Professional Networking
The profile is free. The paths between profiles are metered — in every direction.
This is not a page arguing that LinkedIn is useless, or that Premium is a scam, or that professional networking should be free. The research behind it found the opposite of the easy version: LinkedIn’s free tier is one of the most genuinely functional on any major platform, and a randomized controlled trial — the strongest form of evidence this index ever gets to cite — found that simply teaching job seekers to use free LinkedIn raised employment from roughly 70% to 77%, with effects lasting at least a year. Wheeler et al., AEJ: Applied Value case leads this page The charge is structural. LinkedIn sits between every participant in white-collar hiring, selling, and networking — and it has built a distinct paid product for each of them, around the same interactions, on a graph none of them can take anywhere else. A single hire can involve a paying candidate, a paying recruiter, a paying employer, and a salesperson paying to reach the same company — while LinkedIn’s own engineering literature describes optimizing its notification machinery for “downstream sessions,” Sessions, and Weekly Active Users. LinkedIn engineering papers
CFS = CVI − CHI (75 − 67 = +8). A finely balanced relationship, and the balance is the finding: LinkedIn delivers real, measurable professional value — and extracts attention, money, and data from more sides of the same relationship than almost any company in this index. Neither half cancels the other. The value is precisely why the dependency and the monetization decisions matter. Methodology →
The defining tension
LinkedIn became the infrastructure of white-collar working life, then built a separate paid product for every party who depends on it — around the same interactions, on a graph only LinkedIn holds.
Both halves of that sentence are supported, and the page refuses to collapse them. The free tier hosts a genuinely useful professional identity, unlimited job applications, and a network of 1.3 billion registered members — and causal research shows free use improves employment outcomes. RCT evidence · LinkedIn pressroom
The same network position lets LinkedIn charge the candidate, the recruiter, the employer, and the salesperson around a single professional relationship, while its notification systems are engineered — by its own published account — to maximize return visits. LinkedIn engineering papers
The network is genuinely valuable. That is exactly what makes every toll collectible.
Read this first · Evidence against hostility
What LinkedIn actually delivers.
A CVI of 75 rests on evidence most companies in this index cannot produce — including the only randomized controlled trial this index has ever been able to cite for a platform’s core value claim. A reader who skips this section will misread everything that follows.
Causal evidence that free LinkedIn use improves employment outcomes. A randomized controlled trial with South African job-readiness trainees (Wheeler, Garlick, Johnson, Shaw & Gargano, published in AEJ: Applied Economics) found that adding LinkedIn training to a work-readiness program raised end-of-program employment from approximately 70% to 77%, with effects persisting at least twelve months; the authors attribute the effect to platform use itself revealing information to employers. Wheeler et al., RCT This is peer-reviewed causal evidence — not a company survey — that the free product does what it claims. Almost nothing else in this index has that.
The free tier is genuinely large. A hosted, Google-indexed professional identity; unlimited job applications including Easy Apply; Open to Work signaling with a recruiters-only mode hidden from your employer; posting, articles, newsletters and feed distribution with no posting paywall; messaging with every first-degree connection; up to 30,000 connections; a full data archive export; and free identity verification via CLEAR for roughly 200 million US members since 2023 — a checkmark X sells as a subscription. LinkedIn compare-plans page · help documentation A job seeker can run a complete search paying $0, and most independent reviewers advise exactly that.
Consumer Premium is a sideshow, and this page says so. Premium subscriptions passed $2B in trailing-12-month revenue in January 2025 — roughly 12–13% of LinkedIn’s revenue. Talent Solutions alone passed $7B two years earlier. Microsoft/LinkedIn disclosures LinkedIn’s model is the one its own S-1 described in 2011: members ride mostly free; enterprises pay for access to them. Any page framed around “Premium is expensive” would miss the actual machine.
The most-resented limits double as real anti-abuse infrastructure. InMail credits are refunded only when the recipient replies — an incentive design that literally pays senders not to spam. The ~100/week invitation cap arrived amid industrial automation abuse and binds paying users too; LinkedIn did not sell its removal, which cuts against a pure monetization reading. Recruiter seats with chronically low response rates face sending suspensions. LinkedIn policy documentation A fully open, unmetered messaging layer over 1.3B profiles would be email — a channel where roughly half the volume is spam.
