はじめる
Cold Outreach Board — 3-Lens Email Review

Cold Outreach Board — 3-Lens Email Review

Review a cold email before sending: spam-filter triggers, self-focused tone and weak asks, with findings the same model cross-checks against each other.
#営業
評価
さらに評価が必要です
販売数
0
利用方法
ダウンロード

Review a cold email before sending: spam-filter triggers, tone that reads as about-me, and weak asks that get no reply, with findings the same model checks against each other (CONFIRMED/CORROBORATED/DISPUTED).

How it works

Reviews a cold outreach email (sales, partnership, guest post, investor, or job outreach) using three reviewer lenses: a Spam Filter Simulator, a Recipient Empathy Check, and a Response Rate Realist. Each lens flags issues in turn, then the lenses cross-examine each other's findings. A board verdict tags every finding CONFIRMED, CORROBORATED, or DISPUTED, ending with a send recommendation. It grounds every finding in the email text you paste, does not invent facts about the recipient, and asks for recipient context if none is given. It does not send emails, look up recipients, or test real deliverability; it reviews the text as pasted, so redact anything you do not want shared. It reviews wording and deliverability signals only; it does not check legal requirements for commercial email (consent, sender identification, opt-out), so check the rules that apply to you before sending. The lenses are roles applied by one model (or by sub-agents if your agent supports them), so the cross-checks are self-checks; spot-check the findings.

Sample output

Sample input (an invented example):

Use the Cold Outreach Board on this (a sample I wrote). I'm sending it to a marketing lead I've never spoken to; no other context on her beyond her title at a mid-size retailer.

Subject: FREE audit!!! Guaranteed 3x ROI on your email campaigns

Hi there,

I hope this email finds you well. My name is Sam and I am the founder of BrightLoop Analytics (brightloop.example.com), the leading provider of AI-powered email optimization. We have helped hundreds of companies increase revenue, and we would love to do the same for you.

Our platform offers A/B testing, send-time optimization, predictive segmentation, churn scoring, a dashboard, and a Slack integration. Check out our site, our case studies at example.com/cases/1, example.com/cases/2, and our pricing at example.com/pricing. I've also attached our 24-page brochure.

LIMITED TIME: sign up this week and get a FREE audit worth $2,000!

Would you be open to a 45-minute demo call sometime next week? Or I can send over more info, or you could just sign up on the site, whichever works.

Best,
Sam

Real, unedited output:

## Phase 1 — Lens findings
Mode: one lens at a time (each lens finished before the next started; no sub-agents)

*Context note: the only thing known about the recipient is her title at a mid-size retailer. Personalization findings are best-effort. Her name, company, and one recent campaign or launch would allow a sharper second pass.*

### 🚫 Spam Filter Simulator
- **S1. Subject line is a trigger stack.** "FREE audit!!! Guaranteed 3x ROI" combines all-caps FREE, triple exclamation, "Guaranteed," and a numeric ROI promise. That is close to a textbook filter pattern. Consequence: spam folder or Promotions tab before she ever sees it. Fix: use a plain, specific subject, e.g. "Question about [Company]'s email program."
- **S2. "LIMITED TIME: … FREE audit worth $2,000!"** This line adds caps, urgency, "FREE," and a dollar-value lure. Consequence: it compounds S1's score and reads like a promo blast. Fix: delete the line, or restate the offer plainly (see R6).
- **S3. Too many links for a first touch from an unknown sender.** There are four: brightloop.example.com, /cases/1, /cases/2, /pricing. They also sit on two different domains (`brightloop.example.com` vs. `example.com`). Consequence: link density plus a domain mismatch is a phishing-style signal. Fix: use zero links, or one link to a single relevant case study on the sending domain.
- **S4. A 24-page brochure is attached to a cold email.** Consequence: unknown-sender attachments raise filter and security flags, and many corporate gateways strip or quarantine them. Fix: remove it and offer to send it only if she replies.
- **S5. "Hi there."** There is no name and no company reference anywhere in the email. Consequence: it reads as a mail-merge blast, which filters and humans both discount. Fix: use her name and her company.
- **S6. The subject doesn't match the body.** The subject promises a "Guaranteed 3x ROI," but the body never mentions 3x or a guarantee again. Consequence: bait-and-switch patterns hurt sender reputation when recipients mark the email as spam. Fix: subject and body should make the same claim, or neither should make it.

