The Outbound Engine, Anonymized

Never send cold email from your main domain

Everything else in this system exists to serve one rule: cold email never touches the domain your business lives on. Not once, not "just this batch of 15". Cold email gets spam-flagged sometimes even when it's good and wanted. Those flags damage domain reputation, and domain reputation is shared by every email the domain sends: your newsletter, your receipts, your password resets. One bad cold campaign from yournewsletter.com can tank the deliverability of the exact product you're selling sponsorships for.

Architecture: five components, two human gates

Five components, none of them exotic; the value is in how they're wired. The design principle: everything cheap and reversible happens early (searching, judging, drafting), everything expensive and irreversible happens late (credit spends, live sends). Human judgment sits at exactly two gates: picking the contact, and approving the drafts. Everything else is automatable and should be.

The flow, end to end

1. 1

   Scout (free)

   For every target company, pull ALL plausible candidates via the contact database's free search: marketing, partnerships, demand gen, community. No credits spent. Dump everything to a candidates file.

2. 2

   Judge

   A human (or an LLM with a tight rubric, spot-checked) reads each company's candidate list and picks one primary and one alternate. This is where the buyer-vs-marketer call gets made.

3. 3

   Reveal (paid)

   Spend credits revealing verified emails for the chosen two per company only. Never reveal whole rosters. Verified-only by default; bounces are poison on young domains.

4. 4

   Draft

   The sequencer renders every email to a file. You read them. You will catch broken merge fields and stale fit lines every single batch; this gate has never once been useless for me.

5. 5

   Send

   Initial, then follow-up 1 at +3 days, follow-up 2 at +8 days. 10 to 15 cold sends per mailbox per day, hard cap in code.

6. 6

   Escalate

   No reply from the primary after 2 days: the alternate contact gets a fresh intro from a different mailbox. One escalation per company, ever.

7. 7

   Reply kills everything

   Any reply from either contact, including "no", immediately stops every scheduled message to that company. The conversation continues by hand from the mailbox that owns the thread.

8. 8

   CRM stage advances

   queued, contacted, replied, interested, negotiating, booked, live or passed. Update it on every touch, even if the CRM is a JSON file.

Domains and DNS: the records that make you deliverable

Buying the lookalikes

Buy 3 to 5 lookalikes of your real brand. Patterns that work: get<brand>.com, try<brand>.com, <brand>hq.com, <brand>-media.com, read<brand>.com. Stick to .com; bargain TLDs like .xyz carry spam baggage from every spammer who saved $9 before you.

• [ ]3 to 5 domains, 1 to 2 mailboxes each (more domains distributes risk, more mailboxes per domain concentrates it)
• [ ]Buy them the day you decide to do outbound: warmup takes 2+ weeks and older domains warm better
• [ ].com only, ~$10 each, this is not the line item to save on
• [ ]Mailbox names should look human: firstname@, not sales@ or hello@

SPF: one record, hard fail

Type:  TXT
Host:  @
Value: v=spf1 include:_spf.yoursendingplatform.com -all

Use -all (hard fail), not ~all. On a domain whose only job is outbound there is no legitimate "other" sender, so fail them hard; softfail invites spoofing on a domain you're trying to build reputation on. And exactly ONE SPF record per domain: two TXT records starting with v=spf1 is a permerror, which means SPF is effectively broken.

DKIM: your ESP gives you the record

Type:  CNAME
Host:  s1._domainkey
Value: s1.dkim.yoursendingplatform.com

# or, if your ESP hands you a raw key:
Type:  TXT
Host:  selector1._domainkey
Value: v=DKIM1; k=rsa; p=MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8A...KEY_HERE...IDAQAB

The selector name (s1, selector1) is dictated by the ESP, don't invent one. Take 2048-bit keys where offered, and verify green in the ESP dashboard after adding; propagation is usually minutes, occasionally hours.

DMARC: p=reject, and why that is correct here

Type:  TXT
Host:  _dmarc
Value: v=DMARC1; p=reject; rua=mailto:dmarc@yournewsletter-hq.com; adkim=s; aspf=s; pct=100

p=reject scares people, but on a single-purpose outbound domain it's the obvious call: every legitimate email from this domain is yours and authenticated, so anything failing alignment is spoofing. Reject it. Gmail and Yahoo require at least p=none from bulk senders (5,000+/day); you'll be far under that volume, but reject exceeds the bar and signals a serious sender. Strict alignment (adkim=s; aspf=s) is fine because the From domain and the SPF/DKIM domains are identical here.

