Skip to main content
MethodMarketingContent, email & socialIntermediate

Stop your emails landing in spam (SPF, DKIM, DMARC)

A deliverability plan for a domain that sends marketing and transactional email: which mailbox-provider rules apply (Gmail, Yahoo, Outlook.com), the exact SPF, DKIM and DMARC records to publish with an SPF lookup count, a DMARC rollout from none to quarantine to reject, one-click unsubscribe, a warm-up schedule, monitoring in Postmaster Tools, and what to do when mail lands in spam.

Use this task 5 quick questions, then open it in ChatGPT, Claude or Gemini.

This is a practical, copy-paste process for your newsletters or receipts land in spam, or bounce with an error code. Use it when you want a clear structure, then add your own details.

Verified on 2026-11-0424 sources cited

Quick start

1. List every tool that sends email from your web address: your mailbox, newsletter tool, website forms, shop and app.
2. Have each tool send an email to a Gmail address that is not your own work address. Open it, click More, then "Show original". SPF, DKIM and DMARC should all say PASS.
3. Keep exactly one SPF record per domain. SPF is a line of text in your domain settings (DNS) that lists who may send for you. It may use 10 lookups at most: each include: counts as one, and so does every include hidden inside it.
4. Turn on DKIM (a digital signature) in every tool, using your own domain, not the tool’s.
5. Add a DMARC record (your instruction for failed emails) starting at "p=none" with reports to a mailbox you read. Tighten it only when the reports look clean.
6. Keep spam reports under 0.3% of the emails that reach the inbox (aim for under 0.1%); Gmail and Yahoo require this of every sender. If you send about 5,000 or more emails a day to Gmail, marketing emails also need a one-click unsubscribe built into the email headers (ask your email tool whether it adds this for you), and unsubscribes must be processed within 2 days.
7. Send newsletters from a subdomain such as news.yourdomain.com, grow volume slowly, and never buy lists.
8. A person, not an AI agent, makes every DNS change. Consent and anti-spam law are not checked here.

Before you start

  • You need the words of the emails (subject lines, sequence, calls to action). This method covers whether email arrives, not what it says. Write an email nurture sequence
  • You want to check whether you are allowed to email people at all (consent, opt-in rules, CAN-SPAM, UK PECR, EU ePrivacy). That is a legal question this method does not answer. Check data-privacy compliance gaps
  • You plan to email people who never signed up (bought, scraped or appended lists). No DNS record fixes that: Google says not to send to them, and M3AAWG treats cold email as abusive.
  • You cannot receive email at all (missing or wrong MX records, mailbox setup). That is mailbox setup, not sending reputation.

Full method

How to use this method

  1. 1. Copy or downloadCopy the prompt below, or download the method file and attach it to your chat.
  2. 2. Paste into any AIChatGPT, Claude, Gemini or Copilot. Replace each [PLACEHOLDER] with your details, or let the AI ask you.
  3. 3. Add your materialAttach your own files, data or brief so the AI works from your facts, not guesses.

The prompt

Paste it as-is. The AI will ask for anything still in [BRACKETS].

Download .md
# Stop your emails landing in spam (SPF, DKIM, DMARC)
Method by Deepstory Research: deepstoryresearch.com/method/email-deliverability-setup

You are an email deliverability specialist who sets up domain authentication and sending practices for small businesses and startups. Follow the method below step by step. Goal: A deliverability plan for a domain that sends marketing and transactional email: which mailbox-provider rules apply (Gmail, Yahoo, Outlook.com), the exact SPF, DKIM and DMARC records to publish with an SPF lookup count, a DMARC rollout from none to quarantine to reject, one-click unsubscribe, a warm-up schedule, monitoring in Postmaster Tools, and what to do when mail lands in spam.

## My inputs
- Sending domain: [SENDING_DOMAIN]  (example: example.in)
- Tools that send email: [TOOLS_THAT_SEND_EMAIL]  (example: Google Workspace, HubSpot newsletters, app emails via SendGrid)
- Emails per day: [EMAILS_PER_DAY]  (example: one 20,000 newsletter a week, about 55% to Gmail)
- Current DNS records: [CURRENT_DNS_RECORDS]  (example: paste the SPF, DKIM and DMARC TXT records)
- Problem you see: [PROBLEM_YOU_SEE]  (example: newsletters land in Gmail spam)

Before you start: if any input above is still in [BRACKETS] or missing, ask me for it in one short numbered list and wait. Do not guess. If I attached files (data, briefs, Deepstory evidence files), treat them as the primary source.

## Steps to follow
1. List every service that sends email using the domain, and read the Return-Path, DKIM d= and From domains from one message of each.
2. Count messages per day to Gmail, Microsoft and Yahoo and decide which sender rules apply.
3. Choose the sending domains: root domain for staff mail, a subdomain for marketing.
4. Write one SPF record per domain and count its DNS lookups (10 at most).
5. Turn on DKIM with the user’s own domain in every service, 2048-bit where possible.
6. Publish DMARC at p=none with aggregate reports, then plan the move to quarantine and reject.
7. Check one-click unsubscribe headers on marketing mail and the unsubscribe process.
8. Set up Google Postmaster Tools and Yahoo Sender Hub, plan the warm-up and list cleaning, and set the review dates.

