# Wire release study, autumn 2026: method (written before any result was seen) Written 21 October 2026 by Higgins for PitchBud (marketing idea 262, approved by Bo the same day). This file is fixed once results collection starts. Any later change is added at the bottom under "Changes after collection began", with the reason and the date, and is reported on the public page. ## Question When a small business sends a press release through a public newswire service, how often does it lead to independent editorial coverage, meaning an article written by a journalist, editor or independent writer in their own words, within 30 to 90 days? ## Sample frame Releases that appear in the public newsrooms of newswire services and were published between 23 July and 21 September 2026, which is 30 to 90 days before the check date of 21 October 2026. Target: about 100 releases, spread across services so that no single wire is the whole study: about 40 from EIN Presswire (its public small-business channel), and about 60 from other public wire newsrooms that this server can read (for example PRWeb, Newswire.com, ACCESS Newswire, PRLog and 24-7 Press Release). The final count per service is whatever the newsrooms supply under the rules below, and is reported. ## Which releases count as small-business releases Included: the issuer is a privately held business, a self-employed professional, or an author or creator selling their own work, and the release announces something about that business (a product, service, opening, hire, award, book, event or milestone). Excluded, with the reason logged for each one: - publicly traded companies and their subsidiaries (a stock ticker or a listed parent); - government bodies, universities, schools, hospitals, charities, associations and political campaigns; - market-research report releases ("the X market will reach $Y billion by 20ZZ"); - releases not in English; - gambling, casino, adult, cryptocurrency token or investment promotions; - releases about a press release or newswire service itself; - a second release from an issuer already in the sample (one release per issuer). ## How releases are chosen For each newsroom, listings are read in date order, newest first within the window, and every release that meets the rules is taken until that newsroom's quota is filled. No release is skipped because it seems likely or unlikely to have been covered. Every release looked at is logged, with the reason for any exclusion. ## The coverage check Each release goes through PitchBud's own coverage check, the same code customers use (services/pickup_tracker.py as fixed on 21 October 2026): - the same search model and the same three searches: the exact headline in quotes; the business name with the first word of the headline; and the first sentence of the subtitle or summary when it is long enough; - one extra search for the business name on its own, so coverage that does not repeat the headline has a fair chance to be found (this favours finding coverage); - the same exclusions: the wire's own pages, PitchBud's list of free syndication networks, social networks, and the issuer's own website; - the same verification: each candidate page is fetched and must mention the business name (or the first six words of the headline) to count as a mention. ## Classifying every verified mention Each verified page is put into one of three groups: 1. Wire copy: the page reproduces the release itself, wholly or largely. This includes wire partner networks, the "press release" sections of news and TV sites, financial portals and automated feeds. It is detected by text overlap with the release and confirmed by reading. 2. Own or listing pages: the issuer's other sites and profiles, partners' sites, directories, listings, event calendars and shops. 3. Independent editorial coverage: an article, segment, column, review or post by a journalist, editor or independent writer that reports or comments on the news in their own words. Each one is read before it is counted. When a page is genuinely uncertain, it is counted as independent coverage, which favours the wires, and the doubt is noted. ## What is reported - Main figure: the share of releases with at least one independent editorial article, with a 95% confidence interval (Wilson method). - Also: the share with at least one wire copy, the average number of wire copies per release, and the kinds of outlets where independent coverage appeared. - Per-service figures only where a service has 30 or more releases in the sample. - The result is published whatever it shows, including if it goes against PitchBud's homepage claim. - Individual businesses and releases are not named on the public page. The full list is kept in this folder and can be shared with a journalist who asks. ## Known limits, stated in advance - Search-based checks miss coverage that is behind a paywall, in print or on air without a web page, or not yet indexed by the search engine. The study measures findable web coverage. - Coverage a business earned through its own pitching after the release is counted as coverage of the release; the study cannot tell how the journalist heard of it. - About 100 releases gives a margin of error of several percentage points; the confidence interval is reported rather than a single number alone. - The sample describes these services' small-business releases in this window, not all press releases. ## Changes after collection began 21 October 2026, about 10:47 UTC, before any release was screened or checked. Two clarifications of the sampling frame, made because of how the newsrooms are organised, not because of any result: - EIN Presswire has no public small-business channel. Its public day-by-day calendar (einpresswire.com/calendar/YYYY/MM/DD), which lists every release, is used instead, and the small-business rules above do the filtering. - Sample days are fixed as every Tuesday and Thursday in the window: 23, 28 and 30 July; 4, 6, 11, 13, 18, 20, 25 and 27 August; 1, 3, 8, 10, 15 and 17 September (17 days). On each sample day, listings are read newest first and the first qualifying releases are taken: two a day from EIN Presswire (34), and one a day from each of PRWeb, PRLog, 24-7 Press Release and Newswire.com (17 each, 68 in all), for 102 releases. If a newsroom cannot be read for a given day, that is logged and the slot stays empty. 