MX and the redirect

Type:     MX
Host:     @
Priority: 10
Value:    mx.yoursendingplatform.com

# root redirect (registrar forwarding or any host):
yournewsletter-hq.com/*  ->  301  ->  https://yournewsletter.com

No MX means replies bounce and the entire system is pointless. And the redirect is not optional cosmetics: prospects check the domain before replying. A parked page or an error screams throwaway spam infrastructure; a 301 to a real site with a real audience page closes the credibility gap in one click.

The domain math

At 10 to 15 sends per mailbox per day, 4 domains x 2 mailboxes is roughly 96 emails/day of capacity. That's more pipeline than a solo operator can handle replies for anyway, no single domain carries filter-tripping volume, and none of them is your real one. The full copy-paste record set is in dns-records-example.md at the bottom of this page.

Warmup: the mandatory 2 weeks you cannot skip

New domains and new mailboxes have zero reputation. A zero-reputation sender that suddenly blasts 50 emails a day looks exactly like a spammer, because that is what spammers look like. Mail providers act accordingly. What a warmup network actually does: your sending platform enrolls your mailboxes in a pool of real inboxes that email each other automated but human-looking messages. Those messages get opened, replied to, rescued from spam folders, and marked important. To Gmail and Outlook, your mailbox looks like a normal human account building normal relationships. That's the whole trick.

1. 1

   Start warmup the day DNS verifies

   Every day of delay is a day added to your go-live date. I bought domains and started warmup on day one; my live-send date was gated 12 days out and enforced in code.

2. 2

   Wait a minimum of 2 weeks

   3 to 4 weeks is better. Make the date a literal interlock: my send script refuses to run with the live flag before the warmup-ready date. Willpower is not an interlock; an if-statement is.

3. 3

   Watch health scores per mailbox

   You're ready when: 2+ weeks elapsed, the platform's health score is sustained in its green zone (typically 90%+), and warmup mail is landing in inbox rather than spam across providers.

4. 4

   Ramp, don't jump

   Start at ~5 cold sends per mailbox per day, add ~2/day every few days up to your 10 to 15 cap. Full volume on day one of live sends undoes the warmup story you just spent 2 weeks telling.

5. 5

   Keep warmup running after go-live

   At reduced volume. Warmup traffic keeps the positive-signal ratio healthy alongside real cold sends, which get fewer opens and replies by nature.

The sequence: 3 touches, 1 escalation, reply kills all

Four mechanics carry this whole table, and none of them is optional. 1. A reply kills everything. Any reply, from either contact, including "not interested", stops every scheduled message to that company the moment it's detected. An automated follow-up landing after a human reply is how you get spam-flagged and embarrassed in the same afternoon. 2. Escalation is one-shot and from a different mailbox. If the primary is silent after 2 days, the alternate gets a fresh initial with a different fit angle. Not a forward, no "my colleague didn't answer" (that reads as surveillance). One escalation per company, ever. Two people ignoring you is an answer; take it.

3. Daily caps of 10 to 15 per mailbox. This is the load-bearing deliverability number, enforced in code, not vibes. Across 4 domains x 2 mailboxes you still get 80 to 120 sends/day, which will out-produce your ability to handle replies. 4. Follow-ups reply to your own thread. Same subject, In-Reply-To header set. Threading raises reply rates and lowers spam signals versus a fresh subject line every touch.

Targeting: scout free, judge hard, spend last

The expensive mistake in B2B prospecting is paying to reveal contact data for someone you later realize is the wrong buyer. The fix is to invert the order: on the major contact databases, searching is free and reveals cost credits, so search everyone and reveal almost no one.

1. 1

   Stage 1: scout everyone (free)

   Per target company, pull all plausible candidates via free search endpoints: every title in marketing, partnerships, demand generation, community. You're building a complete menu per company before spending a cent.