## Deliver the output in this structure
- Which sender rules apply and why
- Sender inventory with alignment
- DNS records to add or change, with the SPF lookup count
- DMARC rollout plan
- Unsubscribe check
- Warm-up and list-cleaning plan
- Monitoring and spam-folder fixes

## Quality rules
- Never publish a second SPF record; merge into the one record.
- Never recommend p=reject before DMARC reports show every legitimate sender passing.
- Every DNS change is written out for a person to add; nothing is changed by the AI.
- Say that anti-spam and consent law (for example CAN-SPAM, UK PECR, EU ePrivacy) is not checked here.
- Label every number as FACT (with source and date), ESTIMATE (show the calculation) or ASSUMPTION.
- If a Deepstory evidence file is attached, use its figures as the reference values and cite its source references; tag claims that are not from it as [UNVERIFIED].
- End with: the 3 assumptions that matter most, what would change the answer, and the single best next step.

What you need to give the AI

InputPlaceholderExample
Sending domain[SENDING_DOMAIN]example.in
Tools that send email[TOOLS_THAT_SEND_EMAIL]Google Workspace, HubSpot newsletters, app emails via SendGrid
Emails per day[EMAILS_PER_DAY]one 20,000 newsletter a week, about 55% to Gmail
Current DNS records[CURRENT_DNS_RECORDS]paste the SPF, DKIM and DMARC TXT records
Problem you see[PROBLEM_YOU_SEE]newsletters land in Gmail spam

Use it when

  • Your newsletters or receipts land in spam, or bounce with an error code.
  • You are adding a newsletter tool, CRM or app that sends email from your domain.
  • You send (or will send) about 5,000 or more emails a day to Gmail, Yahoo or Outlook.com addresses.
  • You are moving marketing email to a new domain or subdomain.

Best with: Any chat AI; paste your current DNS TXT records and the full headers ("Show original" in Gmail) of one email from each tool that sends for you.

What you get

  • Which sender rules apply and why
  • Sender inventory with alignment
  • DNS records to add or change, with the SPF lookup count
  • DMARC rollout plan
  • Unsubscribe check
  • Warm-up and list-cleaning plan
  • Monitoring and spam-folder fixes

The AI will ask you first

Answer these before the AI advises you; each one changes what it recommends.

  1. 1

    Which tools send email using your domain (your mailbox, newsletter or CRM tool, website forms, shop, app, helpdesk)?

    Each one needs SPF or DKIM with your domain, and any tool you forget will fail DMARC once you tighten the policy. For each tool, say whether its mail is marketing or subscribed content (newsletters, product news) or transactional (receipts, password resets): if you send about 5,000 or more a day to Gmail, it requires one-click unsubscribe for marketing and subscribed messages, and Google says not to mix promotions into transactional messages.

    If unknown: Ask the user to open one email from each tool in Gmail, click More, then "Show original", and paste the Authentication-Results lines. Until then, assume there is at least one forgotten sender and keep DMARC at p=none.

  2. 2

    On your busiest day, about how many emails go to Gmail addresses, to Outlook.com, Hotmail or Live addresses, and to Yahoo or AOL addresses?

    About 5,000 or more a day to personal Gmail accounts makes you a bulk sender permanently; 5,000 or more to Microsoft consumer addresses brings in Outlook.com’s rules.

    If unknown: Export the list and count addresses ending @gmail.com or @googlemail.com, @outlook.com, @hotmail.*, @live.* or @msn.com, and @yahoo.*, @ymail.com, @rocketmail.com or @aol.com; company addresses on Google Workspace or Microsoft 365 do not count. Until then, if the whole list is over 5,000 and goes out in one day, treat the user as a bulk sender at every provider, but do not compute spam-report numbers from the whole list.

  3. 3

    Who can change your DNS (your domain settings), and where is it managed (registrar, Cloudflare, web host)?

    Every fix is a DNS record. Cloudflare proxying can break the DKIM CNAME records some tools use.

    If unknown: Write each record out in full for the person who manages DNS; the AI does not change DNS.

  4. 4

    What are your current SPF, DKIM and DMARC records? Paste them, or the headers of one email.

    The most common faults are two SPF records, too many SPF lookups, and a tool that signs DKIM with its own domain instead of yours.

    If unknown: Look them up with a public checker (for example Google Admin Toolbox Check MX) and show the user what was found. If you cannot run DNS lookups, count each top-level term, mark every include "nested includes not verified", and ask the user to run a public SPF checker or `dig +short TXT <include>` for each include before relying on the total.

  5. 5

    Does your main domain host your website and staff mailboxes, and do you already send marketing from a subdomain (for example news.yourdomain.com)?

    Some tools (HubSpot, for example) cannot authenticate a domain that hosts your website, and a separate marketing subdomain keeps its reputation and DMARC policy apart from staff mail.

    If unknown: Rule of thumb: if the marketing tool already signs DKIM with your domain or a subdomain of it, and you have sent from this From address for months, keep it while you diagnose a problem: a new From domain needs weeks of warm-up and hides the cause. Plan a marketing subdomain for a new tool, a new programme, or when the tool cannot authenticate the main domain (for example HubSpot when the domain hosts your website). Keep staff mail on the main domain.

  6. 6

    Do you run your own mail server, or does every email go through a provider (Google, Microsoft, HubSpot, Mailchimp, SendGrid)?

    Reverse DNS (PTR) and TLS on your own servers are your job; with a provider they are the provider’s.

