---
title: Policies
description: The sending standards, email categories, and content rules every Mailfully account is held to, what happens when an account breaches them, and how to appeal.
---

> **For AI agents:** the complete documentation index is at [llms.txt](/docs/llms.txt). Append `.md` to any page URL for its markdown version.

Mailfully sends on shared infrastructure. One account with a bad list degrades delivery for everyone behind the same IPs and domains, so the standards below are enforced automatically and without prior notice.

Every suspension can be appealed and is reviewed by a person. See [Appeals](#appeals).

## Sending standards

Two important metrics can lead to an account being suspended.

| Metric | Threshold | Definition |
|---|---|---|
| Complaint rate | **0.07%** | Spam complaints ÷ delivered emails |
| Bounce rate | **4%** | Hard (permanent) bounces ÷ accepted emails |

Cross either one and Mailfully may suspend the account **immediately and without notice**. 

Soft bounces do not count toward the bounce rate. Only permanent failures do, matching what lands on your [suppression list](/guides/suppressions).

### How the rates are measured

- On a **rolling 7-day window**, recalculated continuously.
- Only after **500 sent emails** in the window. Below that floor, rates are informational and no automatic action fires. Manual review still applies to obvious abuse.
- Live sending only. [Test mode](/guides/test-mode) traffic is excluded.

Both rates are visible in the dashboard and through [`GET /v1/analytics/daily`](/api-reference/analytics/daily) and [external reputation](/api-reference/analytics/external-reputation). Watch them; do not wait to be told.

### Maintain your lists

- Delete suppressed addresses. Never import a suppressed address.
- Never send to an address you have not confirmed. Signup typos are a leading source of hard bounces.
- Validate addresses at collection time, not at send time.
- Warm new domains gradually rather than jumping volume significantly in one day.

## Email categories

Every email sent through Mailfully must be classified as **transactional** or **marketing**. Send a reserved `category` tag on every message:

```json
{
  "from": "receipts@yourdomain.com",
  "to": ["ada@example.com"],
  "subject": "Your receipt",
  "tags": [{ "name": "category", "value": "transactional" }]
}
```

Accepted values are `transactional` and `marketing`. Nothing else.

| Category | What it covers |
|---|---|
| **Transactional** | Triggered by a specific action the recipient took, sent to one recipient, and containing information they need: receipts, password resets, order and shipping updates, security alerts, account notices, one-time codes. |
| **Marketing** | Anything whose primary purpose is commercial: newsletters, product announcements, promotions, re-engagement campaigns, onboarding drips, digests, event invitations. |

The test is the message's **primary purpose**, not the system that triggered it. A receipt with a discount code stapled to the footer is marketing. An automated drip is marketing even though it is automated. If a message would be unwelcome to someone who did not ask for it, it is marketing.

<Note>
Misclassifying marketing email as transactional to dodge a plan limit counts as quota circumvention and is grounds for suspension.
</Note>

### Free plan allowances

| Category | Included per month |
|---|---|
| Transactional | 4,000 |
| Marketing | 1,000 |
| **Total** | **5,000** |

The two allowances are separate and do not pool. Exhausting the marketing allowance does not release unused transactional volume, and the reverse is also true. The free plan hard-stops at its allowance with no overage; see [Rate limits and quotas](/concepts/rate-limits) for how the quota errors surface.

## Legal compliance

You are the sender. You are responsible for the legality of every message you put through the API, in every jurisdiction your recipients are in.

### CAN-SPAM

Every commercial message must:

- Use accurate `from`, reply-to, and routing information. No forged headers, no misleading domains.
- Carry a subject line that reflects the contents of the message.
- Identify itself as an advertisement where the recipient has not clearly opted in.
- Include a valid physical postal address.
- Offer a clear, working way to opt out.
- Honor opt-outs within **10 business days**, and keep honoring them. Selling or transferring an address that has opted out is prohibited.

You remain responsible when someone sends on your behalf.

<Warning>
Mailfully may suspend an account **without notice** for CAN-SPAM violations. CAN-SPAM enforcement does not wait for a complaint threshold. One confirmed violation is enough.
</Warning>

Other regimes apply on top of CAN-SPAM depending on your recipients, including GDPR and ePrivacy in the EU, CASL in Canada, PECR in the UK, and the Australian Spam Act. Several of these require opt-in consent, which CAN-SPAM does not. Meet the strictest rule that applies to your list.

### Bulk sender requirements

If you send marketing email at volume, the receiving networks impose their own rules and Mailfully enforces them:

- **Authenticate:** SPF and DKIM must pass, and the sending domain needs a published DMARC record. [Verify your domain](/guides/verify-a-domain) before sending.
- **One-click unsubscribe:** marketing messages must carry `List-Unsubscribe` and `List-Unsubscribe-Post` headers, and the unsubscribe must complete without a login or a confirmation step.
- **Unsubscribe processing:** within 2 days.
- **Complaint ceiling:** under 0.3% as reported by Google Postmaster Tools, a separate ceiling from Mailfully's own 0.07%.

## Permission and lists

Send only to people who asked to hear from you.