2. 2

   Stage 2: judge (the human gate)

   Read each company's candidate list and pick one primary and one alternate, using the title tiers below. An LLM with a tight rubric can do the first pass; spot-check it, because titles lie and one bad pick wastes both a reveal credit and a company.

3. 3

   Stage 3: reveal picks only (paid)

   Spend credits on the chosen two per company. Verified emails only by default: one batch of guessed addresses bouncing can undo two weeks of warmup on a young domain.

Who actually buys sponsorships

Pick the sponsorship buyer, not the first marketer you find. Growth product and performance marketing people build funnels; they don't buy placements. My batch 1 run: 15 companies scouted, 13 came back with a verified primary who plausibly owns the budget, and the misses were companies with no marketing candidates surfaced at all (which is itself a signal to deprioritize).

The fallback ladder: when your pick won't verify

1. 1

   Verified email revealed

   Use as-is. This should be the default outcome for most picks at real companies.

2. 2

   Unverified but a real address is present

   Usable. Tag it guess in your state file and watch bounces on that domain.

3. 3

   No email at all

   Infer from the company's dominant pattern, learned from VERIFIED colleagues at the same domain. If jane.doe@acme.com verified, john.smith@acme.com probably follows. Tag it pattern.

4. 4

   Only a senior person findable

   Use them, but switch the opener from a pitch to a routing ask: "are you the right person for this, or can you point me to whoever is?" Senior people forward; they rarely buy directly. Tag it routing: senior so the sequencer picks the right template.

You are picking cold-outreach contacts for newsletter sponsorship sales.

My newsletter: [NEWSLETTER NAME], audience: [AUDIENCE IN ONE CLAUSE].
Goal: find the person who BUYS sponsorship placements, not who builds growth funnels.

Here is the candidate list for [COMPANY] (title, seniority, department per person):
[PASTE CANDIDATES]

Rank using these tiers, best first:
1. Titles containing "creator", "sponsorships", "newsletter", "media buying"
2. Demand generation
3. Partnerships / business development
4. Brand marketing, field marketing
5. Developer relations
6. Community

Hard exclusions: talent acquisition, growth product, analyst/ops-only roles, brand ambassadors.

Seniority preference within a tier: manager/lead/head > director > VP (senior people route rather than buy; only pick a VP+ if nothing in tiers 1-4 exists below director, and flag them "routing: senior").

Output exactly:
PRIMARY: [name] | [title] | tier [n] | one-line reason (max 15 words)
ALTERNATE: [name] | [title] | tier [n] | one-line reason
FLAGS: ["routing: senior" if applicable, "weak company: no tier 1-4 candidates" if applicable, else "none"]

If no candidate lands in tiers 1 to 4, say SKIP COMPANY and one line why.

The templates: short, specific, no stats you can't screenshot

The full anonymized template set is in the download below. The rules baked into every one of them:

• [ ]60 to 110 words for the initial, shorter for follow-ups. Nobody reads cold essays
• [ ]One specific fit sentence per email referencing something checkable (a launch, a feature, a category). Never "I love what you're doing"
• [ ]No stats claims in email 1 unless you can screenshot them on the first call. Offer the numbers instead of asserting them
• [ ]One CTA, phrased as a question. No calendar link in the first touch
• [ ]Minimal signature: name, one-line identity, link to the real site. No images or banners (they hurt deliverability on young domains)
• [ ]A one-line opt-out in every email, no exceptions

Subject: sponsoring [YOUR NEWSLETTER NAME]

Hi [FIRST NAME],

I run [YOUR NEWSLETTER NAME], a newsletter for [AUDIENCE IN ONE CLAUSE, e.g. "people building software products with AI coding agents"].

[ONE SENTENCE OF REAL FIT: why THEIR product matches YOUR readers. Specific, e.g. "A lot of my readers deploy on [THEIR PRODUCT CATEGORY] and your launch of [SPECIFIC FEATURE] is exactly the kind of thing they ask me about."]

I'm opening [MONTH] sponsorship slots and thought of [THEIR COMPANY] first for [SPECIFIC REASON: a recent launch, a fit with an upcoming issue, an audience overlap].

Worth a short chat? Happy to send audience details and pricing either way.

[YOUR FIRST NAME]
[ONE-LINE IDENTITY, e.g. "Writer, [NEWSLETTER NAME]"]
https://yournewsletter.com

If sponsorships aren't your area or you'd rather not hear from me, just say so and I won't follow up.