    If unknown: Assume providers only, and ask the user to confirm with whoever runs the website or app.

  7. 7

    Where did your list come from, when did you last remove bounces and people who never open, and do new subscribers confirm their address?

    High complaint and bounce rates undo good authentication.

    If unknown: Rule of thumb: treat addresses not emailed for a year or more, or (if you send weekly or more) addresses with no click or purchase in the last 6 months, and any bought or imported list, as high risk and leave them out of the first sends.

  8. 8

    What exactly is going wrong? Copy any bounce message or error code.

    Gmail and Outlook.com codes name the failed requirement (for example 5.7.26 unauthenticated or DMARC failure, 4.7.27 SPF, 4.7.30 DKIM, 4.7.32 alignment, 550 5.7.515 at Outlook.com).

    If unknown: Ask for one full bounce message; without it, start with the header check in step 1.

The procedure, step by step

Read this before you run it, so you can check the AI actually followed the method.

  1. 1

    Inventory every sender

    For each tool, open one received message in Gmail (More > Show original) and note the From domain, the Return-Path domain, the DKIM d= domain and the SPF, DKIM and DMARC results.

  2. 2

    Count volume per provider

    Messages on the busiest day to personal Gmail, to Outlook.com, Hotmail, Live and MSN, and to Yahoo and AOL. Decide which rule sets apply.

  3. 3

    Choose sending domains

    Main domain for staff mail; a subdomain for marketing (and optionally another for app or transactional mail). Check the marketing tool can authenticate it.

  4. 4

    Fix SPF

    One record per sending domain or subdomain; include only senders whose Return-Path is this domain; count lookups including nested includes; stay at 10 or fewer.

  5. 5

    Turn on DKIM everywhere

    Google Workspace (2048-bit, selector google), the marketing tool (for example HubSpot’s two DKIM CNAMEs, plus the MX, SPF and DMARC records it generates), the app or transactional provider. Confirm d= is your domain in a fresh message.

  6. 6

    Publish DMARC at p=none

    At least 48 hours after SPF and DKIM work. rua goes to a dedicated mailbox or report service. Set a review date one week out (Google’s minimum) and plan a month for any domain heading to reject.

  7. 7

    Check unsubscribe

    Marketing mail has List-Unsubscribe (HTTPS) and List-Unsubscribe-Post headers covered by DKIM, a visible body link, and removal within 48 hours (Gmail) and 2 days (Yahoo). Test that the unsubscribe URL is not blocked by a firewall or CDN.

  8. 8

    Set up monitoring

    Google Postmaster Tools (verify domain and subdomains), Yahoo Sender Hub with the Complaint Feedback Loop, Microsoft SNDS if you run your own sending IPs, and the sending tool’s bounce and complaint reports.

  9. 9

    Warm up and clean the list

    Start with the most engaged subscribers, grow without sudden doubling over about 6 weeks, and remove repeat bounces and long-term non-openers.

  10. 10

    Tighten DMARC

    When reports show every legitimate sender passing: main domain to quarantine; marketing subdomain to quarantine, then reject after an equal period. Re-check after any new tool is added.

Platform settings(show)
PlatformPathSettingNote
Google Workspace Admin consoleMenu > Apps > Google Workspace > Gmail > Authenticate emailGenerate New Record (2048-bit, prefix selector google), add the TXT record at your DNS host, then Start authenticationSuper administrator sign-in; needs the Gmail Settings administrator privilege. Wait 24 to 72 hours after turning on Gmail before generating; status should change to "Authenticating email with DKIM".
GmailOpen the message > More (next to Reply) > Show originalAuthentication-Results: check DKIM=pass (and SPF and DMARC)Send the test to another Gmail or Google Workspace user; you cannot verify DKIM by sending to yourself.
DNS host (domain provider)DNS TXT records > Host @ (or the subdomain)One SPF record, for example v=spf1 include:_spf.google.com ~allEach domain and subdomain that sends needs its own SPF record; can take up to 48 hours to work.
DNS host (domain provider)DNS TXT records > Host _dmarcv=DMARC1; p=none; rua=mailto:dmarc-reports@yourdomainv and p must come first. Add only after SPF and DKIM have worked for 48 hours.
Google Postmaster ToolsManage Domains > Add new domain; then More options > Verify domainAdd the DKIM d= domain or SPF Return-Path domain; copy the TXT or CNAME verification record to DNS, then click VerifyVerification is usually instant but can take up to 10 minutes. Add subdomains separately for their own dashboards.
Google Postmaster ToolsTop menus: Domain > Dashboard type > Time rangeCompliance status (with Deliverability analysis), Spam rate, Feedback loop, Authentication, Encryption, Delivery errors (7 to 120 days)Compliance status shows the primary domain only and can take up to 7 days to update after a fix. Google is retiring the Domain and IP Reputation dashboards in Postmaster Tools v2, so do not plan around a reputation label.
HubSpotSettings > Content > Domains & URLs > Email Sending tab > Connect sending domainAdd the generated MX, DKIM (two CNAME), SPF and DMARC records; check the domain shows "Authenticated"Needs Domain settings permission. The sending domain must not host your website; merge HubSpot’s SPF include into your existing SPF record.
Yahoo Sender HubManage Services > Complaint Feedback LoopEnrol each verified DKIM (d=) domainYour email provider may already enrol you; check with them.

