Methods

How the lab works.

The Pipeline Index is produced by an AI newsroom under a human editorial gate. This page is the standing protocol: where the evidence comes from, how articles are made, and exactly how we measure the funnels we spotlight. If we cannot defend a number on this page, we do not print it.

01 / Evidence

Three evidence classes. Nothing else prints.

Literature Published research from the real directories: arXiv, SSRN, NBER, OpenAlex, and open-access journals. We feature a paper only when we can read its full text, and figures use the paper's own reported numbers. Vendor whitepapers and abstract-only papers are disqualified.

Field The record of a decade of Tenbound client engagements, the podcast and conference archive, and thousands of CIENCE campaigns. Stated as field evidence, never dressed up as a study.

Intent telemetry De-identified graph8 intent data, always published with the n and the time window. When the week's pull is thin, the section is skipped, not faked.

Citation policy: every claim carries a citation. References include author, year, venue or identifier (arXiv, SSRN, NBER, DOI), a link, and the access date. Figures carry a source record (name, license, retrieval date, evidence class) checked into the repository with the data; a figure without provenance does not build.

02 / The newsroom

Five days, one gate.

The newsroom is a roster of specialized AI agents (research scout, intent analyst, profiler, pillar writer, data visualization, picture editor, fact-checker, copy editor) coordinated by a managing editor agent. The publish decision is human, every week, with the editor of record named in the issue.

Monday Plan The editor reviews the week's paper candidates (pulled from arXiv, OpenAlex, Semantic Scholar, NBER, and standing SSRN alerts), scores them on the published rubric, and sets the issue run sheet.

Tuesday Gather The newsroom pulls full texts, dossiers, and telemetry. Spotlight measurements fire at 10:00 in the subject's local timezone. Everything lands in the evidence vault.

Wednesday Make Articles are drafted to fixed templates. Every figure is built from the source's own numbers. Photos clear the licensing tree or the typographic card runs instead.

Thursday Verify The fact-check pass: every claim traced to its source, every figure to its data file, every photo to its license record. Anything unverifiable is cut, not shipped.

Friday Gate A human editor reads the assembled issue and approves or holds it. No issue publishes without the gate. The editor of record is named in each issue.

03 / The Pressure Test

The funnel instrument behind every spotlight.

Each issue spotlights two leaders and one company, and each spotlight includes a standardized measurement of the public inbound funnel: the Pressure Test, version PT-v1. Five dimensions of 20 points, one composite Pressure Score of 100, the same protocol every week so a cross-company record accumulates.

Clarity (20) Above-the-fold test at 1440px and 390px (what is it, for whom, what outcome), the timed 5-Minute Test, and pricing transparency.

Friction (20) Form field count, steps from primary CTA to a bookable calendar slot, and CTA focus above the fold.

Velocity (20) Speed-to-Lead: minutes from form submit to the first human response. Chat-to-human seconds, with AI disclosure noted. The Buyer's Mile: clicks and seconds from landing to a live human or a confirmed slot.

Follow-Through (20) A 9-day touch log after one submission: touches, channels, spacing, personalization depth, and whether a reply to their email reaches a human.

Craft (20) PageSpeed Insights mobile performance, the mobile experience rubric at 390px, and the trust surface: named customers, working legal pages, https hygiene.

Scores map to the maturity ladder as funnel maturity: 0 to 40 Manual, 41 to 60 Assisted, 61 to 80 Orchestrated, 81 to 100 Autonomous. The lens is the public funnel, not the whole organization, and we say so in every article.

04 / The measurement protocol

Disclosed, fair, and stricter than the precedent.

Response-time field studies are an established research genre: published secret-shopper studies have measured thousands of company funnels. Our protocol is stricter than that precedent: real identity, immediate disclosure at human contact, no consumed meetings, right of reply before publication, and absolute opt-out.

Spotlights are written to honor the work first: what the subject does well leads every teardown, with at least three items, and improvement notes are written as the lab's observations with the expected impact cited from the literature. Subjects see their full scorecard before publication.

Truthful identity, always. A real researcher name, a real tenbound.com email, a real phone number. No fake personas, no invented buying intent beyond researching for The Pipeline Index.
One submission per subject, fired around 10:00 on a Tuesday in the subject's headquarters timezone. Business hours, mid-week: the fairest window we can give everyone.
Immediate disclosure at first human contact: this is a disclosed research measurement by the Tenbound research lab, there is no purchase intent, and the response time has been recorded. We never advance a sales process and never consume a rep's meeting. Any auto-booked meeting is cancelled at once with the disclosure note.
Right of reply. Every spotlighted subject receives their full scorecard at least 5 business days before publication, with an invitation to correct facts, add quotes, and provide photos. Opt-outs are honored absolutely, through publication day, and never editorialized.
Evidence vault. Every published number traces to stored evidence: screenshots, screen recordings, email headers, and public report URLs, checked in the Thursday verification pass.

Questions, corrections, or an opt-out request: write to [email protected]. Corrections are noted in place in the archive, never silently edited.