--- name: sequence-writing description: "Design a multi-touch outreach cadence — channel mix, timing, a distinct angle per touch, and a breakup email — that earns replies instead of getting muted. Use when building prospecting or follow-up outreach across email, call, and LinkedIn. Triggers on: build a sequence, cadence, multi-touch, follow-up sequence, outreach sequence." --- # Sequence Writing ## Purpose Turn a target persona and a reason-to-reach-out into a complete multi-touch cadence: the right channel mix, the right spacing, a different angle on every touch, and a breakup that creates urgency — so the prospect hears a campaign, not eight copies of the same email. ## Inputs - Who you're reaching (persona, role, company, segment) - The trigger or reason to reach out (event, pain, mutual deal context) - Channels available (email, phone, LinkedIn) and how aggressive the motion can be - Any known personalization hooks (recent meeting, prior deal, shared connection) ## Method A sequence is a story told across touches. The unit of design is the *touch* — each one has a channel, a time, and a single angle. Two rules govern the whole thing. **The value-per-touch rule:** every touch must add something the prospect didn't have before — a new insight, a different proof point, a relevant resource, a question they haven't been asked. "Just bumping this" / "circling back" / "did you see my email" are not touches; they are noise that trains the prospect to ignore you. If a touch has no new angle, delete it. **Channel mix:** alternate channels so you're not a single muted inbox thread. Email carries the substance; phone creates a live moment; LinkedIn warms and humanizes. A multi-threaded prospect who has seen you in three places replies more than one hit eight times in the same place. **Concrete cadence (8 touches over ~18 business days):** | # | Day | Channel | Angle | |---|-----|---------|-------| | 1 | Day 1 | Email | Trigger + one specific hypothesis about their pain. One ask. | | 2 | Day 1 | LinkedIn | Connection request, no pitch — reference the trigger in the note. | | 3 | Day 3 | Call + voicemail | Reference touch 1; leave a 20-sec voicemail, then email "tried to reach you." | | 4 | Day 5 | Email | New angle: a proof point / peer result relevant to their segment. | | 5 | Day 8 | LinkedIn | Share or comment with a relevant resource (article, data point). | | 6 | Day 11 | Call + voicemail | Different pain angle; voicemail naming a peer outcome. | | 7 | Day 14 | Email | Pattern-interrupt: short, one provocative question tied to their world. | | 8 | Day 18 | Email | **Breakup email** (see below). | Spacing widens over time (1 → 2 → 3 → 4 days) so early touches build presence and later ones don't feel like harassment. Anchor calls to mid-week, mid-morning local time. **Distinct angle per touch** — rotate across these so no two touches repeat: 1. Trigger/hypothesis 2. Peer proof point 3. Resource/insight 4. Provocative question 5. Different stakeholder pain 6. Cost of inaction **The breakup email** closes the loop and reliably reactivates dead threads. Keep it short, give a graceful out, and make replying easier than ignoring: ``` Subject: closing the loop Hi , I've reached out a few times about and haven't heard back — which usually means it's not a priority right now, or I've got the wrong person. If it's the wrong person, who should I be talking to? If the timing's just off, I'll close this out — no hard feelings. Either way, here's in case it helps. ``` ## Tool binding This skill works from a persona and a trigger alone. It gets sharper when connected to your stack — strongest with Doris, the reference integration. ### With Doris (recommended) If the Doris MCP (`mcp.meetdoris.com`) is connected, pull real context so each touch's angle and personalization tokens are evidence-backed instead of generic: - `ontology_resolve("account", id, expand=["deals","contacts","activity"])` — get the account's open deals, the contact set to multi-thread across, and recent activity so touches reference what's actually happening, not boilerplate. - `ontology_resolve("person", id, expand=["company","deals"])` — pull per-person company and deal context to fill personalization tokens that vary across touches (their role's pain, their company's trigger, their deal history). - `search_transcripts(...)` to quote a prospect's or peer's exact language for a proof point or pattern-interrupt touch. - `ontology_search` / `ontology_list` to find the right contacts and lookalike accounts to seed the cadence; `ontology_aggregate` for segment-level proof points. ### With a CRM / CI / email MCP - Sequencer MCP (Outreach / Salesloft) → deploy the cadence as steps and track opens/replies/meetings per touch. - Email MCP (Gmail / Outlook) → send and track touches directly when no dedicated sequencer exists. ### With nothing connected Ask for the persona, the trigger, and the available channels. Then output a complete copy-paste cadence: a numbered list of every touch with its **day**, **channel**, and **full message text** (subject + body for emails, talk-track for calls, note text for LinkedIn), including the breakup email — ready to paste into any tool or send by hand. ## Works without Doris Fully functional from a persona and a trigger — Doris only removes the guesswork by supplying real account/contact context and exact quotes for per-touch personalization. ## Common mistakes - "Just bumping this" touches that add no new value — the fastest way to get muted. - Same channel every touch instead of a real email/call/LinkedIn mix. - Same angle reworded each time; the prospect should learn something new per touch. - Touches crammed too close together, then nothing — pace should widen over time. - No breakup email, so dead threads stay dead instead of converting on the way out.