in sixty.
- 72% of the largest clean site's pages sit four or more clicks deep. The healthiest graphs in this cohort hide their content in depth instead of in orphans.
Walled and Unlisted
Telecom and media is the sector where the two failure modes of this series finally meet. Retail was walled. Industrial was open but unwired. Telecom is both at once, split down the middle of its own business.
The consumer half is retail all over again, but harder. The major wireless carriers, cable operators, studios, and tower REITs on the candidate list fell to hard anti-bot walls before a single page was measured: six of the sector’s biggest consumer names returned nothing but challenge pages at both their storefront and corporate domains. One entertainment giant blocked every request despite advertising a two-thousand-URL sitemap. What survived to be measured is the sector’s back office: corporate media sites, holding companies, tower infrastructure REITs, plus two consumer carriers whose front doors happened to open.
And those two carriers are the report’s sharpest exhibit. They crawled, and what the crawl found behind the open door is the worst link topology in this series. One carrier’s measurable web is its commerce funnel: hundreds of buy-flow pages of which 92.5% have no inbound link of any kind. The other’s is its support library: thousands of words per page, meticulously written, with 66.5% of it unlinked. Across the six measurable sites, link-invisibility averages 44.9%, the highest of any cohort and half again technology’s rate. The sector that runs the network has not wired its own web to it.
What Could Not Be Measured
Most of the sector’s consumer giants never entered the crawl: hard anti-bot walls at carriers, cable, studios, and related domains. The eight-site cohort is the measurable minority (2,276 pages, 3.5 million tokens): two consumer carriers, three corporate media properties, one advertising holding company, and two tower-infrastructure REITs. As with retail, that selection is a finding. For most of this sector an AI agent cannot form any picture at all; the picture below is only what leaks out from behind the wall. The full drop list is in Methodology.
Three Tiers of Reachability
Every page falls into one of three reachability tiers, judged against the full link graph (navigation included) for reach and the content link graph for editorial authority:
- Reachable via editorial links. An in-content link path leads to the page.
- Navigation-only. Reachable, but only through chrome an agent strips as boilerplate.
- Link-invisible. Zero inbound links of any kind; the page survives only in the sitemap.
Two of the eight sites did not expose a usable sitemap. Averaged across the six measurable sites:
Barely half the sector’s published content is reachable by following links, the lowest reachable tier in the series. And unlike technology, almost none of the gap is navigation-only: these pages are not routed through menus an agent might partially recover. They are simply unlisted.
View data
| Site | Link-invisible page rate (% of site's published inventory) |
|---|---|
| C3 | 3.8% |
| C8 | 18.6% |
| C4 | 21.4% |
| C2 | 66.5% |
| C6 | 66.9% |
| C1 | 92.5% |
| Sample mean | 44.9% |
The cohort splits with its business model. Corporate media properties sit in an ordinary band; the consumer-facing surfaces dominate the right tail of the strip. The two carriers fail in opposite directions.
The funnel (C1) is a homepage and a cloud of disconnected commerce endpoints (average click depth 0.15). Pages are generated per plan, per device, and per market, each an entry in the sitemap and a target of exactly zero links, reachable only by search, ad landing, or in-app deep link. As a conversion machine that may be deliberate; as a knowledge surface an agent asked about plans, coverage, or devices can walk to almost none of them.
The library (C2) is support documentation: long-form pages with strong metadata and structured-data markup, published into an archive nothing points at. It is industrial’s newsroom pattern transplanted to support content. The irony is that support is exactly what AI assistants are best positioned to relay, and exactly what a link-following agent here cannot find.
Fragility: The Star and the Chain
Betweenness concentration averages 37.6%, the mildest in the series, but the range (1.8% to 98.4%) is the widest, and both ends are instructive.