Hi [FIRST NAME],

Last note from me on this. If newsletter sponsorship isn't a fit for [THEIR COMPANY] right now, all good, I'll close the file.

If it ever becomes relevant, the audience details live here: [LINK TO MEDIA KIT PAGE ON yournewsletter.com].

Thanks either way.

[YOUR FIRST NAME]

You are writing the [ONE SENTENCE OF REAL FIT] line for a cold sponsorship email.

Newsletter: [YOUR NEWSLETTER NAME], audience: [AUDIENCE DESCRIPTION].
Prospect company: [COMPANY], product: [WHAT THEY SELL], recent news: [PASTE 1-3 BULLETS FROM THEIR BLOG/CHANGELOG/LAUNCH].

Write 3 candidate fit sentences. Constraints:
- Max 25 words each.
- Must reference ONE concrete, checkable fact from the bullets (a feature, a launch, a pricing change). No flattery, no "I love what you're doing".
- Must connect that fact to a specific reader behavior or reader problem.
- Plain words. No "synergy", "leverage", "align". No exclamation marks.

Output: a numbered list of the 3 sentences, then one line saying which you'd pick and why in under 15 words.

Reply handling: one inbox, allowlist-filtered, thread-consistent

Run 4 domains with 2 mailboxes each and you have 8 inboxes; checking 8 webmail tabs is not a system, it's a way to miss a reply and lose a deal to latency. The whole reply side runs through one unified inbox pulled via the sending platform's API.

1. 1

   Poll the unified inbox on a schedule

   Every 15 minutes via cron works, with a daily hosted-cron backstop in case the local machine sleeps. Add a notification (Slack webhook, push, whatever you already run) on new replies.

2. 2

   Filter by allowlist, not by pattern

   Only surface messages from contacts you have actually emailed, by matching the sender against your prospect state file. The non-obvious reason this matters: warmup traffic arrives from random third-party domains and looks like real email. Allowlist membership excludes all of it cleanly; content pattern-matching never will.

3. 3

   Reply from the mailbox that owns the thread

   Always, via the platform's API, with In-Reply-To set. Never from your main-domain personal account and never from a different lookalike: it breaks threading, confuses the prospect, and leaks the architecture.

4. 4

   Let reply detection drive the sequencer

   The same poll that surfaces replies flips the company's reply flag in the state file, which is exactly what stops scheduled follow-ups. One source of truth, two consumers.

5. 5

   Advance the CRM stage on every touch

   queued, contacted, replied, interested, negotiating, booked, live or passed. When a sponsor asks "what did we agree in March", you'll have it.

Compliance: decide deliberately, not accidentally

Not legal advice; a map of the tradeoffs so you make each call on purpose instead of by omission. The practical minimum bar wherever you operate: a working opt-out in every email, every "no" honored permanently in your state file, verified emails only, no deceptive subjects, and a real website behind the sending domain.

Build order and what it costs

• [ ]Buy 3 to 5 lookalike domains (~$10 each)
• [ ]Add SPF with -all, DKIM, DMARC p=reject, and MX per domain; verify PASS/PASS/PASS via Gmail "Show original"
• [ ]301-redirect every lookalike root to the real site over HTTPS
• [ ]Create 1 to 2 human-named mailboxes per domain on the sending platform; start warmup immediately
• [ ]While warming (2+ weeks): build the target list, run scout searches, judge picks, reveal verified emails for picks only
• [ ]Write the sequence, render every email to drafts, review by hand
• [ ]Wire the interlocks in code: live sends refuse before the warmup-ready date, daily cap 10 to 15 per mailbox, reply flag stops sequences
• [ ]Go live at ~5/day per mailbox, ramp by ~2/day to the cap
• [ ]Poll the unified inbox every 15 minutes, allowlist-filtered, with notifications
• [ ]Reply same-hour from the owning mailbox and advance the CRM stage

Total cost to stand this up: roughly $30 to 50 in domains, $30 to 100/month for the sending platform, plus the contact database's credit plan at your volume (as of July 2026). One booked placement pays for months of the whole stack, which is the correct way to size the decision.

The downloads