### 🎭 Recipient Empathy Check
- **E1. The opening is entirely about the sender.** "I hope this email finds you well. My name is Sam and I am the founder of BrightLoop…" The first two sentences are a filler pleasantry and a self-introduction. Consequence: she has no reason to keep reading past the preview pane. Fix: open with something about her or her company's email program.
- **E2. No evidence of homework.** Nothing references retail, her company, her campaigns, or anything she has done. Consequence: she feels targeted, not understood. Fix: add one specific, true observation, e.g. a recent promo email of theirs or a seasonal retail angle.
- **E3. "The leading provider" and "helped hundreds of companies" are vague claims.** Consequence: they cost credibility, especially with a marketer who writes this kind of copy herself. Fix: replace them with one concrete, verifiable result from a similar retailer.
- **E4. A feature dump.** "A/B testing, send-time optimization, predictive segmentation, churn scoring, a dashboard, and a Slack integration" is six features and zero of her problems. Consequence: it reads as a brochure. Fix: pick the one capability most likely to matter to a mid-size retailer and tie it to an outcome.
- **E5. The email asks a stranger for a lot of effort.** She would have to read 24 pages, open four links, and give up a 45-minute call. Consequence: the cost of engaging is far higher than her interest in an unknown sender. Fix: ask for one small thing.
- **E6. The tone is wrong for the audience.** "FREE!!!" and "LIMITED TIME" are growth-hack tactics aimed at a marketing lead who uses and recognizes them. Consequence: it signals low sophistication about exactly the field being sold into. Fix: use a calm, peer-to-peer tone.

### 📊 Response Rate Realist
- **R1. Three asks at once.** "45-minute demo call… Or I can send over more info, or you could just sign up on the site, whichever works." Consequence: when every option is offered, none is clearly the next step, and "whichever works" invites "none." Fix: make one ask.
- **R2. The ask is too big for a first touch.** A 45-minute demo is a heavy commitment to a stranger. Consequence: the default answer is silence. Fix: ask for a yes/no reply or a 10–15 minute call.
- **R3. The ask is buried.** It doesn't appear until the last paragraph, after the intro, the features, the links, and the promo. Consequence: someone skimming the first two lines never learns what is being asked. Fix: state the ask, or at least the reason for writing, in the first two lines.
- **R4. No genuine "why you" or "why now."** The only time pressure is the sender's own promo ("sign up this week"). Consequence: there is no reason this matters to her this week. Fix: tie timing to her world, e.g. holiday or Q1 campaign planning for retail.
- **R5. "Just sign up on the site" asks for a purchase decision before any conversation.** Consequence: it skips the trust-building step entirely. Fix: drop it from the first email.
- **R6. The best hook is in the wrong place.** The free audit is actually a strong low-friction offer, but it's framed as a sign-up incentive and buried in a spammy promo line. Consequence: the most reply-worthy part of the email is the part most likely to be ignored. Fix: make the audit the single ask, e.g. "Want me to send three quick observations on your last few campaigns? No call needed."

## Phase 2 — Cross-examination
- **[S1 Subject triggers]**
  - Empathy corroborates: "the hype tone is wrong for a marketer (E6)."
  - Response Rate corroborates: "even if it lands in the inbox, a clickbait subject lowers open rates."
  - No challenge.
- **[S2 LIMITED TIME line]**
  - Response Rate corroborates: "this is the artificial urgency from R4."
  - Empathy corroborates: E6.
  - No challenge.
- **[S3 Link count/domain mismatch]**
  - Response Rate partially challenges: "one relevant case-study link can help credibility, so the problem is the count, not links in general." This sharpens the fix; it isn't a dispute.
  - Empathy challenges the domain mismatch: "the user said this is a sample, so `brightloop.example.com` vs. `example.com` is likely a placeholder artifact, not a real mismatch."
  - Empathy corroborates the link count via E5.
- **[S4 Attachment]**
  - Response Rate challenges the filtering mechanism: "a single PDF from a clean domain isn't reliably filtered. The more certain harm is that nobody opens a 24-page attachment from a stranger."
  - Empathy corroborates the effort side via E5.
- **[S5 "Hi there"]**
  - Empathy corroborates: E2.
  - No challenge.
- **[S6 Subject/body mismatch]**
  - Empathy corroborates: "bait-and-switch reads as manipulative."
  - No challenge.
- **[E1 Self-focused opening]**
  - Response Rate corroborates: R3, the first two lines contain no ask or reason.
  - Spam corroborates: "the boilerplate opener is a template signal."
- **[E2 No homework]**
  - Spam corroborates: S5.
  - Response Rate corroborates: R4, no "why you."
- **[E3 Vague claims]**
  - Response Rate challenges: "social proof can raise replies." This is not substantiated as a reason to keep it, because the claim here is unspecific. The finding stands, with the fix framed as "make it specific," not "remove it."
- **[E4 Feature dump]**
  - Response Rate corroborates: "length buries the ask (R3)."
  - No challenge.
- **[E5 Effort]**
  - Response Rate corroborates: R2.
  - Spam corroborates: S3 and S4.
- **[E6 Tone mismatch]**
  - Spam corroborates: S1 and S2.
  - No challenge.
- **[R1 Three asks]**
  - Empathy corroborates: "'whichever works' shifts the work of deciding onto her."
  - No challenge.
- **[R2 45-min call]**
  - Empathy corroborates: E5.
  - No challenge.
- **[R3 Buried ask]**
  - Empathy corroborates: E1.
  - No challenge.
- **[R4 No why-now]**
  - Spam corroborates: "the only urgency offered is the trigger-word kind."
  - No challenge.
- **[R5 Sign-up ask]**
  - No corroboration and no challenge.
- **[R6 Audit as hook]**
  - Spam corroborates with a caveat: "only if reframed. 'FREE … worth $2,000!' is the trigger phrasing."
  - Empathy corroborates: "an audit of *her* campaigns is the only recipient-centric element in the email."