View data
| Site | Share of total betweenness carried by the top 1% of pages (%) |
|---|---|
| C4 | 1.8% |
| C8 | 4.8% |
| C6 | 17.1% |
| C3 | 24.6% |
| C7 | 36% |
| C2 | 55.7% |
| C1 | 62.5% |
| C5 | 98.4% |
| Sample mean | 37.6% |
At the top, C5, a media conglomerate, routes 98.4% of its structural load through 1% of its pages: 243 pages joined by just 320 editorial links, a graph that is functionally a star around a single news hub. Remove one page and the site falls apart into fragments. At the bottom, a media corporate (C4) spreads load so evenly that no page carries 2%, the healthiest circulation number in the series.
The healthiest graphs here hide a different cost: depth. The largest clean site in the cohort (C3, a media corporate) orphans almost nothing (0.6%) but buries 72% of its pages four or more clicks from home, with a maximum depth of 11. Its sibling (C4) reaches depth 17. These are chronological news archives: every article is linked from the next, in a chain an agent must walk month by month backward. Nothing is invisible; everything is far. The same trade industrial’s deep trees made, at media publishing volume.
Machine-Readability: The First Sector With a Pulse
The structured-data switch splits the cohort as it did retail and industrial: five sites at or near 100%, three at zero (both tower REITs and the media conglomerate publish none at all). But composition finally shows something the last two sectors could not:
250 answer-shaped annotations (1.7%): a small number that towers over retail’s zero and industrial’s three. Nearly all of it sits on the carrier support library (C2), which marks up its FAQ content properly, the one machine-readability investment in the cohort that matches what an assistant actually needs. The pattern completes a three-sector arc: the companies with answer-shaped content are the ones whose business is answering (support operations), and even they forget to link to it.
Render dependency averages 6.8%, with one outlier: a tower REIT (C6) serving 31% of its content client-side. Combined with high link-invisibility and weak title coverage, this makes it the least machine-legible site in the cohort on every axis at once.
Agent-readiness averages 10.7%, matching retail’s level. Content signal, API catalog, and markdown negotiation: zero of eight, the fourth consecutive cohort where no site passes any of the three checks that would signal deliberate agent support.
What Every Telecom & Media Site Shares
The organizing section of the measurable cohort is the newsroom: on the three sites where a news path exists, it holds 37% of pages on average, and on the media corporates it dominates outright. This is the industrial pattern again: the open web of a consumer sector is its press office. What is different here is what sits behind the wall. Industrial’s product content lives in distributor portals and PDFs; telecom’s lives in the app and the account dashboard. The sector has, more completely than any other, moved its customer-facing web off the web; what remains public is either a press release or an unlinked funnel page.
The Industry Scorecard
Five-lens green / amber / red scores for each of the eight measurable sites:
| Site | Skeleton Size, density, and average path length. How big and how connected the site is at the body level. | Circulation PageRank distribution and structural bottlenecks. How importance flows between pages, and which hubs hold it all together. | Organs Community detection. Whether the site's topical clusters cleanly separate, or whether one mega-cluster dominates everything. | Health Islands, orphans, and dead-ends. Where content is structurally dying: unreachable, unlinked, or terminating. | Nervous Sys. Click depth, bridges, and cross-community linking. Whether the site is a well-designed building or a pile of disconnected rooms. |
|---|---|---|---|---|---|
| C1 | |||||
| C2 | |||||
| C3 | |||||
| C4 | |||||
| C5 | |||||
| C6 | |||||
| C7 | |||||
| C8 |
- Green: healthy
- Amber: moderate concern
- Red: critical
Content Quality at Scale
| Content metric | Sample mean (n=8) | Notes |
|---|---|---|
| Pages per site | 285 | Median 203, range 129–571 |
| Avg word count per page | 981 | Second only to tech; one support library averages 3,440 |
| Avg token count per page | ~1,546 | Character-based estimate (length / 4) |
| Pages with thin content (under 200 words) | 26.2% | Between industrial (20%) and retail (38%) |
| Internal-link share of all links | 77.8% internal | In line with the series |
| Title tag coverage | 95.9% | One site drags it (67.5%) |
| Meta description coverage | 67.4% | Wide spread: 0% to 99.5% |
What This Means for AI Search Readiness
1. The wall now has a hole in it, and the hole is the finding. Retail’s wall was clean: nothing measurable behind it. Telecom’s wall leaked two consumer surfaces, and both are structural disasters on the open web. The reasonable inference about the sector’s walled properties is not “probably fine behind the wall.” The two data points we have say the wall and the wiring fail together: companies that treat the open web as a threat surface also stop maintaining it as an information surface.