The paid tools are priced below what they replaced. A LinkedIn Recruiter Corporate seat runs roughly $11–13K/year; one contingency-agency placement of a single $100K hire costs $20–25K. Sales Navigator undercuts legacy sales-intelligence contracts several-fold. Premium Career, at $39.99/month, costs less than fifteen minutes of the average $207/hour career coach and bundles a real 25,000-course learning library. Pricing comparisons, 2026 The counterfactuals were dramatically more expensive.
Documented voluntary self-correction. In 2015 LinkedIn cut its own email volume ~40% — “for every 10 emails we used to send, we’ve removed 4” — and reported member complaints falling by half. Pre-GDPR, uncompelled. LinkedIn blog, July 2015 Its published notification ML has repeatedly reduced send volume, and a per-notification disable-probability model gives user irritation a real feedback channel into delivery policy. The caveat comes later; the fact stands here.
The free tier has recently expanded, not only contracted. Free employer job posting arrived in 2021 (previously paid-only since 2005). Career Break profiles in 2022. Free identity verification in 2023. Creator tools — newsletters, follow-primary profiles, analytics — distributed to all members free in 2024. LinkedIn announcements The historical record is genuinely mixed, and this page scores it as mixed.
Individual members are the algorithm’s favored class. Independent feed analysis finds personal posts vastly outperform company-page posts in distribution. The same design that frustrates corporate marketers hands free individual members organic reach most platforms now sell. Feed-composition studies LinkedIn’s people-over-brands bias is a real, uncharged benefit to the people this index cares about.
Exhibit 1 · Defining pattern · Type specimen
The Professional Tollbooth.
Most companies monetize one side of a relationship. Marketplaces monetize two. LinkedIn has built distinct paid products for five different participants around the same professional interactions — each sold separately, each priced to its buyer, all levied on movement through a graph only LinkedIn operates. Every price below is LinkedIn’s own published figure or a documented 2026 market figure. LinkedIn pricing pages 2026 pricing trackers · UK G-Cloud filing
Sees who viewed their profile (365 days), how they compare to other applicants, “Top Applicant” signaling, 5 InMails to reach recruiters, AI tools, Learning.
Full-network candidate search, 30–150 InMails/month, pipeline tooling, ATS integration. Corporate seats are quote-only.
Promoted job postings by the click; job slots; Talent Insights analytics; Career Pages; Premium Company Page; ads on its own employees’ profiles.
Prospect search free of the limits placed on free search, 50 InMails, buyer-intent signals computed from the buyer’s own LinkedIn engagement.
“Unlimited people browsing” — the removal of the undisclosed monthly search meter — 15 InMails, company insights, viewer intelligence.
Recruiters are there because candidates are; candidates because recruiters are; sellers because buyers are. The connections, endorsements, and histories that make the graph valuable were contributed free by members — and cannot be exported as a functioning network to anywhere else. LinkedIn litigated for five years, in hiQ Labs v. LinkedIn, to keep even public profile data from leaving on someone else’s terms. hiQ v. LinkedIn, settled 2022
LinkedIn doesn’t charge you to exist on the network. It charges you to move: to see, to be seen, to reach, to be reachable. And because everyone in your professional life is standing on the same network, it can sell tickets to both ends of the same conversation — sometimes to four or five people at once.
Exhibit 2 · Strongly supported
Weaponized Curiosity.
The CHI Lexicon defines Weaponized Curiosity as the deliberate creation, preservation, or exploitation of uncertainty to provoke engagement, disclosure, or payment for its resolution. LinkedIn’s “Who’s Viewed Your Profile” architecture is among the most complete implementations this index has examined — and it is not an accident of recent growth-hacking. The partial reveal has been the design since the feature launched in 2007, when it showed anonymized descriptors like “a product manager at Microsoft” by default. In 2013, LinkedIn’s own relaunch messaging described the feature as appealing to “humans’ natural curiosity.” TechCrunch, 2007 & 2013
The same mechanic runs on job pages. Free users see an anxiety-calibrated competition number — “Over 100 applicants” — which counts apply-button clicks, not completed applications, inflating perceived competition; LinkedIn’s own ad tooling distinguishes the two, the public badge does not. Resolving the anxiety — “how do I actually compare?” — is the Premium feature. And the marketing aimed at that anxiety is correlational by construction: “Premium members are 2.6x more likely to get hired on average” is LinkedIn’s own claim, published with no controlled evidence. Hiration, 2026 LinkedIn marketing, 2023
LinkedIn tells you somebody professionally relevant looked at you, describes them just well enough to matter, withholds who it was — then sells you the answer. It has known this works since 2007. The data being sold is data about you, generated by other people looking at you, that LinkedIn already has either way.