Worked example

The figures below are illustrative. Use your own numbers and facts.

Scenario: The figures below are an illustration; use only the user’s own numbers. A Bangalore B2B SaaS startup (CRM software for small businesses) sends a 20,000-contact newsletter once a week from HubSpot with From hello@example.in. Staff use Google Workspace on example.in, which also hosts the website. The app’s password-reset emails go through SendGrid. Newsletters are landing in Gmail spam.

Input: HubSpot export: 11,000 contacts end in @gmail.com, 1,400 in Outlook.com, Hotmail or Live, 600 in Yahoo. DNS for example.in has two TXT records starting v=spf1: (1) v=spf1 a mx include:_spf.google.com include:servers.mcsv.net include:sendgrid.net include:zoho.in ~all (Mailchimp and a Zoho trial are no longer used); (2) v=spf1 include:123456.spf03.hubspotemail.net -all, added by the agency for HubSpot. No DMARC record. HubSpot shows the domain as not authenticated.

Output excerpt: Rules that apply: 11,000 messages to personal Gmail in one day is over about 5,000, so example.in is a Gmail bulk sender, permanently. Microsoft: 1,400 is under 5,000, so Outlook.com’s high-volume rule does not apply yet, but the same setup meets it. Spam rate: if all 11,000 reach the inbox, 0.10% is 11 spam reports per send and 0.30% is 33; 11 or more reports on the day of a send already misses Google’s target of below 0.10% (Gmail counts only DKIM-signed mail that reached the inbox, so treat this as an approximation). SPF: two v=spf1 records give permerror, so SPF fails for all mail from example.in today. Lookups in record (1), counted in DNS on 4 November 2026 (providers change these): a 1, mx 1, _spf.google.com 1, servers.mcsv.net 1, sendgrid.net 1 plus its nested ab.sendgrid.net 1, zoho.in 1 plus spf.zoho.in 1, total 8. Fix: (a) Move newsletters to news.example.in, because HubSpot cannot authenticate a domain that hosts the website. In HubSpot, Connect sending domain hello@news.example.in and add the MX record, the two DKIM CNAMEs, the SPF record and the DMARC record HubSpot generates (1 SPF lookup for HubSpot’s include). (b) Replace both root records with one: v=spf1 include:_spf.google.com include:sendgrid.net ~all = 1 + 2 = 3 lookups (Mailchimp and Zoho removed as unused; a and mx removed as Google’s record covers Workspace, once the web host confirms the website’s contact forms do not send mail from the web server; check the Return-Path of a SendGrid message: if it is SendGrid’s domain or your own em subdomain, remove include:sendgrid.net from the root). (c) Turn on 2048-bit DKIM in Google Workspace and SendGrid domain authentication. (d) 48 hours later publish _dmarc.example.in: v=DMARC1; p=none; rua=mailto:dmarc-reports@example.in. (e) Check a test newsletter’s headers for List-Unsubscribe (HTTPS) and List-Unsubscribe-Post. Warm-up for news.example.in, most engaged first, no step doubling: 2,000, 3,500, 6,000, 10,000, 15,000, 20,000 over six weekly sends (growth 1.75x, 1.71x, 1.67x, 1.5x, 1.33x). Tell subscribers about the new From address first. Until their week comes, pause the contacts not yet moved: HubSpot mail from example.in is not aligned with it, so it fails Gmail’s bulk rules, and the warm-up rule says not to keep mailing them from it for weeks. No one should get two copies. DMARC: root to quarantine after reports are clean (staff use mailing lists, so not reject); news.example.in gets its own _dmarc record and moves none, quarantine, reject a month apart. A person adds every DNS record.

Check your result

Before you trust the AI's output, confirm each of these.

  • ☐ Every sending tool is listed with its From, Return-Path and DKIM d= domains.
  • ☐ Each sending domain has exactly one SPF record with 10 or fewer DNS lookups, counted including nested includes.
  • ☐ A fresh message from each tool shows SPF pass, DKIM pass and DMARC pass with an aligned domain.
  • ☐ DMARC starts at p=none with a working rua address, and the plan does not move to reject before reports are clean.
  • ☐ Marketing messages carry List-Unsubscribe (HTTPS) and List-Unsubscribe-Post headers covered by DKIM.
  • ☐ Spam-rate targets are written as a number of reports per day of sending, computed on personal-Gmail recipients only (below 0.10%, never 0.30%).
  • ☐ Every DNS and account change is listed for a named person to make; nothing was changed by the AI.
  • ☐ The plan says consent and anti-spam law were not checked.
What you'll get(show template)
## Email deliverability plan: [DOMAIN], [DATE]

### 1. Rules that apply
| Provider | Messages on busiest day | Threshold | Applies? |
|---|---|---|---|
| Gmail (personal) | [n] | about 5,000 a day, permanent once reached | [Bulk sender / All-sender rules] |
| Outlook.com, Hotmail, Live, MSN | [n] | 5,000 or more, same From domain | [Yes / Not yet] |
| Yahoo, AOL | [n] | no fixed number | [Treat as bulk if large] |
| Other providers in your list (for example GMX, WEB.DE and T-Online in Germany, Orange in France) | [n] | not covered by this method | Check each provider’s own sender rules |

### 2. Senders
| Tool | From address | Return-Path domain | DKIM d= | SPF aligned? | DKIM aligned? | Fix |
|---|---|---|---|---|---|---|

### 3. DNS records to add or change (a person makes these)
| Host | Type | Current value | New value | Why |
|---|---|---|---|---|

SPF lookup count for [domain]: [term = n, ...] Total [n] of 10 (checked [date] by [DNS lookup / user’s checker / not verified]).

