# Cold Email Campaign Playbook **Portable operational workflow for running a cold-email launch campaign for any new SaaS app in the portfolio.** This is NOT a coded subsystem — it's a discipline + folder convention + decision-log pattern. The reference implementation lives at `reference app` (Kompozy → BILT-warm RE list, ~100k addresses). For SPF/DKIM/DMARC, IP/domain warmup mechanics, and inbox provisioning, see the sibling spec: `templates/s-tier/EMAIL-INFRASTRUCTURE-PLAYBOOK.md`. --- ## 0. Use this spec **Inputs:** - App name (becomes the folder slug — `-COLD-EMAIL`) - Audience source: cold (no prior contact) / warm (engaged with prior campaigns) / hot (paying users of sibling apps) - Offer: tier name, price, expiry, link destination - Sending stack: Instantly / Smartlead / lemlist + how many inboxes are warmed - Stripe products that EXIST today (not what you wish existed — see "Pricing-promise hygiene") **Outputs:** - Working folder under `Active/-COLD-EMAIL/` with versioned email copy, append-only daily metrics, and a rolling DECISIONS log - A/B/C subject test result that picks the next batch's variant on **reply rate**, not opens - Landing-page mismatch diagnosis the moment CTR > 5% with conversion = 0 - A defensible ramp schedule from 20 emails/inbox/day to scaled batches --- ## 1. Folder structure Every campaign starts as: ``` Active/-COLD-EMAIL/ ├── README.md # Campaign goal + status + index of all files ├── email-v1.md # Current canonical email copy (subject + body) ├── email-v2.md # Iteration after first batch result ├── subject-line-test.md # A/B/C variant table + test recipe ├── metrics-v1.md # First-batch send (initial baseline) ├── metrics-2026-05-18.md # Daily append-only — never overwrite ├── metrics-2026-05-19.md ├── metrics-2026-05-20.md ├── landing-diagnosis.md # Filed when CTR is healthy but conversions are zero ├── DECISIONS.md # Rolling log: date · change · hypothesis · result └── assets/ # Screenshots from Instantly/Smartlead, landing-page captures ``` **The hard rule: metrics files are append-only.** Each day = a new file. Never edit yesterday's file. The timeline IS the data — overwriting destroys the empirical basis for every later decision. Reference: `Active/KOMPOZY-COLD-EMAIL/README.md`. --- ## 2. README.md template ```markdown # Cold Email Campaign One-line campaign description (single email? sequence? to what list?). ## Campaign goal - **Audience:** ~k (cold / warm / hot). - **Offer:** — /mo, . - **Format:** Single email / 3-step sequence / etc. - **Landing destination:** (dedicated /founders route? or homepage?). ## Files - [email-v1.md](./email-v1.md) — Current email body. - [subject-line-test.md](./subject-line-test.md) — A/B/C test plan. - [metrics-v1.md](./metrics-v1.md) — First-batch baseline. - [landing-diagnosis.md](./landing-diagnosis.md) — Filed if CTR healthy + conversions zero. - [DECISIONS.md](./DECISIONS.md) — What changed when and why. ## Status - **Sent so far:** emails across batches. - **Avg open rate:** <%> (unreliable — see §5). - **Avg reply rate:** <%> — primary metric. - **Avg click rate:** <%>. - **Conversions:** trial / paid. ## Next decisions 1. 2. ``` --- ## 3. email-v1.md template ```markdown # Email v1 — cold to ## Subject ## Preheader <60-90 chars — first thing shown after subject in iOS Mail / Gmail preview> ## Body Hey {{first_name}}, <2-3 sentence cold open. Reference the segment specifically — "real estate agents", "Shopify store owners", etc. Pattern-match the reader into the page.