## Phase 3 — Board verdict
1. **[CORROBORATED] The subject line is a spam-trigger stack and promises a claim the body never makes** (S1 + S6 + E6). Fix: use a plain, specific subject, drop "FREE," "!!!," "Guaranteed," and "3x ROI," and keep the subject consistent with the body.
2. **[CORROBORATED] The opening is about Sam, not her, with a generic greeting and no homework** (E1 + E2 + S5 + R3). Fix: use her name and company, and open with one specific observation about their email program.
3. **[CORROBORATED] There are three asks, the main one is a 45-minute demo, and it's buried at the end** (R1 + R2 + R3 + E5). Fix: make one small ask, stated early.
4. **[CORROBORATED] The "LIMITED TIME: FREE audit worth $2,000!" line** (S2 + R4 + E6). It's a trigger phrase, the urgency is artificial, and the tone is wrong. Fix: delete it as written.
5. **[CORROBORATED] The audit should be the hook, not a sign-up incentive** (R6 + E-lens + Spam with caveat). Fix: "Happy to send 3 quick observations on your recent campaigns. Want them?"
6. **[CORROBORATED] Link and attachment load** (S3 + S4 + E5). Fix: no attachment, and at most one link.
7. **[CONFIRMED] The feature dump** (E4). Fix: name one capability tied to one retail outcome.
8. **[CONFIRMED] Vague superiority claims** (E3). Fix: replace with one concrete result from a comparable retailer, or cut them.
9. **[CONFIRMED] "Just sign up on the site"** (R5). Fix: remove it from the first touch.
10. **[DISPUTED] The domain mismatch between links** (S3, partial). Challenge: "likely a placeholder artifact, since this is a sample." Your call: if the real links use different domains from your sending domain, fix it. Otherwise ignore it.
11. **[DISPUTED] The attachment as a deliverability risk specifically** (S4). Challenge: "a single PDF isn't reliably filtered, and the certain harm is effort, not filtering." Removal is still recommended under item 6 either way.

**Send recommendation:** Rewrite first. There are 6 CORROBORATED blockers (spam-bait subject, self-focused opening, buried multi-ask, the promo line, the misused audit hook, link and attachment load). This version would likely be filtered, and if it reached her it would likely be ignored.

For reference, a rewrite applying the fixes (bracketed parts need real details):

> **Subject:** Quick idea for [Company]'s holiday emails
>
> Hi [Name],
>
> I noticed [specific thing, e.g. "your Black Friday preview email went out at 6am ET"]. For mid-size retailers we've seen [one concrete result, e.g. "send-time tuning lift click-through ~18% for [similar retailer]"].
>
> Want me to send three quick observations on your last few campaigns? No call needed, just a reply with "sure."
>
> Sam
> Founder, BrightLoop Analytics

Out of scope for this review: legal requirements for commercial email (sender identification, opt-out, consent rules where she's located). Check those separately before sending.

Legal limits

It reviews wording and deliverability signals only; it does not check legal requirements for commercial email (consent, sender identification, opt-out), so check the rules that apply to you before sending.

Handling sensitive info in the outreach

If the email references real names, companies, or contact details, treat that as context only — do
not echo it back beyond what's needed to explain a finding. This Skill sends the text you paste
as-is to the model; redact anything you'd rather not share before pasting.

Works with

Built for Claude Code; other agents that read SKILL.md files (such as Codex CLI or OpenClaw) should also work. No network access or extra tools needed.

Anthropic, OpenAI and OpenClaw are trademarks of their respective owners. This Skill is an independent product and is not affiliated with, endorsed by, or sponsored by any of them.


プライバシーポリシー

https://glowhalo.github.io/privacy/