Data Inventory, Retention, and Deletion
What SpoofKiller collects, why, how long it is kept, and how to export or delete it.
Updated
This page is the operational companion to the privacy policy: what SpoofKiller stores, why, for how long, which outside services touch it, and how to take it out or delete it. Where the two differ in wording, the privacy policy and the terms of service govern.
What SpoofKiller never receives
Data inventory
| Data | What it contains | Why it is collected | Retention |
|---|---|---|---|
| Account | Name, email address, a password hash (never the password) or a link to a Google sign-in, and whether the address is confirmed | Sign-in; sending alerts only to confirmed addresses | Until the account is deleted |
| Organization and members | Organization name, plan, report branding, members and their roles, pending invitations (token stored as a hash) | Access control and billing | Until the organization is deleted; invitations expire after 7 days |
| Domains | Domain name, verification and reporting tokens, DKIM selectors and brand terms you supply, latest score and policy | Monitoring and routing reports to the right domain | Until the domain or organization is deleted |
| DNS scan results | Public DNS records for the domain and the findings drawn from them; one score per day | The Protection Score, findings, and change detection | The 30 most recent full scans per domain; daily scores for the life of the domain |
| DNS change history | Previous and new values of monitored records, with the time detected | Alerting and audit of record changes | The plan’s history window |
| DMARC aggregate reports (parsed) | Reporting organization and its contact address, period, published policy; per sending IP: message counts, disposition, SPF and DKIM results, the domains involved, DKIM selectors | Analytics, sender identification, detection, and enforcement readiness | The plan’s history window |
| Original report files | The file as received (XML, gzip, or zip), with its name, size, and hash | Reprocessing and review of files that failed to parse | The plan’s raw-file window |
| Sending sources | IP address, reverse DNS, network and registry country, inferred provider with its evidence, your classification, notes, and classification history | Telling you who sends as your domain | Until the domain or organization is deleted |
| Findings and alerts | Detected problems with their evidence and status; alert records and delivery results | Notification and follow-up | Alerts, and findings once resolved: the plan’s history window. Open findings stay until resolved |
| Blacklist and lookalike results | Listing status per checked address; candidate lookalike domains with registration and DNS facts | Reputation and impersonation monitoring | Until the domain or organization is deleted |
| API keys | Name, prefix, scopes, last-used time, and a SHA-256 hash of the key | Authenticating API requests | Until the organization is deleted; revoked keys stop working at once |
| Audit log | Who did what and when: domains added and verified, classifications, role changes, key creation, exports, deletions | Accountability within your organization | 730 days |
| Processing errors | Why a report file was refused or failed to parse, with its name and size | Diagnosing ingestion problems | 90 days |
| Billing | Stripe customer and subscription identifiers, plan, and status. Card details are held by Stripe and never reach SpoofKiller | Subscriptions | Until the organization is deleted, subject to legal record-keeping |
| Contact forms | What you type into a contact or review-request form: name, email, company, domain, message | Replying to you | Kept as correspondence; ask us to delete it |
| IP intelligence cache | Reverse DNS, network, and registry country for IP addresses seen in reports | Avoiding repeated public lookups | Refreshed after 30 days. Shared across customers; contains no customer data |
The free public tools are separate. A lookup in the domain health check reads public DNS, and headers pasted into the email header analyzer are analyzed and not stored. Rate limiting for public pages uses a hash of the visitor’s IP address rather than the address itself, and the counters are discarded within a day or so.
Retention by plan
Two windows depend on the plan. A client workspace uses its agency’s plan.
| Plan | Reporting history | Original report files |
|---|---|---|
| Free | 7 days | 7 days |
| Starter | 90 days | 30 days |
| Pro | 365 days (12 months) | 90 days |
| Agency | 730 days (24 months) | 180 days |
- Reporting history covers parsed reports and their records, daily statistics, DNS change events, alerts, and resolved findings. A scheduled cleanup deletes whatever is older than the window. Date ranges in the dashboard and the API are limited to the same window.
- Original report files are deleted from storage once older than the raw-file window, which is shorter. The parsed data from a file outlives the file.
- The windows follow the current plan. After a downgrade, the next cleanup applies the shorter windows; after an upgrade, data accumulates up to the longer ones but data already deleted does not come back.
Plans and prices are on the pricing page.
How original report files are stored
Each file received by email, upload, or the API is stored before it is parsed, apart from the database of parsed results, in private object storage. Access is by the application only, using the storage credential; files are never given a public URL, and there is no link in the product that serves one directly from storage. Each file is recorded with its organization, domain, size, and SHA-256 hash, and is removed when the raw-file window passes or when its domain or organization is deleted.