2. Support content is the sector’s unclaimed asset. The one carrier that writes and marks up genuine answer content (long-form pages, FAQ schema, full metadata) leaves most of it unlinked. For a sector whose customer questions (“why is my bill higher,” “what does this plan include”) are exactly what AI assistants field daily, linked and answer-shaped support content is the cheapest citation surface available. It is already written. It just is not wired.
3. Depth is the next invisibility. The cohort’s healthiest sites orphan nothing and bury everything: 72% of pages at depth 4+, with archives seventeen clicks deep. As agents get better at boilerplate-stripping and worse at exhaustive traversal, a seventeen-click chain is functionally an archive wall. Chronology is not architecture; a news archive needs topic hubs, not just a next-page link.
The fix is structural, and every piece of it is measurable: link the funnel and support inventories into the editorial graph, collapse the deep chronological chains behind topic hubs, spread betweenness off the single news hub, and put answer-shaped markup on the support content that already answers. This is the entire premise of the Digital MRI service.
Methodology
Eight anonymized Fortune 500 telecom and media sites (two consumer carriers, three corporate media properties, one advertising holding company, and two tower-infrastructure REITs) were each run through the same pipeline: an HTTP-first adaptive crawl, main-content extraction (navigation, header, and footer links filtered out), dual-graph construction, and the five-lens topology analysis (Skeleton, Circulation, Organs, Health, Nervous System). The candidate list was substantially larger: six major consumer names were dropped at the pre-flight probe or crawl on hard anti-bot grounds (both storefront and corporate domains), one entertainment conglomerate was blocked by its WAF despite a published sitemap, one venue site served an empty JavaScript application shell, and one network operator crawled too thin to measure. The cohort is the measurable subset of the sector, which systematically excludes its most consumer-facing properties, a selection effect reported as a finding, as in IR-004. One holding company’s site had rebranded to a new domain mid-study; its crawl followed the redirect and measured the live property.
Reachability and invisibility use only the six sites whose crawl captured a usable sitemap; topology, fragility, content, and the scorecard use all eight. Aggregate figures are exploratory benchmarks for the measurable F500 telecom and media web, not point estimates for the sector.
The dual-graph model. Each page is classified against two graphs from the same crawl: the full graph (navigation included) for reach, and the content graph (in-content editorial links only) for authority. A page is reachable via editorial links if it has an inbound content link, navigation-only if reachable in the full graph but not the content graph, and link-invisible if nothing links to it in either, surviving only in the sitemap. This matches IR-002 through IR-005; per-site rates are not directly comparable across cohorts, but the shared finding now holds across six sectors: a measurable share of F500 content sits outside any link path an unaided agent can walk.
Disclaimers:
- Methods. PageRank, Louvain community detection, and betweenness centrality over crawled page structure.
- Robots & ethics. Site-level robots directives respected; disallowed pages never fetched; anti-bot walls honored, never circumvented. Publicly accessible page structure only. No content, metadata, or user data stored.
- Anonymization. Codenames C1-C8; the codename-to-domain mapping is intentionally not published. A site-level data appendix is available on request.
- Navigation exclusion. Any link target appearing on more than 80% of a site’s pages is treated as global navigation and dropped from the content graph; the full graph keeps it for reachability only.
- Token counts. Character-based estimates (page length divided by four), not tokenizer output.
- Betweenness. Reported as the share of total betweenness centrality carried by each site’s top 1% of pages.
- Scope. Statistical patterns for educational purposes only; not advice about any specific site or company.