### 4. DMARC rollout
| Domain | Policy now | Next policy | Review date | Move on only if |
|---|---|---|---|---|

### 5. Unsubscribe check
List-Unsubscribe (HTTPS): [yes/no] | List-Unsubscribe-Post: [yes/no] | Covered by DKIM: [yes/no] | Body link: [yes/no] | Removal time: [hours]

### 6. Warm-up and list cleaning
| Send | Date | Recipients | Segment | Growth vs last |
|---|---|---|---|---|
Removed before sending: [bounces, inactive since, unconfirmed]

### 7. Monitoring
Postmaster Tools: [verified domains] | Yahoo Sender Hub and CFL: [status] | Spam rate target (Gmail personal accounts only, per day of sending): [G] messages to @gmail.com or @googlemail.com reach the inbox, so below 0.10% = fewer than [G ÷ 1,000] reports and never 0.30% = [3 × G ÷ 1,000]. If G is not known, write "not computed: count Gmail addresses in the export". Never use the whole list: that raises the allowed number of reports.

### 8. If mail lands in spam
[Header results, dashboard findings, recent changes, blocklist check, then content]

### Not checked
Marketing-consent and anti-spam law (for example CAN-SPAM, UK PECR, EU ePrivacy) is not checked here. Rules of mailbox providers other than Gmail, Yahoo and Microsoft (Outlook.com) are not checked.

Quality checks built into the prompt

  • • Never publish a second SPF record; merge into the one record.
  • • Never recommend p=reject before DMARC reports show every legitimate sender passing.
  • • Every DNS change is written out for a person to add; nothing is changed by the AI.
  • • Say that anti-spam and consent law (for example CAN-SPAM, UK PECR, EU ePrivacy) is not checked here.
  • • Every number labelled FACT, ESTIMATE or ASSUMPTION; claims outside attached evidence tagged [UNVERIFIED].

Decision rules

If/then rules the AI should apply. Heuristic rules are a practitioner's rule of thumb, not a platform rule.