> {{cta_url}}?utm_source=cold&utm_medium=email&utm_campaign=-v1&utm_content= — Founder, ## Sender From: > Reply-to: moe@ ``` Reference: `Active/KOMPOZY-COLD-EMAIL/email-v1.md`. --- ## 4. subject-line-test.md template A/B/C protocol — 2,500 recipients per variant, decide by **reply rate**, not opens. ```markdown # Subject Line Test — v1 ## Test slice - Total recipients in test: 7,500 (2,500 per variant) - Sending window: across inboxes - All other variables held constant (same body, same CTA, same landing page) ## Variants | Variant | Subject | Preheader | Hypothesis | |--|--|--|--| | A | "" | "" | Curiosity beats specificity for cold | | B | "I built this for you ()" | "" | Direct segment naming beats curiosity | | C | "" | "" | Benefit-first beats both | ## Decision criteria (in priority order) 1. **Reply rate** — primary. >2% on cold is the "scale this" bar. 2. **Conversion rate** — secondary. Replies are leading indicator; conversions confirm. 3. **Click rate** — tertiary. Useful only if landing page is trusted (see landing-diagnosis.md). 4. **Open rate** — IGNORE for ranking. Apple MPP inflates 20-40%; tracking pixels are blocked by Gmail privacy on a lot of segments. Open rate is noise. ## Result (filled in after test) Winner: . Reply rate <%> vs <%> vs <%>. Logged in [DECISIONS.md](./DECISIONS.md). ``` --- ## 5. metrics-YYYY-MM-DD.md template (append-only) **One file per day. Never overwrite.** User pastes screenshot or numbers from Instantly/Smartlead → new dated file → update README status block → append to DECISIONS.md. ```markdown # Metrics — Cold — 2026-05-19 ## Send window - Batches sent today: - Recipients today: - Cumulative recipients: ## Numbers (from dashboard) | Metric | Today | Cumulative | Rate | |--|--|--|--| | Sent | | | — | | Delivered | | | <%> | | Bounces (hard) | | | <%> | | Bounces (soft) | | | <%> | | Opens | | | <%> (unreliable) | | Clicks | | | <%> | | Replies | | | <%> ← primary | | Conversions (trial) | | | <%> | | Conversions (paid) | | | <%> | | Unsubscribes | | | <%> | | Spam complaints | | | <%> ← if >0.1% STOP | ## Reply breakdown (qualitative — leading indicator) - Positive ("interested", "tell me more"): - Neutral ("what's the price", "how does it work"): - Negative ("unsubscribe", "wrong person"): - Hostile ("spam", "report"): ← if >0 in a batch, pause and review ## Screenshot  ## Notes ``` Reference: `Active/KOMPOZY-COLD-EMAIL/metrics-v1.md`. --- ## 6. DECISIONS.md template (rolling log) The point of this file: correlate "I changed X on date Y → metrics moved Z" so optimization becomes empirical, not vibes-based. Every change goes in. Every change. ```markdown # Decisions Log — Cold Email | Date | Change | Hypothesis | Result (filled in after 3 days of metrics) | |--|--|--|--| | 2026-05-15 | Initial send, email-v1, subject A, to 538 recipients | Baseline | 1.42% click, 0 conversions — see metrics-v1.md | | 2026-05-18 | Filed landing-diagnosis.md — CTR healthy, conversion zero | Promise mismatch on landing | Pending /founders build | | 2026-05-19 | Launched A/B/C subject test on 7,500 slice | Direct segment naming beats curiosity | Pending — read 2026-05-22 | | 2026-05-22 | Variant B won (reply rate 2.3% vs 1.1% vs 0.8%) | — | Rolling B to remainder of list | | 2026-05-23 | Re-pointed CTA from `/` to `/founders` | Promise match → conversions | Pending | ``` **Rule: never change two variables in the same row.