# FIXXO — Email Outreach

**Companion** `outreach/COMPLIANCE.md` · `outreach/SMS.md`

> Email is the one channel where **cold outreach is lawful** in the US. CAN-SPAM is an opt-**out** regime, not a consent-first regime, and the FTC's Telemarketing Sales Rule exempts most B2B calls (and by extension B2B email is not consent-gated the way SMS and AI voice are). That makes email our primary volume channel for the first cold touch — with two hard constraints.

---

## 1 · The two hard constraints

1. **CAN-SPAM (15 U.S.C. § 7701)** — accurate `From`/`Reply-To` and routing; **non-deceptive subject line**; identify the message as an advertisement; a valid physical postal address; a clear opt-out that works for **30 days** and is honoured within **10 business days**.
2. **Washington CEMA (RCW 19.190.020)** — narrower and stricter for WA recipients: no false or misleading subject lines, no misrepresented origin. A single WA resident receiving a misleading subject line is a per-message claim. **Washington recipients get their own subject-line review.**

Email is not a licence to be sloppy. It is a lawful channel that must be operated precisely.

---

## 2 · Sequence E1 — cold B2B to a registry-verified business

| # | Day | Subject | Body intent |
|---|---|---|---|
| 1 | 0 | `Question about lead costs at {{business}}` | One-line context, one question, no pitch attachment |
| 2 | +3 | `No fee to quote — the Fixxo model in 40 words` | The fee structure, plainly |
| 3 | +7 | `{{trade}} work in {{city}} — verification takes 15 min` | The ask: a verification slot |
| 4 | +14 | `Closing the file on {{business}}` | Polite close, leave the door open |

**E1-Message-1 (the whole email — this is the one that matters):**

> Subject: `Question about lead costs at {{business}}`
>
> Hi {{first_name}} —
>
> I run supply at Fixxo, a home-repair platform launching in {{city}}.
>
> Quick question: when you buy leads from the big platforms, roughly how many turn into actual work — and do you get the ones that don't refunded?
>
> We built the opposite model: nothing to see a job, nothing to quote it, 15–18% only when the job is finished and the customer has approved it. Pros keep 82–85%, paid 3–7 days after a photo check.
>
> If that's interesting, I can send the detail. If not, no follow-up from me.
>
> — {{sender_name}}, Fixxo
> {{physical_address}} · Unsubscribe: {{unsubscribe_url}}

Four sentences of substance, one question, honest out. No "I hope this finds you well". No attachments on message 1 — attachments and links on a first cold touch suppress deliverability and look like spray-and-pray.

---

## 3 · Sequence E2 — women-owned partners

Same skeleton, different opening, because the pitch genuinely differs:

> Subject: `Partner opportunity for {{business}} — {{city}}`
>
> Hi {{first_name}} —
>
> Fixxo is launching a women-first home-repair service in {{city}}: customers can request a woman technician, and women-owned businesses get priority matching.
>
> It's a commission model, not a lead fee — nothing to see a job, 15–18% only when it's done and approved. And the pay is published, not negotiated per pro.
>
> If you'd like the detail I'll send it. If the timing's wrong, tell me and I'll leave it.
>
> — {{sender_name}}, Fixxo

Warm, specific, no pink, no "empower". The offer is priority matching and published pay — commercial facts, not sentiment.

---

## 4 · Sequence E3 — referral partners (realtors, property managers)

Not supply. Demand-side local partners who refer women homeowners and busy landlords.

| # | Day | Subject | Intent |
|---|---|---|---|
| 1 | 0 | `Cover for your {{city}} clients` | The offer: vetted trades, fixed price, escrow |
| 2 | +4 | `What a Fixxo job looks like end to end` | One real example, no fluff |
| 3 | +10 | `Closing gift vouchers for your buyers` | A concrete partnership |

---

## 5 · Rules

- **One CTA per email.** Message 1 asks one question. Message 3 asks one thing.
- **Plain text, no tracking pixels** on cold outreach — a pixel in a B2B cold email is a deliverability and privacy liability with no upside.
- **No attachments on touch 1.** Link in a later touch, from a warmed domain.
- **Domain hygiene:** send from a subdomain (`partners.fixxo.com`), warmed over 3–4 weeks, SPF + DKIM + DMARC configured. A single spam complaint rate above 0.3% gets the domain throttled.
- **No false scarcity, no "we're onboarding only 10 pros in {{city}}"** unless it is literally true and provable.
- **No emoji.** Same as every other channel (`BRAND.md` §4).
- **Personalisation tokens must resolve.** A missing `{{first_name}}` or `{{business}}` kills the send; it never ships as `{{first_name}}`.
- **Unsubscribe is one click, no login, honoured within 10 days**, and writes to the same global `suppression_list` as SMS and calls.
- **Emails to Washington recipients get a subject-line review pass** before launch (CEMA).

---

## 6 · The AI's role in email

The agent drafts; a human approves the first send of every new sequence. After that:
- AI writes the first-draft personalisation line per prospect from public data (their licence record, their suburb, their review count).
- AI classifies replies: `interested` · `not_now` · `not_a_fit` · `unsubscribe` · `question`.
- AI drafts the reply for a human to send — it does **not** auto-send replies until the sequence has a 30-day clean compliance record.
- AI never invents a fact about the business. If it does not know their crew size, it does not mention it.

---

## 7 · Metrics

| Metric | Target |
|---|---|
| Deliverability (inbox, not spam) | > 95% |
| Open rate (plain text, no pixel — measured by reply and click only) | n/a |
| Reply rate touch 1 | 3–8% |
| Positive reply rate | 1–3% |
| Unsubscribe rate | **< 0.5%** |
| Spam complaint rate | **< 0.1%** (0.3% kills the domain) |
| Compliance (address, opt-out, honest subject on every send) | **100%** |