IfThenType
You are deciding whether Gmail’s bulk-sender rules apply.A bulk sender sends close to 5,000 messages or more to personal Gmail accounts (ending @gmail.com or @googlemail.com) within 24 hours. Mail from all subdomains counts towards the same primary domain, so 2,500 from example.com plus 2,500 from news.example.com is 5,000. Once reached, bulk status is permanent. The rules apply to mail sent from Google Workspace too, but not to mail sent to Google Workspace accounts.Threshold: about 5,000 a day to personal GmailPlatform rule
You send any email to personal Gmail accounts.Since 1 February 2024 every sender must: set up SPF or DKIM; have valid forward and reverse DNS (PTR) for sending domains or IPs; send over TLS; keep the spam rate in Postmaster Tools below 0.3%; format messages to RFC 5322; and not put a @gmail.com address in the From header (Google says Gmail will use a DMARC quarantine enforcement policy, so this can hurt delivery).Platform rule
You are a Gmail bulk sender.You also need SPF and DKIM both; a DMARC record (p=none is enough); the From domain aligned with the SPF or the DKIM domain; and one-click unsubscribe plus a clearly visible unsubscribe link in the body of marketing and subscribed messages, with unsubscribes honoured within 48 hours. As of November 2026, Google says that since November 2025 it is "ramping up its enforcement": non-compliant messages get temporary and permanent rejections. Google’s FAQ table: failed alignment, SPF and DKIM, forward and reverse DNS, TLS or RFC 5322 format bring temporary or permanent failure codes or spam foldering; a missing DMARC record, missing one-click unsubscribe, unsubscribes not honoured within 48 hours or a spam rate above 0.3% make you ineligible for delivery help (mitigation).Platform rule
You are checking spam complaints at Gmail.Keep the user-reported spam rate in Postmaster Tools below 0.10% and never let it reach 0.30%. It is calculated daily, on mail delivered to the inbox. Above 0.3% you are ineligible for mitigation until it stays below 0.3% for 7 days in a row. A very low or zero rate after a high one can mean Gmail is already sending your mail to spam, so read it with the Compliance status dashboard (including Deliverability analysis) and the Delivery errors dashboard. Gmail and Yahoo require every sender, not only bulk senders, to stay below 0.3%.Threshold: below 0.10%, never 0.30% or higherPlatform rule
You send to Yahoo or AOL addresses.Yahoo gives no volume number for bulk senders. All senders need SPF or DKIM, a spam rate below 0.3%, valid forward and reverse DNS and RFC 5321 and 5322 compliance. Bulk senders also need SPF and DKIM, DMARC of at least p=none that passes (relaxed alignment is fine; a rua tag is strongly recommended), a working List-Unsubscribe header (the RFC 8058 POST method is highly recommended, mailto is acceptable), a visible unsubscribe link, and unsubscribes honoured within 2 days. Yahoo’s spam rate counts only mail delivered to the inbox. Enrol your DKIM domain in Yahoo’s Complaint Feedback Loop.Threshold: spam rate below 0.3%Platform rule
You send 5,000 or more messages to Microsoft consumer addresses (Outlook.com, Hotmail, Live, MSN) with the same From domain.Both SPF and DKIM must pass, and a DMARC record of at least p=none must pass with SPF or DKIM aligned to the From domain (Microsoft prefers both). Microsoft’s support page says non-compliant messages are rejected with "550 5.7.515 Access denied, sending domain [domain] does not meet the required authentication level". Recipients adding you to Safe Senders does not bypass this.Threshold: 5,000 or more to Microsoft consumer servicesPlatform rule
You are writing or editing an SPF record.Publish exactly one record starting v=spf1 per domain or subdomain: two records give "permerror" and SPF fails for every message. Each subdomain that sends needs its own record. The include, a, mx, ptr and exists mechanisms and the redirect modifier each cost one DNS lookup, includes inside includes count too, and more than 10 gives "permerror"; an include that points at a name with no SPF record gives "permerror" by itself, and a, mx or exists terms whose names return nothing count against a separate limit of 2 "void lookups", so remove dead includes and terms. ip4, ip6 and all cost nothing. The ptr mechanism should not be published. Google’s setup pages say to list every server that sends for you; SPF is checked against the Return-Path domain, so add a sender’s include only to the SPF record of the domain that sender uses as its Return-Path (envelope MAIL FROM), which you can read in "Show original". Many tools bounce to their own domain or to a subdomain you set up for them: Amazon SES, for example, uses a subdomain of amazonses.com unless you set a custom MAIL FROM subdomain, which then gets its own record (v=spf1 include:amazonses.com ~all) and an MX record. Their include at your main domain costs a lookup and authorises nothing. Remove senders you no longer use.Threshold: 1 record; 10 DNS lookups at mostProfessional guidance
A tool’s SPF passes but DMARC still fails.DMARC checks SPF against the Return-Path (envelope, MAIL FROM) domain, not the From address you see. If the tool bounces mail to its own domain, its SPF cannot align with yours; get alignment from DKIM signed with your domain (d=yourdomain), or set the tool’s custom return-path to your domain. Relaxed alignment (the default) accepts any subdomain of the same organisational domain. Google recommends aligning both SPF and DKIM and says this will likely become a requirement.Professional guidance
You are turning on DKIM.Gmail needs a DKIM key of at least 1024 bits and recommends 2048; the DKIM standard says signers MUST use at least 1024 bits and SHOULD use 2048. In Google Workspace, generate the key in the Admin console (Authenticate email), add the TXT record at selector._domainkey (default selector google; choose another if google is already used), then click Start authentication. Wait 24 to 72 hours after turning on Gmail before generating a key, and allow up to 48 hours for DKIM to start working. Each sending tool uses its own selector, so several DKIM keys can live side by side.Platform rule
You are adding DMARC for the first time.Have SPF and DKIM authenticating for at least 48 hours first. Publish a TXT record at _dmarc.yourdomain starting v=DMARC1; p=none with rua=mailto: a mailbox or group that only collects reports (they can be many a day). Google recommends reading the reports daily for at least a week, then moving to quarantine, then (where suitable) reject. A report address on a different domain needs an extra DNS record at that domain.Threshold: p=none for at least 1 weekPlatform rule
Your DMARC record or a guide uses the pct tag (for example pct=5).DMARC was republished as a Standards Track RFC in May 2026 (RFC 9989, with RFC 9990 for aggregate reports), replacing RFC 7489. The record still starts v=DMARC1. The pct tag is now "historic": the standard says values other than 0 and 100 were usually not applied accurately. Its replacement is t=y (test mode), which asks receivers to apply one level below your policy (reject is treated as quarantine, quarantine as none). New tag np sets a policy for subdomains that do not exist. Google’s rollout page still shows pct examples, so do not rely on a percentage to limit risk; step the policy itself.Professional guidance
You want p=reject on a domain where staff send everyday email.RFC 9989 says domains whose users might post to mailing lists SHOULD NOT publish p=reject, because forwarding and lists break SPF and can get recipients unsubscribed from lists. If you still want reject: publish p=none for at least a month, then p=quarantine for as long again, compare results, and sign everything with DKIM (a p=reject domain MUST NOT rely on SPF alone).Threshold: none 1 month, then quarantine 1 monthProfessional guidance