21 October 2026, recorded at about 10:55 UTC. The screening instructions given to the collectors at about 10:47 UTC, before any release was screened or checked, also excluded large well-known corporations (roughly over 500 staff, or a household-name brand) even if privately held, and told collectors to include an issuer of uncertain size with an "uncertain size" note. This follows from the study being about small businesses; it was applied once in PRWeb (a national senior-living operator). It is recorded here because it was not in the first-written method. 21 October 2026, about 10:57 UTC, after seeing the verified mentions of the first release checked (one release of 102, and before any classification or count). Many pages that name a business were published before its release, such as earlier profiles of the same person, and some are the issuer's own releases sent through other wires. Neither can be coverage the release produced, so classification gains a fourth group and two rules: - 4. Earlier or unrelated mention: the page names the business but was published before the release date, or is about a different business or person with the same name. These are not coverage of the release. A page with no findable date is read and, when its timing is uncertain, counted as independent coverage, which favours the wires. - A press release by the same issuer on any wire, including a different release, is group 1 (wire copy) when it reproduces this release and group 2 (own or listing) otherwise. - Independent coverage is read generously: any third-party article in its own words about the business, published on or after the release date, counts, even if it does not mention the specific news. This also favours the wires. - Whether a third-party article was paid for is not judged unless the page itself says it is sponsored, advertising or a paid post; such pages go to group 2. Everything else that meets group 3 counts, which again favours the wires. The share of group 3 pages on sites that openly sell article placement is reported separately if it can be established. 21 October 2026, about 11:00 UTC, before any classification or count. The collectors applied "the release announces something about that business" differently, so the rule is settled and applied the same way on every wire, and picks are corrected in listing order: - Qualifies: news about the business itself, such as a new product, service, location, hire, award, book, event, partnership, milestone, or a report, survey or data set it publishes. - Does not qualify: general advice or commentary articles that announce nothing about the business (common search-engine-optimisation releases, such as cleaning or legal tips), and notices that only offer a spokesperson to journalists. - Also excluded: releases by or for a company connected to PitchBud's owner (for example books published by eBookIt.com), so the study has no interest in any sampled release. Where a pick changes, the next qualifying release in that day's listing replaces it. 21 October 2026, about 11:05 UTC. Dating rule for pages that show no publish date: the Internet Archive's earliest saved copy is used as evidence that the page existed by that date. A page first archived two or more days before the release is group 4. A page with neither a date nor an archived copy is read and, if its timing is still uncertain, counted as group 3. 21 October 2026, about 11:31 UTC. Automated rewrites: some sites restate press releases with software, crediting the piece to a bot or to the site itself (for example "GOAI" on GeneOnline, or several versions of one release on one site within a day), with no reporting of their own. They are copies of the release in other words, so they join group 1, and they are counted separately as "automated rewrites". A page is put here only on that kind of evidence; otherwise it is read as group 3. Group 3 pages are also tagged by kind (news article; mention in a list; profile on a site that sells placements; syndicated copy of an independent article) and reported by kind. Correction, 21 October 2026, 11:32 UTC: the clock times on the entries above were first written wrongly (as late as "13:20 UTC" when it was not yet 11:40 UTC). They have been replaced with times reconstructed from the files' own modification times; the order of the entries was always right. 21 October 2026, 11:39 UTC. Applying the automated-rewrite rule: pages under "/agp-article/" on EIN's network of sites carry the label AGP and the byline "By AI", so they are classed by rule. When the page title shares at least 40% of its topic words with the release title (ignoring the business name), it is a rewrite of this release (group 1); otherwise it rewrites another release by the same issuer (group 2). The readings already made by hand agree with this rule.