** If you swap subject AND landing page on the same day, you can't attribute the lift. Stagger by at least 48h or run them as separate A/B slices. --- ## 7. List segmentation Three temperatures. Pick one per campaign. Don't mix in the same send. - **Cold** — no prior contact with anyone in the portfolio. Lowest reply rate (~0.5-1.5% on a good day), highest spam risk, requires the most warmup discipline. Used when launching to scraped/purchased lists. - **Warm** — engaged with prior campaigns from a sibling app (BILT CRM's ~100k RE list is the canonical warm pool). Reply rates 2-4x cold; deliverability much friendlier because most addresses are still active. **Most marketing apps in the portfolio launch warm**, including Kompozy → BILT-RE. - **Hot** — paying customers of a sibling app. Treat as transactional, not cold. Different inbox, different sender (founder-personal vs. brand), reply rates 8-15%. For warm sends: still ramp. The list has been emailed before but not from your new sending domain — recipients haven't seen this `From:` address before, ESPs haven't built a reputation on it. --- ## 8. Ramp schedule New sending stack ramps over **6+ weeks** before hitting full volume. Skipping this kills the domain — and there's no recovery, only buying a new one. ``` Week 0 : Buy domain. Set up SPF/DKIM/DMARC (see EMAIL-INFRASTRUCTURE-PLAYBOOK.md). Week 1-2 : Instantly/Smartlead automated warmup. 14 days minimum, ZERO real sends. Week 3 d1 : 20 emails/inbox/day to warm list only. Reply-rate check on day 3. Week 3 d4+ : If reply rate >1% and spam <0.05%, hold 20/day for 7 days. Week 4 : Ramp 10% week-over-week → ~22/inbox/day. Daily metrics check. Week 5 : ~24/inbox/day. Watch bounce rate — hard bounces >2% means list quality issue, NOT volume issue. Week 6+ : Continue 10%/week ramp until target throughput. Stop ramping if any week shows reply-rate decline or spam-complaint uptick. ``` **Inbox math:** 20/inbox/day × 10 inboxes = 200/day = 6,000/month. To send 100k in 30 days you need ~17 fully-warmed inboxes pulling 200/day each — plan inbox provisioning 6 weeks ahead of the campaign launch date. --- ## 9. Landing-page-mismatch diagnosis **Trigger:** CTR is healthy (>5% among openers, or >1% of delivered) and conversion is zero or near-zero. The email is doing its job; the landing page isn't. File `landing-diagnosis.md` with: 1. **Ground-truth mobile capture.** Open the destination URL on a fresh mobile profile (DevTools iPhone emulation works). Screenshot above-the-fold. Most cold-email clicks are mobile; the desktop view is irrelevant. 2. **List what the email promised.** Subject + opening line + the offer terms. Verbatim. 3. **List what the landing page shows.** Hero, sub, primary CTA, secondary CTA, what's NOT visible without scrolling. 4. **Map the mismatch.** Where does the page fail to deliver the email's promise? Common failures: generic SaaS pitch when email named a segment, free-trial CTA when email built urgency around a paid offer, $X price hidden 3 scrolls down behind "See Pricing". 5. **Prescribe the high-leverage fix.** Usually a dedicated route (`/founders`, `/`, `/`) with hero that mirrors the email subject + single CTA + the offer's price/expiry visible without scrolling. 6. **Re-point the email link, then resume sending.** Don't blast the rest of the list against a known-broken funnel. The Kompozy v1 case: 6,912 sent → 98 clicks → 0 conversions. Diagnosis identified the homepage as the bottleneck, not the email. Stopped the blast at 7k instead of burning the remaining 93k against a broken funnel. --- ## 10. Pricing-promise hygiene **Never promise pricing/features in cold email that don't exist in Stripe today.** The Kompozy v1 email mentioned a tiered ladder ($39 → $59 → $79 every 100 signups) that wasn't wired into Stripe products — replies asking "how do I lock in $39?" couldn't be answered with a working checkout link. Cost: ~3 days of campaign momentum while the Stripe products were built. Before sending v1: - Open Stripe dashboard. Confirm every tier/price/feature mentioned in the email has a live product. - Confirm the signup link routes to a checkout that actually charges the promised price. - Confirm any "founding member" / "early bird" / "locked for life" flag has a DB column or Stripe metadata key that survives renewal. This is the single most expensive mistake to make in a cold campaign — every recipient who replied "interested" while the checkout was broken is a soft-burn on the list. --- ## 11. Reply handling - **Per-app inbox.