Sending Sources and Classification
How senders are identified, what confidence means, and how to classify them.
Updated
A sending source is one IP address that receivers saw delivering mail with your domain in the From line. DMARC reports identify sources by address and nothing else, which is about as helpful as identifying a caller by the exchange they dialed from. SpoofKiller adds what public data can support: the likely provider, the network, and an approximate country. You add what only you know, which is whether the source is supposed to be there.
The evidence
For each new source, SpoofKiller collects five kinds of evidence. None of it comes from your mail; all of it is public data or already in the reports.
| Evidence | Where it comes from | What it can and cannot tell you |
|---|---|---|
| Reverse DNS | A PTR lookup on the IP address | The hostname the network operator assigned. Strong when it falls under a provider’s own domain; absent for many addresses. |
| Network (ASN) | Public IP-to-ASN data, queried over DNS | Which network announces the address and that network’s name. Says who carries the traffic, not which customer sent it. |
| Country | The registry country in the same IP-to-ASN data | Where the announcing network is registered. Approximate. It is not the location of a server or a sender. |
| Authenticated domains | The SPF and DKIM results in your aggregate reports | The domains that passed SPF (the envelope domain) and DKIM (the signing domain) for mail from this source. A provider’s own bounce or signing domain is a strong signal. |
| DKIM selectors | The same reports | Some providers use recognizable selector names. A weak signal, used only to support other evidence, never alone. |
IP facts (reverse DNS, network, country) are cached for 30 days and shared across all accounts; they describe the internet, not any customer. Enrichment runs in the background shortly after a report is processed, so a brand-new source may show as unidentified for a short while.
Provider identification and confidence
The evidence is compared with a catalog of 24 sending providers. Each kind of evidence that matches a provider carries a weight, and the weights combine into a confidence score from 0 to 100.
| Matching signal | Weight | Confidence if it is the only signal |
|---|---|---|
| Reverse DNS falls under the provider’s domain | 0.80 | 80% |
| Reverse DNS falls under your own domain (“Your mail server”) | 0.75 | 75% |
| An SPF or DKIM domain that passed belongs to the provider | 0.70 | 70% |
| The address is announced by a network strongly associated with the provider | 0.35 | 35% |
| A DKIM selector typical of the provider | 0.15 | Not used alone |
Signals combine as independent pieces of evidence: the chance that all of them are wrong is the product of the chances that each is wrong.
confidence = 1 − (1 − w₁) × (1 − w₂) × … capped at 98%
reverse DNS + authenticated domain: 1 − 0.20 × 0.30 = 94%
network only: 35%The provider with the strongest combined evidence is shown, with its confidence and the specific evidence lines (“Reverse DNS mail-xx.example.net belongs to …”). Confidence never reaches 100, and the reason is the point of this whole section:
An IP address alone is not proof
The catalog currently recognizes:
- Mailbox providers: Google Workspace, Microsoft 365, Zoho Mail, Rackspace Email, Fastmail, Proton Mail, Yahoo Mail
- Transactional email: Postmark, SendGrid, Amazon SES, Mailgun, SparkPost
- Marketing email: Brevo, Mailchimp, Klaviyo, Constant Contact, Shopify, ActiveCampaign
- CRM and support: HubSpot, Salesforce, Zendesk
- Email security gateways: Proofpoint, Mimecast
- Hosting: GoDaddy Email
A source that matches none of these is shown as unidentified, with its reverse DNS and network so you can work it out. Unidentified does not mean hostile; plenty of legitimate mail comes from self-hosted servers and smaller services. Provider identification is listed among the features of the Starter plan and above on the pricing page.
The five classifications
Every source has a classification. New sources start as Unknown. Analysts, Administrators, and Owners can change it, optionally with a note, and every change is kept in the source’s history with who made it.
