사용 신호로 사람 개입 시점을 정하는 PLG 세일즈 어시스트 플레이북

Role: You are a product-led growth operator who has built sales-assist motions at three self-serve SaaS companies, moving users from free and team plans to annual contracts, and who has seen what happens when sales interrupts users who were about to upgrade on their own.
Context: [companyName] sells [productService] with a [pricingModel] pricing model. Users sign up and get value without talking to us. We have [salesCapacity] of human selling time per week, so every touch must be earned by a signal. Here are the usage and account signals we can see today: [usageSignals]. The playbook will be used by the sales-assist team and by the growth team that owns in-product messaging. It decides when a human steps in, when the product keeps working on its own, and how a lead is handed over without the user repeating themselves.
Task: Produce a sales-assist playbook: the signals that trigger a human touch with thresholds, three message templates, the handoff rules between product, sales and success, and an explicit list of what not to interrupt.
Inputs:
- [productService]: what the product does and the moment users first see value.
- [usageSignals]: events, account attributes and plan data we track; if a signal you would need is missing, put it in the final section on signals to add rather than assuming we have it.
- [pricingModel]: how we charge (per seat, usage, flat tiers); shape the expansion triggers around it.
- [salesCapacity]: hours or headcount per week; size the thresholds so the volume of touches fits inside it, and show the math.
Constraints:
- Score signals on fit (company size, domain, role) and intent (depth of use, team spread, limits hit, billing page visits); explain the weighting in plain terms.
- Each trigger has a threshold, an owner, a response-time target and the one question the human should ask first.
- Messages under 90 words, written as a person who looked at the account and names the specific signal; no fake personalization, no pretending an automated message was hand-typed, a one-line opt-out in each.
- The do-not-interrupt list must include users in their first session, trial users with rising usage, anyone who declined in the last 30 days, and product-specific cases you infer.
- Handoff: define the fields passed to the CRM, who owns the account after a reply, and what success does when a deal closes.
- Define five metrics for the motion, including assisted conversion rate against the self-serve baseline and touches per rep per week.
- Do not invent benchmark numbers; where a threshold is a guess, label it as a starting threshold to review after 4 weeks.
- Before answering, check that every trigger fits inside [salesCapacity] and that no template could reach a user on the do-not-interrupt list.
Output format:
Numbered sections with headings:
1. Signal scoring model: a table with columns: Signal, Type (fit or intent), Weight, Why it matters.
2. Triggers: a table with columns: Trigger, Threshold, Owner, Response time, First question.
3. Three message templates: product-qualified lead outreach, limit-hit expansion, multi-team consolidation, each with a subject line and body.
4. Do-not-interrupt list: a bullet list with the reason for each item.
5. Handoff rules: product to sales, sales to success, in bullets with the CRM fields.
6. Metrics: a table with columns: Metric, Definition, Starting target, Review cadence.
7. Signals to add and open questions for me, at most 5 bullets.