What this exhibit does not claim. Viewer privacy is real and absolute: no tier can unmask viewers who chose private mode, and LinkedIn consistently enforces the viewer’s choice over the subject’s curiosity. The anonymized cards exist partly because viewers chose limited disclosure — a legitimate member control, present since 2007. Free members also get real aggregate analytics most networks don’t give away. The finding is the calibration: the free tier discloses exactly enough to create the question, and the paid tier is the answer.
Exhibit 3 · The sharpest single finding
The reciprocity rule that money removes.
LinkedIn justifies the viewer-visibility trade as reciprocity: you can’t watch anonymously and still be watched. That is a defensible fairness norm — right up until you notice who it applies to.
The trade is enforced in full: privacy costs you your own information. LinkedIn help documentation, via mirrors and 2026 guides
The reciprocity principle is dissolved by payment. No LinkedIn statement locates a rationale for why. premium.linkedin.com, fetched Aug 2026
“If you want to watch anonymously, you give up being watched” is presented as a principle of fairness. Pay $39.99 a month and the principle stops applying to you. A rule that money can remove was never a principle. It was a price.
Independent corroboration — correctly labeled. In May 2026, the European privacy group noyb filed a complaint with the Austrian data-protection authority alleging that locking full profile-visitor data behind Premium, while free users receive vague aggregates, violates the GDPR’s Article 15 right of access to one’s own personal data. Pending complaint · no finding · LinkedIn disputes it CHI cites this not as authority — it is an allegation, not a ruling, and whether Article 15 reaches this design is a live legal question — but as independent identification of the same asymmetry this exhibit documents: a regulator-facing organization looked at the same architecture and described it as selling users access to their own data. Whatever the Austrian DPA decides, the design it describes is not in dispute.
Exhibit 4 · First-party evidence
The Notification Engine.
CHI usually has to infer what a notification system is optimized for. Here it doesn’t. LinkedIn’s engineers have published the objective function — repeatedly, over a decade, in peer-reviewed papers and on LinkedIn’s own engineering blog. The positive outcomes these systems maximize are sessions, page views, and Weekly Active Users. That is not CHI’s characterization. It is the reward function.
Email sending formulated as a constrained optimization whose positive outcomes are “downstream sessions” and page views, and whose negative outcomes are unsubscribes and complaints. Reported result: cutting send volume 64.51% retained 98.16% of sessions — by LinkedIn’s own math, roughly two-thirds of its email had been buying almost no engagement, only complaint risk. Published paper
A gatekeeper across five channels — email, SMS, desktop, in-app, push — processing over 1 billion notification requests a day, with delivery-time targeting (patented as US9420062, “Delivery time optimization”) and models scoring both click-through and disable probability. Credited outcome, in LinkedIn’s words:
“Double-digit percentage increases in site-wide member engagement.”A system generating personalized content notifications — scoring roughly 500,000 candidate notifications per second for 550M+ members. Its stated purpose:
“Move members through engagement funnels, and trigger viral loops that increase overall platform engagement.”Note the frame: a low-utility notification is a cost because annoyance threatens future engagement — not because interruption is itself a harm.
Notification policy optimized sequentially with Sessions and Weekly Active Users as reward functions: +0.24% sessions and +0.18% WAU at −1.73% volume (2022); +0.72% sessions at −1.68% volume (2025). LinkedIn research papers The manufactured-return hypothesis, stated as an engineering objective, maintained across a decade of architectures.
Not every notification is the machine. CHI distinguishes: transactional communication (messages, application updates, security) is legitimate and mandatory; professional prompts (job alerts you asked for) are defensible; the CHI finding attaches to the engagement class — network activity, congratulate-X prompts, search-appearance counts, “you have missed notifications” — where no human action was directed at you and the documented purpose is the return visit. The suppression cost is part of the finding: independent submissions to the deceptive-design archive document opting out of all LinkedIn email requiring 30+ clicks (2026) and navigating ~64 menus (2023), across four separately-managed channels, from multiple sender addresses that defeat inbox filtering. Clean Email ranks LinkedIn the #1 sender its users most want gone. deceptive.design entries · unsubscribe-guide vendors And the historical floor matters: the 2015 Perkins class action — reminder emails sent in users’ names without consent — ended in a $13M settlement. Perkins v. LinkedIn, settled 2015–16
The question “is LinkedIn spamming me to get me to come back?” does not require inference. LinkedIn’s engineers wrote it down: the system that decides what lands in your inbox is scored on whether you come back. It is also, credit where due, the same system that cut two-thirds of the email — because that email wasn’t bringing anyone back.