You are setting up one-click unsubscribe for marketing email.Add two headers: List-Unsubscribe with one HTTPS URL (a mailto may be added too) and List-Unsubscribe-Post: List-Unsubscribe=One-Click. A DKIM signature must cover both headers. The URL must identify the recipient and the list (and should include a hard-to-guess token), must work without cookies or a login, and must not answer with a redirect. A link in the body or a mailto alone does not meet Gmail’s requirement, though an extra body link may go to a preferences page. Transactional mail (password resets, order confirmations) does not need it. If a firewall or CDN (for example Cloudflare) blocks the POST, Google says you are still responsible for the unprocessed unsubscribe.Professional guidance
You run your own sending server or a dedicated IP.The sending IP must have a PTR (reverse DNS) record that resolves to a hostname, and that hostname’s A or AAAA record must resolve back to the same IP (forward-confirmed). Missing this gives Gmail 4.7.23 (rate limited) or 5.7.25 (blocked). Use one PTR name per IP that clearly names the responsible domain, not a generic pool name, and send over TLS (Gmail 4.7.29 or 5.7.29 means no TLS). If an email provider sends for you on shared IPs, this is the provider’s job.Platform rule
Gmail returns a 4.7.x or 5.7.x code.Read the code: 4.7.26 or 5.7.26 the message is unauthenticated (neither SPF nor DKIM passed), failed your own DMARC policy, or failed an SPF record that ends -all; 4.7.27 or 5.7.27 SPF did not pass; 4.7.30 or 5.7.30 DKIM did not pass; 4.7.31, 4.7.40 or 5.7.40 no DMARC record or no policy (Google’s two pages use different numbers); 4.7.32 or 5.7.32 From not aligned with SPF or DKIM; 4.7.28 or 5.7.28 unusual rate of email or quota exceeded. For 4.7.28 stop sending for at least 10 minutes, then resume on a single connection and add connections one at a time. 4xx codes are temporary (retry later), 5xx are permanent (do not retry that message).Platform rule
You start sending from a new domain, subdomain or IP, or volume will jump.Start with a low volume to your most engaged subscribers and increase slowly at a steady rate; avoid sudden spikes such as doubling previous volume, which can bring rate limiting or reputation drops. Sending daily lets you grow faster than weekly. M3AAWG suggests about 6 weeks of warm-up on average, warming a new subdomain even if the main domain has a good reputation, and spreading each day’s mail out rather than sending it all at once. If bounces or deferrals rise, cut volume and climb again slowly. If the current setup fails Gmail’s bulk rules (the From domain is aligned with neither SPF nor DKIM, or there is no DMARC record), do not keep mailing the not-yet-moved contacts from it for weeks: fix authentication on the current domain first (DKIM signed with your domain, DMARC p=none), or pause those contacts until their turn on the new subdomain.Threshold: about 6 weeks (M3AAWG average); no sudden doublingProfessional guidance
You already send from this domain and spam placement, complaints or deferrals are rising.Rule of thumb: for the next 2 to 4 sends, mail only recipients who clicked, bought or signed up recently (choose a window that fits how often you send, for example 90 days if you send weekly or more). Then add back older segments, most recent first, never doubling the previous send. Run one re-engagement email for long-inactive contacts and stop mailing those who do not respond. Review how often you send.Heuristic
You are choosing which domain each kind of email comes from.Use subdomains of your own domain for different purposes (for example news. for marketing, a different one for transactional mail), not lookalike domains such as example-promo.com; moving to a lookalike domain to escape a bad reputation is what spammers do. Keep the same From address for each category, do not mix promotions into receipts, and if you use several IPs give each message type its own. HubSpot can only authenticate a sending domain that does not host your website, and the subdomain should match the From address. This is for setting up: do not move an existing sender to a subdomain to fix a spam problem until the cause is found.Professional guidance
You are deciding who stays on the list.Send only to people who opted in; confirmed (double) opt-in is the best practice. Never buy lists or pre-tick sign-up boxes. Remove addresses that bounce on at least two consecutive sends over two weeks or more, remove "user unknown" addresses at once, and consider removing people who never open or read.Threshold: 2 consecutive bounces over 2+ weeksProfessional guidance
You are setting up Google Postmaster Tools.Add the domain you authenticate with (the DKIM d= domain or the SPF Return-Path domain) and verify it with the TXT or CNAME record shown. Add subdomains separately to see their own dashboards. Data covers only personal Gmail accounts, is in UTC, updates within about 24 hours and can be missing on low-volume days. The Compliance status dashboard reports the primary domain only and can take up to 7 days to reflect a fix.Platform rule
HubSpot asks you to add an SPF value and you already have an SPF record.Add HubSpot’s include: value to the existing record rather than creating a second one, and keep v=spf1 and the all term only once. HubSpot generates an MX record, two DKIM CNAME records, an SPF TXT record and a DMARC TXT record; add all of them. With Cloudflare, turn off proxying and domain-wide CNAME flattening. DNS changes usually show in 10 to 70 minutes but can take 48 hours.Platform rule
Mail lands in spam and you do not know why.Rule of thumb: the Gmail Promotions tab is not spam. With the Default inbox, Gmail sorts deals, offers and other promotional email there, and no DNS record changes that. Open rates are a weak signal (some mail apps open messages automatically, and Promotions placement lowers opens), so judge by the Postmaster Tools spam rate and delivery errors, clicks and orders. Then check in this order: (1) one received message’s headers ("Show original") for SPF, DKIM and DMARC pass and which domains they used; (2) Postmaster Tools Compliance status (including Deliverability analysis), spam rate and delivery errors, and Yahoo Sender Hub; (3) what changed recently (new list import, volume jump, new tool, new From address, new link domain); (4) whether your domain or link domains are flagged in Google Safe Browsing or a public blocklist; only then (5) the content. Do not move to a new domain to escape the problem.Heuristic
The SPF record would need more than 10 lookups.Rule of thumb: first remove services you no longer use and redundant a and mx terms (Google’s own Workspace record is only include:_spf.google.com); then move a heavy sender to its own subdomain with its own SPF record. Treat "flattening" (copying a provider’s IP addresses into your record) as a last resort, because providers change their IPs without telling you. Recount after every change.Heuristic
You are choosing how the SPF record ends.Rule of thumb: end with ~all, which Google recommends. Use -all (HubSpot’s example) only when you are sure every server that sends for that domain is listed, because Gmail can block a message outright (550 5.7.26) when it fails an SPF record that ends -all. Never use +all, which authorises every server on the internet, and never put the all term twice.Heuristic
The main domain must stay at quarantine because of staff mail, but you want marketing protected more strongly.Rule of thumb: publish a separate _dmarc record on the marketing subdomain (for example _dmarc.news.example.com). A record on the subdomain overrides the one inherited from the main domain, so the subdomain can move to reject once its own reports are clean, while staff mail stays at quarantine.Heuristic
The user asks whether they are legally allowed to send the emails.Rule of thumb (scope): this method does not check marketing-consent or anti-spam law (for example the US CAN-SPAM Act, UK PECR, the EU ePrivacy rules, Canada’s CASL or India’s rules). Say so plainly and point to a consent and privacy check; mailbox-provider rules are not a legal test.Heuristic

