Email deliverability: warm-up, address validation and archive copy
Sending high-volume email from a new sender domain or IP damages your reputation and lands you in the spam folder. For this, Pushizo offers warm-up plans, pre-send address validation and a BCC/archive copy…
Sending high-volume email from a new sender domain or IP damages your reputation and lands you in the spam
folder. For this, Pushizo offers warm-up plans, pre-send address validation and a BCC/archive copy
(docs/adr/0029-whatsapp-rcs-iki-yonlu-sms-eposta-isindirma.md §6). Settings live on the Deliverability
screen in the dashboard and under /v1/apps/{appId}/email-deliverability in the API (admin scope).
1. Warm-up plan
A plan defines daily send limits for a specific domain (domain: sends whose sender address is on this domain) or IP (ip: all sends from the app's email
channel).
POST /v1/apps/{appId}/email-deliverability/warmup-plans
{
"name": "Yeni alan adı",
"kind": "domain",
"target": "mail.ornek.com",
"curve": { "start": 500, "growth": 2, "target": 20000 },
"provider_shares": { "gmail": 0.4, "outlook": 0.3, "yahoo": 0.1, "other": 0.2 },
"timezone": "Europe/Istanbul"
}- Limits are given directly with the
daily_limitsarray (1–120 days) or with acurve(start × multiplier → target). provider_sharessplits the daily limit across recipient providers (summing to 1). Giving large providers such as Gmail and Outlook separate shares makes gradual reputation building easier.- A domain plan takes precedence over an IP plan; there can be only one plan for the same kind + target (
409). - Only marketing sends (campaigns and journeys) are limited; transactional emails sent via the API (password resets, order confirmations) are never delayed.
- Deliveries that do not fit the day's remaining allowance are spread over the following days, up to each day's limit, and distributed across
08:00–20:00 local time on the target day. Even if the campaign shows "sent", these deliveries wait as
pendingwith a futuresend_after. - Pausing the plan (
PATCH … { "status": "paused" }) lifts the limits; when the curve finishes, the plan becomescompleted. - Plan details and day × provider usage:
GET /v1/apps/{appId}/email-deliverability/warmup-plans/{planId}and…/{planId}/usage?days=30.
Recommended start: on a new domain, 200–500 emails a day to your most engaged recipients; roughly double every day to reach your target volume in 2–4 weeks. Verify your SPF, DKIM and DMARC records beforehand with Channels → Email → Verify DNS.
2. Pre-send address validation
Validation is on by default (MX and disposable domain checks; role accounts allowed). Apps that have never
changed the setting use this default; it is off only if "enabled": false has been set explicitly. When on, every
email delivery is checked before sending:
| Check | Setting | Result |
|---|---|---|
| Syntax | always | invalid addresses are not sent |
| Disposable domain | block_disposable | mailinator.com etc. are blocked |
| Role account | block_role | shared addresses such as info@, destek@ are blocked |
| MX record | check_mx | domains that do not accept mail are blocked; DNS errors do not block |
| Third-party verifier | verifier (ZeroBounce, NeverBounce) | a definite "invalid" is blocked; "unknown" (catch-all) is sent |
PUT /v1/apps/{appId}/email-deliverability
{
"validation": { "enabled": true, "check_mx": true, "block_disposable": true, "block_role": false },
"verifier": { "provider": "zerobounce", "api_key": "…" }
}- A blocked delivery is suppressed with reason
invalid_address; the event details state which check rejected it. - Checks run with limited concurrency (8) and are capped by a 20-second time budget per dispatch job: if the budget runs out, the remaining addresses are sent unchecked (validation never delays delivery; clearly bad addresses are suppressed anyway through provider feedback). MX and verifier results are cached, so the second send is fast.
- The verifier API key is stored with envelope encryption; responses only return
has_key; if left empty, it is kept. - Test tool:
POST …/email-deliverability/verify-address { "address": "ad@pushizo.com" }runs all checks and returns the result (ok,reason, checks run). - An address found invalid at send time is marked permanently (subscription
invalid); for an address that passed a full check in the last 7 days, DNS/verifier are not queried again.
3. BCC / archive address
If bcc_address is set, a blind copy of every email goes to this address (compliance and archiving); null removes it. Supported on SMTP, Amazon
SES and Resend channels.