| Classification | Use it when | What it changes |
|---|---|---|
| Approved | The source is yours: your mail platform, your invoicing tool, your newsletter service. | Never flagged as a potential unauthorized sender. If it fails DMARC on more than 20% of at least 20 messages in a report, you get an “Approved sender failing authentication” finding at high severity. Counts as a legitimate sender for enforcement readiness. Open unfamiliar-source findings for it are resolved. |
| Unknown | Nobody has decided yet. The default. | Eligible for detection. Unknown failing sources with meaningful volume hold enforcement readiness at yellow until classified. |
| Under Investigation | Someone is finding out whose it is. | Treated like Unknown by detection. Any source in this state keeps enforcement readiness at yellow, on the reasoning that you should not tighten a policy mid-inquiry. |
| Unauthorized | You are confident the source is not yours and should not be sending as your domain. | Raises the confidence of spoofing findings for it by 25 points. Never treated as a legitimate sender. Risk level is high whenever it has failing mail. |
| Ignored | You have looked and it does not need attention, such as a known forwarder. | Never flagged. Not treated as a legitimate sender that must pass. Risk level is informational. Open unfamiliar-source findings for it are resolved. |
Classification records your judgment. It does not block mail; SpoofKiller is not in the delivery path. What receivers do with mail from an unauthorized source is governed by your DMARC policy, which is why the classifications feed enforcement readiness.
Rules that remember a decision
Large providers send from many addresses, and classifying them one at a time is a job nobody should have. When you classify a source you can ask SpoofKiller to remember the decision as a rule:
| Rule | Matches | Scope |
|---|---|---|
| This IP, this domain | The exact IP address | The one domain |
| This IP, whole organization | The exact IP address | Every domain in the organization |
| This provider, this domain | Any source identified as the same provider | The one domain |
| This provider, whole organization | Any source identified as the same provider | Every domain in the organization |
- A rule is applied at once to matching sources that are still Unknown, and to new sources as they are identified.
- Rules never overwrite a classification someone set by hand. Only Unknown sources are touched.
- When several rules match, the most specific wins: a domain rule beats an organization rule, and at the same scope an IP rule beats a provider rule.
- Saving a rule replaces any earlier rule with the same match and scope.
- A provider rule needs an identified provider; it is not offered for unidentified sources.
A provider rule approves a provider’s infrastructure, not your account on it. Other customers of the same provider send from the same addresses, so an Approved provider still has to authenticate as your domain to pass DMARC. That is the protection; the rule only saves clicks.
Risk levels
Each source shows a risk level for the period you are viewing. It is derived from the classification, the share of the source’s mail that failed DMARC, and whether a provider was identified.
| Classification | Condition | Risk level |
|---|---|---|
| Ignored | Always | Info |
| Unauthorized | Any failing mail | High |
| Unauthorized | No failing mail | Medium |
| Approved | More than 20% failing | Medium |
| Approved | Otherwise | Info |
| Unknown or Under Investigation | More than 50% failing, provider not identified | High |
| Unknown or Under Investigation | More than 50% failing, provider identified | Medium |
| Unknown or Under Investigation | More than 5% failing | Medium |
| Unknown or Under Investigation | Otherwise, provider not identified | Low |
| Unknown or Under Investigation | Otherwise, provider identified | Info |
How sources become findings
When a report is processed, each source in it that is not Approved or Ignored and has mail failing DMARC is assessed for possible impersonation. The assessment produces a confidence figure: confidence that this is impersonation rather than forwarding or misconfiguration. It starts at 35 and moves with the evidence.
| Raises confidence | Points | Lowers confidence | Points |
|---|---|---|---|
| The source has not been seen before | +15 | The receiver noted a forwarding or mailing-list override | −30 |
| It is on a network new to this domain | +10 | A DKIM signature for your own domain was present but did not verify | −20 |
| It matches no recognized provider | +10 | The source belongs to a large mailbox provider, where forwarding is common | −15 |
| No DKIM signature was present at all | +10 | It matches a known sending service at 70% confidence or more | −10 |
| SPF authenticated an unrelated domain | +5 | Fewer than 5 failing messages | −10 |
| 100 or more failing messages | +10 | ||
| You marked the source Unauthorized | +25 |
The result is held between 5 and 95. At 40 or above a finding is raised; at 60 or above it is titled “Potential Domain Spoofing Detected,” and below that “Unauthenticated mail from an unfamiliar source.” Severity is high at 75 or above with at least 50 failing messages, medium at 60 or above, and low otherwise. The finding lists both the reasons it was flagged and the reasons for doubt. How findings turn into email is covered in Alerts and Notifications, and the reasoning behind the rules in email spoofing detection.
A DMARC failure is not proof of spoofing, and the scoring is built around that. The commonest innocent explanations are covered in DMARC Alignment Explained. To trace one particular message, paste its headers into the email header analyzer.