Failure modes

Common mistakes with this method, how to spot them, and how to fix them.

MistakeSymptomFix
Adding a second SPF record for a new tool.SPF permerror: every message from the domain fails SPF, and DMARC fails wherever DKIM is not aligned.Merge all senders into one v=spf1 record per domain, or give the new tool its own subdomain.
Too many includes in SPF.More than 10 DNS lookups gives permerror at some receivers; failures look random.Count lookups including nested includes, remove unused senders, move a sender to a subdomain.
A tool signs DKIM with its own domain and bounces to its own domain.SPF and DKIM both "pass" but DMARC fails (Gmail 4.7.32, Outlook.com 550 5.7.515).Set up the tool’s domain authentication so DKIM d= (and ideally the Return-Path) uses your domain.
Jumping straight to p=reject.A forgotten sender (invoices, helpdesk, forms) or forwarded mail is rejected without anyone noticing.Start at p=none, read reports, move through quarantine; keep staff domains off reject.
Relying on a body link or mailto for unsubscribe.Postmaster Tools shows One-click unsubscribe as "Needs work"; spam reports rise.Add List-Unsubscribe (HTTPS) and List-Unsubscribe-Post headers covered by DKIM, and process removals within 48 hours.
Sending the whole list from a new subdomain or IP on day one.Gmail 4.7.28 rate limiting, deferrals, mail in spam.Warm up over weeks, engaged contacts first, without sudden doubling.
Mailing an old, imported or bought list.Hard bounces and a spam rate above 0.3%; Gmail stops offering mitigation.Remove unconfirmed and long-inactive addresses, use confirmed opt-in, and remove repeat bounces.
Leaving DKIM CNAMEs proxied in Cloudflare.DKIM lookups fail and the tool shows the domain as not authenticated.Set those records to DNS only and turn off CNAME flattening, as HubSpot advises.
Sending bulk mail with a free mailbox From address (@gmail.com, @yahoo.com, @aol.com, @outlook.com) through a newsletter tool.Mail From @yahoo.com or @aol.com is rejected at major providers because Yahoo publishes a DMARC reject policy; Google warns that impersonating Gmail From headers can hurt delivery as Gmail moves to a DMARC quarantine policy.Send from your own authenticated domain or subdomain.
Adding Google Workspace’s DKIM TXT record but never clicking Start authentication.Your Workspace mail has no DKIM line for your domain in Show original; DMARC depends on SPF alone and fails when mail is forwarded.In the Admin console (Apps > Google Workspace > Gmail > Authenticate email), click Start authentication and confirm the status reads "Authenticating email with DKIM". Check with a message sent to someone else, not to yourself.
Switching to a lookalike domain to escape a bad reputation.The new domain quickly gets the same treatment, and it looks like spammer behaviour.Fix the cause (list, complaints, authentication) on the real domain and recover slowly.

Agent mode: if an AI agent runs this unattended

May do without asking

  • • Look up public DNS records (SPF, DKIM, DMARC, MX, PTR) with read-only lookup tools and count SPF lookups.
  • • Read message headers, bounce messages and DMARC reports the user pastes or shares.
  • • Read Google Postmaster Tools and Yahoo Sender Hub dashboards in a read-only session.
  • • Draft DNS records, a DMARC rollout calendar, a warm-up schedule and list-cleaning rules for the user to review.

Must stop before (the user decides and does these)

  • • Adding, editing or deleting any DNS record, including SPF, DKIM, DMARC and verification records.
  • • Using a sending tool’s "Sign in with [DNS provider]" button, which writes DNS records automatically.
  • • Clicking Generate New Record or Start authentication in the Google Workspace Admin console, or changing any account or domain setting in a sending tool.
  • • Sending, scheduling or publishing any email, campaign or test send to real contacts.
  • • Uploading, importing, exporting or deleting contact lists, or suppressing contacts.
  • • Accepting terms to join Postmaster Tools, Yahoo Sender Hub, a feedback loop or a new email service.
  • • Anything that would spend money or change billing, such as buying a dedicated IP, a warm-up service or a DMARC report tool.
  • • Contacting a mailbox provider’s support or submitting a mitigation request on the user’s behalf.

Evidence for this method

This method runs on your own material: your files, brief or code. If your task needs market or economic numbers, attach a sourced context file from the evidence library alongside the prompt.

Browse evidence

Run next

Methods that chain well after this one.

Deepstory methods are process guidance for AI tools. Review AI output before relying on it; legal, tax, medical and investment decisions need a qualified professional.