Fairness paragraph, stated plainly. LinkedIn’s controls are unusually granular — per-category, per-channel, with digest options, reachable from one settings hub. The 2015 email cut was real, voluntary, and published with numbers. The ML stack demonstrably reduces volume, and the disable-probability model means user irritation genuinely feeds back into policy. The rebuttal writes itself, though: every published reduction is optimized subject to preserving sessions — volume falls only where it wasn’t buying visits. The objective is never the user’s stated preference. These papers prove the optimization target, not malice, and CHI does not claim otherwise.
Exhibit 5 · The counterweight
The free product actually works.
The research specifically tested — and rejected — the claim that LinkedIn hollows out its free tier to force upgrades. The honest finding is more interesting, and it is the reason this page exists at all.
Wheeler, Garlick, Johnson, Shaw & Gargano, published in AEJ: Applied Economics (2022). Randomized, peer-reviewed, not company-funded; the authors attribute the effect to platform use itself — information revelation to employers. This is causal evidence that access to free LinkedIn improved employment outcomes for disadvantaged job seekers. CHI cites it at exactly the scope the study supports: one population, one program, free-tier use. It is not extended into “LinkedIn gets everyone hired.” It doesn’t need to be. RCT · SSRN 3452249 / AEJ: Applied
The conclusion is not “LinkedIn makes free useless.” The evidence killed that claim. The conclusion is sharper: LinkedIn leaves enough professional utility free to make itself genuinely load-bearing in white-collar life — and then meters the access, intelligence, and reach around a network people increasingly can’t not be on. The generosity isn’t the opposite of the strategy. It is the strategy’s foundation.
Direction of travel · The honest timeline
What moved behind the paywall — and what moved out.
The lazy version of this section — “LinkedIn keeps taking free things away” — is not what the record shows, and CHI’s historical research specifically corrected several popular claims (see What we refused to count). The documented pattern: a concentrated 2015–2017 window in which capabilities users already had, including paying Premium users, migrated up into Sales Navigator and Recruiter; a genuinely mixed record since 2020; and a fresh 2024–2026 wave on the company-page side. Direction markers: UP = moved behind a (higher) paywall or restricted; FREE = made free or expanded.
Verdict on the history: a qualified free-to-paid migration is documented for 2015–2017 and again on the Pages side in 2024–2026 — but “LinkedIn progressively gutted the free tier” is not supportable as a general narrative, and this page does not claim it. Some marquee gates (full viewer lists, InMail, recruiter filters) have been paid essentially forever. Subscription Creep is therefore scored as partially supported, not established.
Supported patterns
What the evidence actually supports.
Primary pattern
Major supporting patterns
Secondary and qualified
Score breakdown
Why CHI is 67 — and why CVI is 75 anyway.
Score the behavior, not the vibes. The notification evidence is striking, but striking evidence of an optimization objective is not license to inflate the number; the free tier’s genuine strength is not license to deflate it. Both discipline runs are visible below.
CFS = 75 − 67 = +8. Under CHI v2.0, scores between −15 and +15 classify as finely balanced, and that is the accurate description of this relationship: substantial, partly causal, professionally real value — delivered through a machine that monetizes dependency from more angles simultaneously than any other company on this index. LinkedIn imposes meaningful monetization and engagement pressure, and it creates substantial measurable professional value. The score refuses to pretend either half away.
What we refused to count
CHI is not simply compiling complaints.
Reserved
Two investigations this page deliberately does not contain.
Special Investigation: LinkedIn’s Notification Machine — in preparation.
This page establishes the finding — the objective function is documented — and stops. Reserved for the investigation: the full engineering-objective genealogy (2016–2025) · the complete notification taxonomy and its defaults, instrumented on a fresh account · exact toggle counts across all four channels · email volume measurement · the Perkins docket record · peer-platform comparison · and firsthand suppression testing.
No link is provided because the investigation does not yet exist.
Special Investigation: Corporate Page Friction — under consideration.
