Your domain tells every receiving server what to do with mail that claims to be from you. We read what it actually says.
SPF and DMARC are published in public DNS, and a record that is broken usually looks fine from the outside. We read yours, count the DNS lookups your SPF record actually requires against the limit of ten the specification sets, and tell you what a receiver is told to do with mail that fails. Free, in seconds, no account and nothing to cancel.
Every finding is read twice. If two tools disagree, it is not a defect.
Links decay quietly. A restaurant lets a domain lapse and somebody in another country re-registers it. A site builder disconnects a domain and serves its own error page. A platform shuts down and every link to it becomes a 404 on the same day. The directory keeps printing the link, and the first person to notice is usually the member, or a visitor who lands somewhere embarrassing.
We do not read category or alphabetical listing pages and call that a directory. Those pages omit some members and print some links that differ from the member's own page. We fetch each member page and take the link that page prints. Every candidate defect is then re-opened with a second, independent tool before it is written down.
Thirty-five accusations we did not make
A machine can flag a page as parked or expired by the words on it. We treat that as a list of candidates, never findings. On the most recent full directory read, that scan flagged 35 of 160 working member sites. Every single one was rejected on inspection: incidental wording inside working sites, ordinary brand redirects, and three pages that were simply small.
We also refuse to claim what we cannot see. Large hotel and restaurant brands run bot walls that answer 403 or serve a captcha to any automated request. A link behind one of those is reported as not readable from here, never as broken. In one 59-member directory, 18 of 58 links fell into that bucket and were left alone.
Most of it answers 200 OK. A link checker will tell you it is fine.
Expired domain, re-registered
The old address now serves an online casino or a crypto page, sometimes in another language.
checker catches it: no — answers 200Registrar parking
A "this domain may be for sale" lander. Often a tiny 200 that redirects in JavaScript.
checker catches it: noSite builder disconnected
The builder's own error page, branded with the builder's name.
checker catches it: sometimes — usually a 404Dead platform
A plain "page not found". Websites built with Google Business Profiles were switched off in March 2024; after 10 June 2024 every one of those addresses returns an error. That cause is documented by the platform, not inferred by us.
checker catches it: yes, if anyone is checkingName rot
The link works; the business is trading under a different name than the one printed. Nothing is broken except the fact.
checker catches it: noRedirect chain to an unrelated business
Seen once, in a single directory: the printed address forwarded through eight unrelated restaurants in five states before dying. Every tool we had used until that day called it a bot wall, because they report only where a link ends. We now keep the whole chain.
checker catches it: noThree directories have been read in full. The first: all 37 of its public categories, 597 member website links, of which 523 could be read and 11 are broken. The second: all 83 categories, 193 member pages, 177 published links, of which 165 could be read and 3 are broken — a better-kept directory than the first, and we told them so in those words. The third: all 433 member listings, 396 publishing a website, 394 unique addresses opened, 345 readable, 22 broken on the day we last checked. Three of that directory's broken links return a perfectly healthy 200 OK to any checker, and a fourth is published with no hostname at all.
Reading in full also killed one of our own theories: dead Google business.site links, which we expected to be common, appear just twice across the 1,168 links of the three full reads.
Counts, not rates.
We are not going to tell you that "X% of chamber directories are broken", because we have not audited a sample large enough or random enough for that sentence to mean anything. Each figure traces to a working ledger we keep, and every defect in it was confirmed twice.
| What was read | Links opened | Confirmed defects |
|---|---|---|
| Ten local chamber directories, member-page method | ~1,270 readable | 49, in seven of the ten |
| A statewide business association's index of local chambers | 260 | 17 |
| A national organization's Wisconsin chamber finder | 194 | 5 |
| An association index of independent inns | 86 | 0 |
The last row matters as much as the others. That index was audited in full and was clean, and we said so rather than stretching it into a reason to send someone an email. Three of the ten chamber directories were clean too.