Third-party processors and services
These are the outside services involved in running SpoofKiller, with what each one sees.
| Service | Role | What it handles |
|---|---|---|
| Vercel | Application hosting and private file storage | All requests to the application; original report files |
| Neon | PostgreSQL database | All stored account and monitoring data |
| Stripe | Payments and subscriptions | Billing contact and payment details, entered on Stripe’s own pages |
| Postmark | Email: outbound alerts and account mail; inbound aggregate reports | Recipient addresses and the content of mail SpoofKiller sends; report emails addressed to your reporting address |
| Cloudflare and Google public DNS (DNS over HTTPS) | Resolving DNS for scans and checks | The names being looked up: your domains, candidate lookalike names, blocklist query names. No account information |
| Team Cymru (IP-to-ASN, over DNS) | Network and registry country for sending IP addresses | The IP address being looked up |
| RDAP (public registry data) | Registration and expiration dates | The domain name being looked up: yours for expiration, a lookalike’s for its registration date |
| SpamCop and, optionally, Spamhaus (Data Query Service) | Blocklist checks on the Pro and Agency plans | Your domain name and the IP addresses of approved or self-hosted senders, as DNS queries |
| DataForSEO (optional) | WHOIS data when a registry offers no RDAP answer | The domain name being looked up |
| Google (optional) | Sign-in with Google, if you choose it | The identity exchange for your sign-in |
“Optional” services are used only where they are configured; Spamhaus is queried solely through its Data Query Service, since its public mirrors are not licensed for commercial use. None of the lookup services receives report contents or anything identifying your account; they see a name or an address, as any DNS resolver does.
Exporting your data
- Reports. Members with the Read-only User role or higher can download any of the seven report types as CSV, and as PDF on the Pro and Agency plans. Between them the DMARC Authentication, Suspicious Sender, and DNS Change reports cover authentication totals, every sending source, and the change history. Each download is recorded in the audit log.
- API. On the Pro and Agency plans, the REST API returns domains, scan results, daily DMARC series, sending sources, DNS records and changes, findings, and alerts as JSON.
- History limits apply. An export covers at most the plan’s reporting history, because older data has been deleted.
If you need something in a form these do not provide, ask us.
Deleting data
| Action | Who | What happens |
|---|---|---|
| Delete a domain | Owner or Administrator; confirm by typing the domain name | The domain and everything attached to it (reports, sources, findings, alerts, scan results, DNS history) are deleted. Its reporting address stops accepting reports. |
| Delete an organization | A direct Organization Owner; confirm by typing the organization name | Access ends immediately for every member and API key, and for an agency, for all of its client workspaces too. The data is permanently removed within 30 days by a scheduled cleanup. |
| Revoke an API key | Owner or Administrator | The key stops working on the next request. |
| Remove a member | Owner or Administrator | The person loses access to the organization, and to its client workspaces if access was inherited. |
Before deleting an organization with a paid plan, cancel the subscription from the billing portal. Remember also to remove the SpoofKiller address from your DMARC record’s rua tag. Reports sent to the address of a deleted domain are not accepted, but there is no reason to keep asking the world’s mailbox providers to send them. The DMARC checker will confirm the record afterward.
Reporting limitations
Knowing what the data cannot tell you is part of handling it properly.
- Aggregate reports are incomplete. They come only from receivers that choose to send them. Figures are observations from those receivers, not a complete record of all mail sent as your domain. See Understanding DMARC Reports.
- Reports arrive late. A report usually covers the previous day, so what you see is about a day behind the mail.
- Monitoring is scheduled. DNS is scanned every six hours; blacklist, lookalike, and expiration checks run daily. A change is detected at the next check, not when it happens.
- A failure is not an accusation. DMARC failures also come from forwarding and mailing lists. Findings state a confidence and their reasons for doubt.
- Provider and country are inferences. A provider name is an evidence-based estimate with a confidence figure, and the country is the registry country of a network, as explained in Sending Sources.
- DKIM selectors cannot be listed from DNS. Scans check the selectors you supply and common provider selectors.
Security measures
Aggregate reports are files sent by arbitrary mail servers on the internet, which makes them untrusted input. The measures below are the ones that bear on your data.
| Area | Measure |
|---|---|
| Hardened XML parsing | Any document containing a DOCTYPE or ENTITY declaration is refused before parsing, and the parser runs with entity processing disabled. This removes external-entity and entity-expansion attacks. Control characters are stripped from report text and field lengths are capped at 512 characters before anything is stored or displayed. |
| Decompression limits | Files are limited to 15 MB as received and 60 MB decompressed; a zip may hold at most 25 files; nested archives are not opened; a report may contain at most 100,000 records. The format is determined from the file’s content, never from its name or declared type. |
| Background processing | Decompression and parsing never happen inside a web request. Files are stored and queued, then processed by a job. |
| Tenant isolation | Every query for customer data is scoped to the caller’s organization, and a domain is loaded only if it belongs to that organization. Access comes from membership, never from knowing an identifier. Details are in Agency Workspaces, Clients, and Roles. |
| Ownership verification | Report data for a domain is accepted and shown only after DNS ownership verification. |
| Hashed keys and tokens | API keys, invitation tokens, and email-verification and password-reset tokens are stored only as hashes. An API key is shown once, at creation. Passwords are stored as hashes. |
| Audit log | Security-relevant actions are recorded with the acting user and time: domains added, verified, paused, and deleted; sender classifications; finding status changes; invitations and role changes; API key creation and revocation; report uploads and exports; alert pauses; organization deletion. |
| Rate limiting | Domain additions, verification attempts, uploads, exports, invitations, and API requests are rate limited. |
| Safe exports | Text in CSV exports that begins with a formula character is neutralized, so that report-derived strings cannot execute in a spreadsheet. |
| Alert email | Sent only to members whose email address has been confirmed. |
None of this makes a system invulnerable, and this page does not claim it. If you believe you have found a security problem, please tell us before telling anyone else.
