Dormant accounts are not free inventory. Every autonomous re-entry needs a receipt that proves why contact is legitimate now, who owns the risk, and what will stop the machine if the account pushes back.
Consent receipts are operational records that prove an autonomous GTM action has a valid reason to contact a dormant account again.
Old accounts carry human context that systems often lose: promises made, objections raised, timing agreed, and boundaries set.
Teams restart conversations through governed evidence instead of letting agents mine silence as if it were consent.
Autonomous GTM workers are excellent at finding neglected revenue surfaces. They can detect stale opportunities, old product-qualified leads, inactive expansion targets, and customers who stopped engaging after a pricing conversation.
The danger is that machine workers often see absence as opportunity. No open task. No recent activity. No owner response. The system concludes that reactivation is available.
But dormant does not mean permissionless. A paused account may be waiting for budget, cooling off after a support failure, honoring a legal review, or intentionally protected by an account owner. If an agent re-enters without that context, the company spends trust to create a meeting.
What changed since the last touch: new intent, product usage, funding, hiring, renewal timing, content engagement, or explicit inbound interest.
Whether the account has opt-outs, sensitive stakeholders, active support pain, pricing objections, legal constraints, or account-owner notes.
Who is accountable for the re-entry: named rep, lifecycle owner, RevOps policy, or a machine-worker queue with a human escalation path.
Which channel is allowed, which copy constraints apply, how many touches can happen, and when the worker must stop.
What happened after contact: reply, unsubscribe, complaint, opportunity movement, silence, handoff, or a rule update for future accounts.
The worst version of agent-led growth is not a bad email. It is a system that repeatedly discovers the same forgotten accounts and reopens them with no memory of why the last motion stopped.
A human seller usually remembers the texture: the CFO was annoyed, the champion left, the buyer asked for Q3, the support ticket made outreach inappropriate, or the discount conversation needs VP approval. Agents need that texture translated into receipts.
Without receipts, the control plane cannot distinguish a high-fit dormant account from an account the company should leave alone.
Define why the account is dormant: no response, closed-lost, delayed timing, support sensitivity, contract pause, pricing objection, or owner hold.
Name the evidence that allows re-entry now. A generic stale-date rule is not enough for sensitive accounts.
Specify the permitted channel, message type, touch limit, personalization source, and human-approval threshold.
Define what halts the worker: unsubscribe, negative reply, support escalation, owner objection, missing evidence, or no response after the cap.
Store the outcome so future agents know whether this pattern created revenue, annoyance, cleanup, or policy change.
This is not bureaucracy for its own sake. A receipt lets the worker move quickly when the evidence is clean and pause cheaply when the history is risky.
For Heads of Growth, the practical question is not whether agents should revive dormant pipeline. They should. The question is whether the system can explain why this account, why now, why this channel, and who owns the consequences.
If those answers are missing, reactivation is not leverage. It is trust arbitrage.
Accounts reply with context the worker should have known: previous promises, unresolved tickets, opt-outs, or timing requests.
Reps, CSMs, or executives discover agent outreach only after the customer reacts.
The system records sends and replies, but not why re-entry was allowed or what future agents should avoid.
Separate quiet, closed-lost, delayed, sensitive, support-risk, and owner-held accounts before agents touch them.
Do not allow re-entry unless the worker can attach a current signal stronger than elapsed time.
Name the human, queue, or policy owner accountable for the outreach and for stopping it.
Store the result in a way the next worker can use, not just a vanity activity count.
Pick one dormant-account workflow and require a consent receipt before any autonomous touch goes out.
The blunt rule: a dormant account is a relationship with missing context, not a blank slot for an agent to fill.