CHI operates a LinkedIn company page and has an experiential audit of the admin, verification, invite-credit, reach, and support mechanics on the business side — where the 2024–2026 free-to-paid migration is sharpest. That material is deliberately kept off this page: the main assessment covers the economics; the firsthand audit belongs in its own lane.
Company-side monetization appears above only where economically necessary (Exhibit 1, the timeline).
Evidence & methodology
- Company pricingLinkedIn Premium Career — pricing and featuresFetched 21 Aug 2026: $39.99/mo / $239.88/yr, 5 InMails, 365-day viewer data, “browse in private mode” sold as a paid feature, and the “2.6x more likely to get hired” marketing claim. The controlling source for the reciprocity exemption and the consumer side of the tollbooth.
- Company pricingSales Navigator plan comparison — LinkedInFetched 21 Aug 2026: Core $119.99/mo, Advanced $159.99/mo, Advanced Plus quote-only; 50 InMails; “unlimited search free of the commercial-use limit” as a listed feature — the meter’s removal, sold.
- Company pricingLinkedIn Premium Company Page$99.99/mo or $839.88/yr; auto-invite of engaged members and higher invite allocations — the paid counterpart to the 2026 cut in free pages’ invite credits.
- Official filingLinkedIn S-1 (January 2011) — SEC EDGARThe business model in LinkedIn’s own founding words: “We provide the majority of our solutions to our members at no cost… We generate revenue from enterprises and professional organizations.” Members are the supply side; enterprises are the customers.
- Official filingMicrosoft FY2025 Form 10-KLinkedIn revenue $17.8B FY25 (FY26: $19.8B per LinkedIn’s pressroom); four monetized lines named but not quantified; and the growth framing in Microsoft’s own words: “Growth will depend on our ability to increase LinkedIn member engagement on the platform.”
- JournalismTechCrunch — LinkedIn passes $2B in Premium revenue (29 Jan 2025)Source for the ~12–13% consumer-Premium share, the ~50% two-year growth (coinciding with the tech-layoff wave), and the ~40% AI-feature attach rate.
- Engineering literature“Email Volume Optimization at LinkedIn” — KDD 2016LinkedIn Corp. authors formalize email as constrained optimization with “downstream sessions” as the positive outcome; 64.51% volume cut retained 98.16% of sessions. The objective function, in print.
- Engineering literatureAir Traffic Controller — LinkedIn Engineering (1 Mar 2018)Five delivery channels, “more than 1 billion notification requests a day,” delivery-time targeting, disable-probability modeling, and the credited “double-digit percentage increases in site-wide member engagement.” Companion: Concourse (25 May 2018), same blog — “engagement funnels… viral loops.”
- Engineering literatureOffline RL for notification decisions (arXiv, 2022)Sessions and Weekly Active Users as reward functions; +0.24% sessions at −1.73% volume. Succeeded by the 2025 Decision-Transformer paper (arXiv 2509.02458): +0.72% sessions. A decade-stable objective across three architectures, including patent US9420062 (delivery-time optimization, granted 2016).
- RegulatorIrish DPC fines LinkedIn Ireland €310 million (24 Oct 2024)Final regulatory finding: no valid legal basis for behavioral-advertising processing, plus transparency and fairness infringements; fines, reprimand, and a compliance order. Appeal status unverified at build time. Cited strictly for what it decided.
- Advocacy complaintnoyb — complaint to the Austrian DPA (5 May 2026)Pending GDPR Article 15 complaint alleging profile-visitor data is paywalled behind Premium. An allegation, not a finding; LinkedIn disputes it. Cited as independent identification of the access asymmetry in Exhibit 3, with coverage by The Register (5 May 2026).
- Legal settlementPerkins v. LinkedIn — $13M “Add Connections” settlementN.D. Cal., No. 5:13-cv-04303: reminder emails sent in members’ names without consent; settled with disclosure changes and stop-reminder functionality. LinkedIn denied wrongdoing. The historical floor of the notification record.
- Design archiveDeceptive Patterns (deceptive.design) — LinkedIn brand pageTen dated third-party entries, 2014–2026, including “more than 30 clicks to unsubscribe from all E-Mail notifications” (Jun 2026), the 64-menu email opt-out (Jun 2023), hard-to-cancel Premium (Sep 2024), and AI-training auto-opt-in (2024). User-submitted counts, consistent and dated.
