Every industry has a shadow curriculum of numbers that get repeated until they feel like laws. This registry is the index for our autopsy series. One row per statistic. Each row links the first known appearance we could verify on the live web, states the original sample where one exists, and grades the claim on our evidence ladder: Observed, Corroborated, Derived, Inferred, defined on our method page. Where a defined search located no primary source, the row is graded Bounded absence, the ladder's fifth class. It records the search (the linked autopsy, run November 9, 2026), not proof that no source exists. Full workings for most rows are in the linked autopsies.
How should you read this table?
Three rules. First, the evidence class grades the claim as originally published, not the way it circulates; many Observed rows circulate in mutated forms that would grade no better than Inferred. Second, "what it actually measured" is the load-bearing column: most abuse of these numbers comes from quietly swapping the measured quantity for a sexier one. Third, status is dated. Numbers rot. This table says November 9, 2026, and means it.
What is in the seed edition of the registry?
| # | The claim | First known appearance | Original sample | What it actually measured | Evidence class | Status, Nov 9, 2026 |
|---|---|---|---|---|---|---|
| 1 | Calling in 5 minutes vs 30 makes contact 100x more likely | InsideSales/MIT Lead Response Management study, 2007 | 6 companies, 15,000+ leads, 100,000+ calls, vendor's own dialer | Odds of making contact, by response time | Observed (vendor-funded, narrow sample) | Routinely misquoted as a close-rate claim |
| 2 | Qualification odds drop 21x from 5 to 30 minutes | Same 2007 study | Same as row 1 | Odds of qualifying a lead, not closing | Observed (same caveats) | Same misquote pattern as row 1 |
| 3 | 37% of companies respond to a web lead within an hour | HBR, The Short Life of Online Sales Leads, 2011 | 2,241 US companies, one audited test lead each | Response latency distribution | Corroborated (independent reconstruction) | Solid for 2011; latency landscape since changed |
| 4 | 23% of companies never respond to a web lead at all | Same HBR 2011 audit | Same as row 3 | Share of audited firms with no response | Corroborated | The least quoted and most damning number in the family |
| 5 | Average lead response time is 42 hours | Same HBR 2011 audit | Same as row 3 | Mean among responders within 30 days only; excludes never-responders | Corroborated, commonly misread | Quoted without the exclusion, which flatters it |
| 6 | Contacting within an hour makes qualifying ~7x likelier than the next hour | HBR 2011 companion analysis | 1.25M leads, 42 companies (29 B2C, 13 B2B) | Qualification likelihood ratios, no baseline rate given | Corroborated | Cannot be converted to revenue; often is anyway |
| 7 | Calling within 1 minute lifts conversion 391% | Leads360 study via Chief Marketer, Nov 27, 2012 | ~3.5M leads, vendor data | A lift with no stated baseline or conversion definition | Inferred at best as published | Circulates unattributed on vendor pages |
| 8 | Wins are 3x more likely at 10-60 minutes than under 10 | VanillaSoft/Telfer study, 2017-2018 | ~45M leads, 550 companies, vendor data | Win likelihood by first-response window | Observed, self-labeled preliminary | Promised full study never surfaced in our search |
| 9 | 48% of salespeople never follow up | No primary source; attributed to a "National Sales Executive Association"; debunked by VentureBeat, Aug 15, 2014 | None located | Nothing; no study found | Bounded absence | 11+ live pages still citing it this month |
| 10 | 80% of sales require 5+ follow-up contacts | Same family; nearest historical root is a 1942 chapter survey, sample under 40, per SMEI (accessed Nov 9, 2026) | Under 40 members, 1942, different question (calls per sale) | Nothing matching the modern claim | Bounded absence | Actively laundered through 2026 statistics roundups |
| 11 | 44% of salespeople give up after one follow-up | Same family, same non-source | None located | Nothing; no study found | Bounded absence | Travels bundled with rows 9 and 10 |
| 12 | Buyers are 57% through the purchase decision before engaging a supplier rep | CEB, The Digital Evolution in B2B Marketing, 2011, quoted and contextualized by LinkedIn, 2016 | CEB buyer research, 2011, updated 2013 | Average decision progress before supplier-rep engagement | Observed (owner later revised toward 65% per this trace) | Misused as "sales enters late"; owner disagrees |
| 13 | 67% of the buyer's journey is done digitally | SiriusDecisions, 2013; owner's correction by Megan Heuer, mirrored here (accessed Nov 9, 2026) | SiriusDecisions buyer research, 2013 | Share of journey interactions in digital channels | Observed, widely misread | Owner's correction is now offline at the original URL |
| 14 | Buyers are two-thirds to 90% through their journey before contacting vendors | Forrester blog language, 2012-2013; author's own hedge in a CustomerThink interview (undated, accessed Nov 9, 2026) | Directional synthesis, not a single measurement | A rhetorical range, per its author: "like any average, it's wrong" | Inferred, per the author herself | Original Forrester posts now redirect to category pages |
| 15 | Pipeline coverage should be 3x quota | No primary source; earliest candid account is Kellblog, Apr 19, 2013 | None; sales-floor folklore | Nothing; it is a convention, and a self-fulfilling one | Bounded absence (no empirical source located) | Replace with a derived multiple |
| 16 | Only 27% of leads ever get contacted | Ken Krogue, Forbes, Jul 12, 2012 | InsideSales.com in-house research, unpublished | Vendor-internal contact rate, sample unknown | Inferred (unpublished methodology) | Circulates as anonymous plural "studies" |
| 17 | Bad data costs the US $3.1 trillion per year | IBM estimate, 2016; methodology never published; trace at Analytics India Magazine, 2022 | Not disclosed | Unknown; no public methodology | Inferred at best | Quoted as settled fact across data-quality marketing |
| 18 | B2B data decays 70.3% per year (and the older 25-30% broker estimate) | John Coe, Biznology, Feb 13, 2015 | 1,200+ self-reported business cards, 2002 study updated 2009 | Self-reported annual changes to title, phone, address, email; combination math not shown | Observed inputs, Derived headline with undisclosed method | Both figures circulate undated as current benchmarks |
What patterns does the registry expose?
Four, so far. Vendor gravity: thirteen of eighteen rows were produced or promoted by companies selling the implied fix. Measure-swapping: contact odds become close rates, digital interaction share becomes sales avoidance, a lift with no baseline becomes a planning number. Date-stripping: the newest primary research in the table is from 2017, yet every one of these claims circulates today in present tense. And correction decay: the debunks and owner corrections (rows 9, 12, 13, 14) are consistently harder to find than the claims they correct, and two of the corrections have gone offline entirely while the myths stay up.
What gets a statistic into the registry?
Three criteria: it is widely quoted in sales, marketing, or revenue operations; it influences real decisions, staffing, budgets, or tooling; and its provenance is checkable, even if checking it reveals there is nothing there. Suggest an entry by the contact routes below. Submissions that arrive with a primary link jump the queue. This is the seed edition; rows will be added, regraded when new evidence surfaces, and never silently edited, with changes logged at the foot of the page.
Change log
| Date | Change |
|---|---|
| November 9, 2026 | Seed edition published with 18 entries. Rows 9, 10, 11 and 15 graded Bounded absence after searches run November 9, 2026. |
Every later change to a row, an evidence class or a status is added to this table with its date and reason. Rows are never edited silently.