** Don't route replies to a personal address — they get lost in the noise and conversion attribution dies. Use `moe@` or `founders@` with a shared inbox tool. - **Reply rate is the primary metric.** Track it daily in metrics files. Cold-RE benchmark: >2% is "scale this", 0.5-1.5% is "iterate copy", <0.5% is "rethink offer or audience." - **Qualitative replies lead conversions by 2-3 weeks.** If reply tone shifts negative (more "unsubscribe" / "wrong person" / "this is spam"), expect conversion rates to drop in 2-3 weeks — pause and iterate now, don't wait for the numbers to confirm. - **Conversion attribution.** UTM params on every CTA link (`utm_source=cold&utm_medium=email&utm_campaign=-v&utm_content=`). Signup form persists UTMs into `profiles.signup_source` so attribution survives even when the user comes back days later via direct URL. --- ## 12. Common pitfalls - ❌ **Don't blast cold without warmup.** Fresh IP + fresh domain + cold list = inbox provider auto-spam in under 100 sends. 14-day warmup is non-negotiable. - ❌ **Don't optimize on opens.** Apple MPP inflates open rates 20-40%; Gmail privacy strips the tracking pixel on a meaningful share of recipients. Opens are noise in 2026. Rank by reply rate. - ❌ **Don't change multiple variables at once.** A/B/C tests must isolate ONE variable per round (subject only, OR body only, OR CTA only, OR landing only). Otherwise the result is unattributable. - ❌ **Don't overwrite metrics files.** Append-only. Each day = new file. The timeline is the data. - ❌ **Don't promise pricing/features that don't exist in Stripe.** See §10. Kompozy v1 paid for this mistake. - ❌ **Don't ignore qualitative reply patterns.** They're a 2-3 week leading indicator before quant data confirms. - ❌ **Don't scale through a broken funnel.** If conversion is zero after 5k+ sends with healthy CTR, file landing-diagnosis.md and STOP sending until the funnel is fixed. - ❌ **Don't share an inbox between cold campaigns and personal mail.** Reply attribution breaks; deliverability reputation cross-contaminates. --- ## 13. Kickoff prompt > Execute the Cold Email Campaign Playbook at `COLD-EMAIL-CAMPAIGN-PLAYBOOK.md` for ``. Mirror the Kompozy reference at `Active/KOMPOZY-COLD-EMAIL/`. > > Inputs: > - App name + sending domain: [provided] > - Audience: cold / warm / hot — [pick] > - List source + estimated size: [provided] > - Offer (tier name, price, expiry): [provided] > - Stripe products live for this offer: [yes / building] > - Sending stack (Instantly / Smartlead) + warmed inbox count: [provided] > > Step 1: Create `Active/-COLD-EMAIL/` with README.md, email-v1.md, subject-line-test.md, DECISIONS.md from the templates above. > Step 2: Audit the offer against Stripe before drafting (§10). Halt if any promise lacks a live product. > Step 3: Draft email-v1 + 3 subject variants. Pause for review before sending. --- **Reference files in the Kompozy campaign:** - `Active/KOMPOZY-COLD-EMAIL/README.md` — campaign goal + status - `Active/KOMPOZY-COLD-EMAIL/email-v1.md` — current canonical copy - `Active/KOMPOZY-COLD-EMAIL/subject-line-test.md` — A/B/C variant table - `Active/KOMPOZY-COLD-EMAIL/metrics-v1.md` — first-batch baseline - `Active/KOMPOZY-COLD-EMAIL/landing-diagnosis.md` — 98 clicks → 0 conversions postmortem