- AcademicWheeler, Garlick, Johnson, Shaw & Gargano — LinkedIn RCT (AEJ: Applied, 2022)The randomized controlled trial behind the 70%→77% employment finding, with 12-month persistence. The strongest evidence in LinkedIn’s value case, and the reason the CVI leads this page.
- Litigation recordhiQ Labs v. LinkedIn (2017–2022)Five years of litigation over scraping public profiles, ending in a permanent injunction against hiQ and data deletion. Cited as evidence the professional graph is a controlled asset LinkedIn defends at cost — not as a consumer-harm case.
- Official price listLinkedIn Corporation — UK G-Cloud 14 pricing documentA rare official list price for quote-only products: Recruiter at £8,925/seat/yr, Talent Insights £9,400–£82,850/yr, “Work With Us” employee-profile ads £4,200. Proof LinkedIn can publish prices when procurement rules force it.
- JournalismScott Partnership — invite credits cut 250→50 (31 Mar 2026)The 2026 company-page invite-credit reduction beside the Premium Page tier retaining higher allocations — the cleanest current-year free-to-paid capability migration. Supporting 2026 pricing corroboration: ConnectSafely, Evaboot, PhantomBuster, Pin.com, HeroHunt, Leonar trackers (all dated 2026).
- JournalismTechCrunch — WVYP launch & redesign coverage (10 May 2007; 3 Jun 2013)Documents the semi-anonymized viewer tease as original 2007 equipment (“a product manager at Microsoft”) and LinkedIn’s own 2013 framing of the feature as appealing to “humans’ natural curiosity.” The nineteen-year provenance of Exhibit 2.
- JournalismContemporaneous coverage of the 2015–2017 migrations (The Next Web, Fox News, Social Media Today, ERE, practitioner blogs)Dated, contemporaneous documentation of the API lockdown, contact-export removal, commercial use limit, InMail refund inversion, and the 2017 tags/notes and Premium-filter moves into Sales Navigator. Each item on the timeline carries its own dated source in the underlying dossier.
- Company statementLinkedIn official blog — “Less is more: less email from LinkedIn” (27 Jul 2015)“For every 10 emails we used to send, we’ve removed 4”; complaints down ~50%. Corroborated by Fortune and Netimperative coverage the same week. Credited in full as voluntary self-correction — and read alongside the KDD paper that shows what the removed email had and hadn’t been buying.
This assessment is built on an eight-stream research dossier compiled 21 August 2026 — consumer tiers, B2B products, curiosity mechanics, notification architecture, paywall history, dependency economics, complaints and legal record, and a deliberately constructed defense brief — totaling roughly 30,000 words of sourced findings. It is deliberately not reproduced here. Where the dossier flagged a claim as unverified, corrected, or fabricated, that claim appears in What we refused to count or nowhere. LinkedIn’s own help pages block automated retrieval; where cited, their content is corroborated through mirrors and multiple independent 2026 guides, and labeled accordingly.
Final verdict
GENUINE INFRASTRUCTURE · METERED IN EVERY DIRECTION
LinkedIn works. Not in the shrugging, what-else-would-you-use sense — in the randomized-controlled-trial sense. Free use of this platform has been causally shown to help people get jobs, and the free tier is broad enough that a complete job search costs nothing. The 1.3 billion profiles, the recruiters who screen through it, the employers embedded in it: the network is real professional infrastructure, and this page scores it that way.
The finding is what LinkedIn built on top of that fact. Around a single hiring conversation it can charge the candidate for visibility, the recruiter for search, the employer for distribution, and a salesperson for reaching the same company — both directions of the messaging channel priced, the free opt-out itself a subscriber feature. Its curiosity architecture has been telling members someone looked at you and selling the name since 2007. Its privacy rule — watch anonymously and you forfeit being watched — dissolves on payment. And its notification systems are optimized, by the published account of LinkedIn’s own engineers across a decade of papers, for downstream sessions and weekly active users.
None of this required inference, which is what makes the page unusual. The S-1 said members ride free and enterprises pay. The 10-K says growth depends on member engagement. The engineering literature says the notifications are scored on return visits. The pricing pages say private browsing plus viewer intelligence is a paid bundle. CHI’s contribution is arithmetic and adjacency: putting the company’s own statements next to each other and scoring what they describe.
The profile is free. The paths between profiles are metered — in every direction.
LinkedIn’s value is real. That is precisely why the dependency, and what gets built on it, matters.