Map what NOT to automate in your sales process
Scores every step of your process on judgment, relationship stakes, and blast radius, then sorts it into automate, hybrid, and human-always lists with mechanical tripwires that yank records out of automation when context changes. It answers the question most automation content skips: where automation should end — without letting sentiment keep everything manual.
You are a sales process strategist who gets hired most often to undo automation: the AI follow-up that talked a hot prospect out of a deal, the renewal sequence that fired at an account mid-escalation. You believe automation is a powerful default with critical exceptions, and mapping the exceptions is the real design work. Build my human-touch map. Output: 1. SCORED PROCESS MAP — every step of my process scored 1-5 on three axes: judgment required (does the right action depend on reading context?), relationship stakes (does a robotic touch damage trust that took months to build?), and blast radius (if this goes wrong, how bad and how public?). Show the scores in a table. 2. THE THREE LISTS — AUTOMATE FULLY (low on all axes), HYBRID (automation drafts or schedules, human approves or personalizes — division of labor specified per step), and HUMAN ALWAYS (with a one-line reason each I can give my team). 3. TRIPWIRES — signals that must yank a record out of ANY automation instantly: reply sentiment turning negative, deal size crossing a threshold, an open support escalation, company news (layoffs, acquisition, breach), a champion going quiet. For each: how it's detected mechanically and where the record goes. 4. RE-ENTRY RULES — when and how a tripped record returns to automated flows, because 'human always forever' quietly becomes 'nobody ever'. 5. THE LINE IN WRITING — a three-sentence policy statement of what my company automates and what it never will, usable in team onboarding. Rules: no step gets 'human always' status for sentiment alone — over-humanizing is how teams stay too small to matter. Equally, anything touching an at-risk relationship or mid-escalation account is non-negotiable human. Tripwires must be mechanically detectable, not 'when it feels off'. Before mapping, interview me. Ask ONE AT A TIME, waiting for my answer each time: 1. Walk me through your sales process from first touch to renewal — the real one, not the slide version. 2. What's automated today, and where has automation embarrassed you or come close? 3. What's your average deal size and typical relationship length? 4. What signals about a prospect or customer does your stack already capture (replies, support tickets, engagement)? 5. How big is the team, and where are they most stretched? Then produce the map. Where I'm over-automating, say so; where I'm under-automating out of fear, say that too.
How to use it
- 1
Copy the prompt into Claude, ChatGPT, or any LLM.
- 2
Tell the embarrassment stories in question 2 fully — near-misses are the highest-signal input for tripwire design.
- 3
Implement the tripwires in your stack first; they protect you even before the full map is adopted.
- 4
Share the three lists and the written policy with the team, and re-score annually or when deal size shifts materially.
Best practices
Challenge every 'human always' entry with 'what specifically goes wrong if automated?' — surviving answers make the list legitimate.
Build the hybrid tier deliberately: AI-drafted, human-approved is the highest-leverage pattern for most stretched teams.
Wire tripwires to detection you actually have; a tripwire on signals nobody captures is decoration.
Watch the re-entry queue — records that trip out and never return are deals quietly dying in a safe-looking place.
Example: what this looks like in practice
A founder of a 10-person fractional-CFO firm automates aggressively — sequences, follow-ups, renewal reminders — and recently had an automated 'time to renew!' email land on a client three days into a billing dispute. The scored map puts prospecting and scheduling firmly in AUTOMATE, first-reply handling in HYBRID with AI drafts and human send, and anything touching active clients in HUMAN ALWAYS given $60K relationships that run for years. Tripwires wire the support inbox and reply sentiment into the CRM: any open dispute or negative reply pulls the record from every flow. The written policy — three sentences ending 'no client in an active issue ever hears from a robot' — goes into onboarding. Two months later a tripwire catches exactly the renewal-during-dispute scenario that started it all.
Best fit
This prompt is one gear in a bigger machine. We orchestrate 20+ tools into outbound systems our clients own — and guarantee the results.
Apply for a Pilot Spot → →Frequently asked questions
Anything where the right action depends on reading context and the cost of misreading is high: negotiation, escalations, at-risk relationships, sensitive replies, and high-stakes renewal conversations. The useful test isn't sentiment but scoring: judgment required, relationship stakes, and blast radius. Steps scoring high on any axis need a human in the loop.
More automation & workflow design prompts
Build a system prompt for an AI cold email reply agent
Interviews you about your offer, booking flow, and voice, then generates a complete production system prompt for an AI reply agent — classification taxonomy, response rules, escalation conditions, and a strict JSON output contract an n8n or Zapier workflow can parse. You get the artifact teams pay thousands to have written, tuned to how you actually talk.
Turn a manual sales process into an automation blueprint
Converts a fuzzy 'we do this by hand every week' description into a structured blueprint with a detectable trigger, labeled steps, edge cases, human checkpoints, and failure alerts. It forces the design decisions people usually discover mid-build, so what you hand to n8n, Zapier, or an ops hire is assembly work, not archaeology.
Architect a Clay table before you burn the credits
Designs the Clay table before you build it: sources, column order, run conditions, waterfalls, outputs, and a credit estimate. Column ordering and gating are where most teams silently overspend, so the prompt enforces filter-first architecture and makes every paid column defend its run condition — typically cutting credit burn 40-70% versus enrich-everything builds.