Updates
Every data run, newest first. Profiles show their own “last checked” date.
w2: research (full pass, with deeper fields)
3,000 candidates checked, 2,931 MSPs kept (1,999 US, 932 UK), 69 left out with a reason, 2,855 full profiles.
Second research wave: the next 3,000 MSPs from the research queue (2,044 US, 956 UK) read in full, now with the deeper fields: services, industries, stack, support hours, certifications, Companies House (682 of 932 UK matched) and corporate parents, LinkedIn size band, open roles from careers pages, job-board feeds and a job-listing search per MSP domain, Google rating, published prices, ownership and outsourcing. 2,931 stay MSPs and 2,855 are new indexable spec sheets; 44 sites need JavaScript and keep their basic profile, and 25 were re-classified as not MSPs (hosting, telecom, repair shops, a web agency). A check of 40 random profiles against the pages they cite found 89.5% of decided facts right; after-hours lines taken from phone-system features and advice copy, 24/7 from staff perks, founded years from testimonials or vendor copy, a reseller brand catalogue and integration logos read as stack, and blog pages read as industries were fixed before publishing, and the pilot and wave 1 were re-read with the same fixes (unchanged facts keep their date). The report, stack, hiring and pricing numbers now use all 4,744 full profiles. No full profile was older than 60 days, so none was due for re-checking.
expand: more seed sources + basic profiles (US + UK)
23,815 candidates checked, 1,652 MSPs kept (1,179 US, 473 UK), 22,103 left out with a reason, 0 full profiles.
More MSPs for the master list: the sites that old MSP domains now redirect to (rebrands and acquisitions; a relation stated on the new site is recorded as ownership), three more Google Maps queries in every US and UK seed place, Companies House smaller companies found by a Maps name search, JavaScript-only sites read with a rendering fetch, and dead sites retried once. Each query and source was tested on a sample first and kept only if it found new MSPs at a quarter of the seed's rate per dollar or better: five Maps queries, US towns of 5,000-9,999 people and "software company" listings fell short and were stopped. Sites that block crawlers stay out. New profiles are basic (noindex) until a research wave reads them in full.
deepen: deeper profiles (hiring, pricing, reviews, ownership, outsourcing)
1,813 profiles re-checked: 0 closed, 0 moved, 0 added, 1,692 updated.
All 1813 full profiles re-read from the cache for five new fields: open roles from public job listings (careers pages, the public feeds of the job boards they link to, and a job-listing search per MSP domain), Google Maps rating and review count, published per-user and per-device prices, ownership (acquisitions, mergers, private-equity backing reported by the MSP, and UK corporate parents on the Companies House register), and what the MSP says about outsourcing its help desk, NOC or SOC. 274 MSPs have a help desk, field technician or NOC role from the last 60 days (/hiring); 122 publish prices (/pricing). Tools named only in job ads are now kept apart from the site stack, 24/7 needs to sit next to support wording, and comparison or "we replace X" copy and testimonials no longer count. Facts that did not change keep their original date.
w1: research (full pass)
1,500 candidates checked, 1,493 MSPs kept (997 US, 496 UK), 7 left out with a reason, 1,487 full profiles.
First research wave: the next 1,500 MSPs from the research queue (1,000 US, 500 UK) read in full: services, industries, stack, support hours, hiring, certifications, Companies House (386 of 497 UK matched) and LinkedIn size band. 1,493 stay MSPs and 1,487 are now indexable spec sheets; 6 were re-classified on the full read (2 hosting, 1 telecom, 1 software vendor, 1 enterprise integrator, 1 not an MSP) and removed, and 1 site did not answer. A check of 40 random profiles against the pages they cite found 89% of facts right; staff counts read from client-size sentences, menu labels read as job titles and legal-notice pages read as a legal-sector page were fixed before publishing. The report and stack shares now use all 1,813 full profiles. No full profile was older than 60 days, so none was due for re-checking.
refresh-2026-11: monthly listing refresh
316 profiles re-checked: 0 closed, 0 moved, 0 added, 0 updated.
Home page of every full profile re-fetched (MSPScopeBot, 1 request/second per host): all 316 answered, none redirects to another domain, so none is marked closed or moved. No profiles added or changed: the pilot and the national seed were collected on 2026-11-07, one day before this check. The next research wave (w1) adds and deepens profiles from the queue.
seed: national seed + basic profiles (US + UK)
44,928 candidates checked, 10,555 MSPs kept (7,823 US, 2,732 UK), 34,373 left out with a reason, 0 full profiles.
Basic profiles (home, about, services, contact): listed and searchable, noindex until a full research wave passes the publish gate.
pilot: research (full pass)
487 candidates checked, 320 MSPs kept (169 US, 151 UK), 167 left out with a reason, 316 full profiles.
Pilot: 81 Google Maps searches (45 US places in 43 states, 36 UK places) via treg, polite crawl of each site, public DNS, Companies House, LinkedIn size band. Sites that block crawlers are left out, not bypassed.