One more finding worth stating: two different indexes listing many of the same organisations rot independently. Of a dozen confirmed defects on one, exactly one appeared on the other, and in at least one case the second index still carried a working address the first had lost. An audit of one index is not an audit of the other.
Public email-authentication records, straight from DNS.
Every domain can publish two short public records, SPF and DMARC, that tell a receiving mail server what to do with a message that claims to come from it. On 2 September 2026 we read those records for 21 Wisconsin chamber and business-association domains. We read published policy and nothing else: we did not touch anyone's mail, we did not attempt to forge a message and we would not, and we make no claim at all about DKIM.
The three that enforce are the useful part. They run the same directory software as most of the others, so this is not a case of one organisation being able to afford something the rest cannot. And p=none is not a fault: it is the monitoring setting DMARC is designed to start at. We report what is published. We do not tell anyone their mail can be forged, because we have not tested that and will not.
A single DNS query is not evidence. On our first pass, four of these domains appeared to publish no SPF record at all. On retry, three of the four publish one perfectly normally; the first answer had been a resolver failure dressed up as an absence. We now ask every name repeatedly, across two independent resolvers, and accept only a definite answer. One domain's SPF is unknown for exactly that reason, and one domain was left out because its own website did not answer when we went to confirm who it belonged to.
The full measured dataset — 35 chamber domains read on 3 September 2026, counts, checks and method is published, and it names nobody.
Count how much of its ten-lookup DNS budget an SPF record spends — free, and it expands every include and prints the chain. RFC 7208 caps those terms at ten and requires a permanent error above it.
Read any domain’s published DMARC and SPF records here — free, nothing to sign up for, and it names no organisation but the domain you type.
| What was read | Domains |
|---|---|
| Chamber and association domains read, organisation confirmed | 21 |
| Publish no enforcing DMARC policy | 18 |
of those, p=none — monitoring only | 11 |
| of those, no DMARC record at all | 7 |
Publish an enforcing policy (quarantine or reject) | 3 |
| Publish no SPF record | 1 |
Nobody.
This page names no chamber, no association and no member business, and it will not, whatever we find. Publishing a list of organisations with broken links is pressure, not reporting, and the people who would be embarrassed by it are usually one or two staff doing the work of five. What we name freely are the parking and gambling destinations themselves, the defect classes, and the counts above.
If you run a directory and want to know whether any row above is yours, email us and we will send you your own rows privately, free, whether or not you ever pay us anything.
Check any page yourself.
This runs on our server, not yours, and asks you for nothing. Paste the address of a page — a directory listing, a members page, a links page — and we open every outbound link it publishes, up to fifteen, read where each one actually lands, and re-open every failure a second time before calling anything broken. Same two-read rule as our paid audits.
The fifteen-link cap is a real limit, not a sales tactic: this runs on a free Cloudflare Worker, allowed fifty outgoing requests per request, and reading every link twice plus a DNS check spends them quickly. Most member directories put each member's website on the member's own page rather than the list page, so on those the fast check mostly finds social and platform links. Reading a whole directory member by member is the slow job — that one you email us about.One fixed price for the written audit. Everything else here is free.
Directory link audits, on request only. We no longer read a directory we were not asked to read. If you operate one, email rjhsignaltech@gmail.com with a link to it — that request is the permission — and you get back a file listing every member page we opened, the link it printed, where that link actually landed, and a verdict for each: defect, not claimed, or fine, with the evidence for the defects.
- We will not commit to work this system cannot finish on its own.
- We will not contact an organisation twice if the first message went unanswered. One email, ever.
- If you pay and then decide it was not worth it, say so and we refund it without asking why.
The mail authentication audit of one domain, every check written out with the clause it evaluates, emailed within one business day. What is in it.
Order the auditThe roster read: the same eight checks across every domain you send mail on behalf of, one row each, plus a summary of which ones publish no enforcing policy at all. What is in it.
Order the roster readThe record reader, the SPF lookup counter and the page checker cost nothing and ask you for nothing. If one of them was worth something to you there is a link, and anything from $5 is fine. Nothing is also fine.
Put something toward the work