**Prohibited outright:**

- Purchased, rented, leased, or exchanged lists, regardless of what the seller claims about consent.
- Scraped or harvested addresses.
- Appended addresses, including addresses guessed from a name and a domain.
- Cold outreach to people who have no relationship with you.
- Addresses carried over from a different company, product, or brand than the one the recipient signed up for.

**Required:**

- A record of when and how each recipient opted in, producible on request.
- A working opt-out in every marketing message.
- Suppression of every address that bounces hard or complains. Mailfully does this for you; do not work around it.
- Re-permission before mailing a list that has gone longer than 12 months without contact.


## Prohibited content and uses

You may not use Mailfully to send:

| Category | Examples |
|---|---|
| **Spam** | Unsolicited bulk email of any kind |
| **Phishing and fraud** | Credential harvesting, impersonation of a person or brand, invoice fraud, fake shipping or security notices |
| **Malware** | Viruses, ransomware, exploit code, or links to any of it |
| **Illegal content** | Anything unlawful where the sender or the recipient is, or content promoting it |
| **Harassment** | Threats, intimidation, doxxing, or coordinated targeting of an individual |
| **Sexually explicit material** | Adult content, escort services, and related solicitation |
| **Regulated goods** | Controlled substances, prescription pharmaceuticals sold without a prescription, tobacco, vaping products, weapons and ammunition |
| **High-abuse financial offers** | Payday and high-interest loans, debt relief and credit repair, work-from-home and get-rich-quick schemes, multi-level marketing, forex and binary options |
| **Cryptocurrency promotion** | Token sales, airdrops, and trading signals. Account and security notices for a licensed exchange are fine |
| **Gambling** | Casinos, sportsbooks, and lotteries, except where you hold a license valid in the recipient's jurisdiction |
| **Deceptive claims** | Miracle cures, unsubstantiated health or income claims, fake countdowns and fake scarcity |

You also may not:

- Send from a domain you do not control, or forge another party's identity in the headers.
- Open multiple accounts to work around a quota, a suspension, or a plan limit.
- Resell raw Mailfully sending capacity as a standalone email service.
- Use the API to test, probe, or validate address lists you do not intend to mail.
- Send data the platform is not built to carry. Mailfully is not HIPAA-eligible and is not in PCI scope: no protected health information, no full card numbers, no government identifiers in message content.

A few categories above turn on who you are rather than what the message says. A regulated but lawful business, such as licensed gambling, a licensed pharmacy, or regulated finance, can be approved case by case. Ask before you send.

<Note>
Mailfully may refuse or stop any sending that threatens platform deliverability, including sending that breaks no specific rule listed here.
</Note>

## Security and account hygiene

- Keep API keys server-side. Never ship them in a browser bundle, a mobile app, or a public repository.
- [Rotate keys](/api-reference/api-keys/rotate) on a schedule and revoke them when someone leaves.
- Scope keys to what they need. A key that only sends should not read suppressions.
- [Verify webhook signatures](/guides/webhooks) on every delivery.
- Report a suspected key compromise to security@mailfully.com immediately. Sending from a stolen key is still attributed to your account, and the resulting complaint rate is still yours.

## Enforcement

What happens depends on what was breached.

| Action | What it means | When |
|---|---|---|
| **Warning** | Email to the org owners and a dashboard banner. Sending continues. | Metrics trending toward a threshold |
| **Throttle** | Daily cap reduced. Sending continues at a lower rate. | Sustained poor metrics below the threshold |
| **Domain pause** | One sending domain refused. The rest of the account keeps sending. | A threshold crossed on one domain |
| **Suspension** | All sending refused. The API and your data stay readable. | A threshold crossed account-wide, or a content or legal breach |
| **Termination** | Account closed. | Repeat or severe abuse, or fraud |

The ladder is not a sequence you are entitled to. Complaint and bounce breaches, CAN-SPAM violations, phishing, and malware go straight to suspension with no warning step.

### What suspension looks like

Sends from a suspended account fail with `403` and error type `forbidden`:

```json
{
  "error": {
    "type": "forbidden",
    "message": "This account is paused."
  }
}
```

Do not retry-loop a `403`. It will not clear on its own; see [Errors](/concepts/errors).

Queued and scheduled messages stay queued. Read routes stay available so you can export message history, events, and suppressions. Webhooks continue to deliver events for mail already in flight. Billing continues through a suspension; no credit is issued for the period.

## Appeals

Every suspension can be appealed. A person reads every appeal; none are decided by automation.

Email appeals@mailfully.com from an address on the account, and include:

1. Your org ID.
2. What you believe triggered the suspension.
3. Where the affected list came from and how consent was collected.
4. What you have changed so it does not happen again.

We respond within **1 business day** and aim to resolve within 3. Sending stays off during the review. If the suspension was ours to fix, it is lifted and nothing is charged for the downtime.

Reinstatement usually carries conditions: a reduced daily cap, a clean re-permissioned list, or a period of monitored sending before full volume returns. There is no reinstatement fee.

<Note>
Contest the facts, not the thresholds. The most useful thing an appeal can contain is evidence about where the list came from.
</Note>