4. Monitoring
The deliverability screen shows each plan's limit/usage for today, provider buckets and sends deferred to later days. Track bounce and complaint rates on the analytics screen; if the rates rise, pause the plan and slow down the curve.
5. Volume spike protection
If volume_spike.multiplier (2–20) is set, the portion of campaign and journey email that exceeds this multiple of the last 7-day average is moved to the next UTC day. If the average is below floor, that number is used as the baseline (default 50). An empty multiplier disables the protection. Transactional email sent via the API is never delayed. The deferred count is separate from the warm-up counter.
6. Google Postmaster
postmaster.domain and a write-only access_token are saved. POST …/email-deliverability/postmaster/sync reads yesterday's reputation from Google. If there is no token, if Google returns 401/403, or if there is no record (404), no figure is produced. In the dashboard, it is on the Deliverability screen.
7. Keeping the bounce rate low
Gmail, Yahoo and email providers (SES, Resend) throttle senders whose bounce rate exceeds 2%, and above 5% this can lead to account suspension. Pushizo also automatically suspends the email channel if the bounce rate is > 5% over ≥ 500 sends in the last 7 days (the suspension email and the channel screen link to the list cleaning tools). Layers to keep the rate below 2%:
- Checks at registration and import time. The SDK, the public/internal API, user/subscription endpoints and CSV / legacy provider
imports immediately check email addresses against syntax, disposable domain and (if enabled) role account rules.
An invalid address is saved but excluded from sending (subscription
invalid,email_check.result = "invalid"). For addresses that pass, the MX check (and ZeroBounce/NeverBounce, if configured) completes in the background (email-check-sweep, every 2 minutes; immediately after an import finishes). The import report providesrejected_emailsandrejected_email_reasonscounts; in the error CSV, rejected rows appear asemail_rejected:<reason>. - Hard / soft bounce hygiene. A hard bounce immediately adds the address to the suppression list (
bounce) and marks the subscriptioninvalid; it is not reactivated even if the same address is registered again. If the same address gets soft bounces on 3 different sends within 30 days (full mailbox, temporary rejection), it is suppressed like a hard bounce. - List cleaning. Validate list on the deliverability screen validates all sendable email subscriptions
in bulk, rate-limited (50 addresses per second) and resumable; progress and results (valid /
invalid by reason / unknown) appear on screen. Addresses marked invalid are already excluded from audiences;
Suppress invalid excludes them from sending and adds them to the suppression list (
manual, sourcelist_hygiene). Endpoints (session,channels.manage):GET /internal/v1/apps/{appId}/email-deliverability/hygiene,POST …/hygiene-runs,GET …/hygiene-runs/{runId},POST …/hygiene-runs/{runId}/suppress. Starting a run and suppressing are written to the audit log (email_hygiene.started,email_hygiene.suppressed). - Send guard. When a campaign containing email is sent or scheduled, the app's observed invalid rate (the greater of the list validation result and
the last 30-day bounce rate) is applied to never-validated / unknown addresses in the audience.
If the estimated rate exceeds the threshold (
max_invalid_ratio, default0.05) and the audience has ≥ 50 recipients, the request is rejected with409 HS_CONFLICT(details.reason = "email_invalid_ratio"). The dashboard shows a warning and a Validate list first link. An owner/admin (or an API key withadminscope) can proceed with?force=true; this is written to the audit log ascampaign.email_guard_overridden. Audience risk summary:GET /v1/apps/{appId}/campaigns/{campaignId}/email-risk. - Monitoring. The deliverability screen shows the daily bounce rate for the last 30 days against a 2% target line, the breakdown by type (hard, soft,
complaint, blocked before sending) and provider reason, and the recipient domains with the most bounces
(not addresses) (
GET /internal/v1/apps/{appId}/email-deliverability/bounces?days=30). The campaign builder shows the 7/30-day bounce rate for email sends.
Recommended steps for an app with a high bounce rate: (1) Pause campaigns. (2) Run Validate list and wait for it to finish. (3) Suppress invalid. (4) Fix the source of invalid addresses (form, import); use double opt-in. (5) Resume sending with a warm-up plan, starting with your most engaged recipients.
The email channel — SMTP, Amazon SES and Resend providers, sender domain verification, content fields (MJML/HTML, Liquid), open/click tracking, List-Unsubscribe and inbound feedback webhooks.
Drag-and-drop email editor
Dashboard → Messages → Email Editor (/{org}/email-designs). Decision record: ADR-0027.