Prompt Templates — Role Starter Pack (PM, Analyst, Support)

Role Starter Pack: Ready-to-use Prompts for PMs, Analysts, and Support

This resource provides practical, testable prompt templates plus short guidance so frontline roles can begin using AI quickly and consistently. Use the templates as starting points — adapt voice, constraints, inputs, and expected outputs to match your product, data, and workflow.

How to use these templates

1. Copy the template and fill the placeholder fields (bracketed items like <PRODUCT>).
2. Run the prompt against your chosen assistant or model. Save the result and compare against the test cases below.
3. Refine prompts by tightening constraints, adding examples, or supplying relevant documents or data.

PM: Feature spec drafting template

When to use: Rapidly generate a first-pass feature spec, acceptance criteria, user stories, and a short implementation notes section to kick off planning.

Prompt template

You are a product manager writing a concise feature spec for <PRODUCT NAME / AREA>. Provide: 1) one-sentence summary, 2) user personas impacted, 3) goals & success metrics, 4) user stories (3–5), 5) acceptance criteria for each story, and 6) implementation notes & potential risks. Use short, actionable language. Assume the engineering team knows our platform: <TECH STACK OR CONSTRAINTS, e.g., "mobile-first, React, no backend changes">.

Example filled prompt: Replace placeholders with specifics such as product name and constraints.

Quick customization notes: Add a small sample UI mockup text or link if you want the assistant to extract labels. When privacy is a concern, redact customer data or use synthetic examples.

Test cases / validation:

• Spec includes at least 3 user stories and matching acceptance criteria.
• Goals are measurable (e.g., % lift, time saved) rather than vague.
• Implementation notes mention at least one risk or dependency.

Common pitfalls: Vague goals, missing constraints, or prompts that assume unavailable data. Remedy by including explicit metrics or attaching mock data.

Analyst: Data-exploration prompt + test cases

When to use: Generate a first-pass exploratory analysis plan, key questions, recommended visualizations, and testable checks for a dataset.

Prompt template

You are a data analyst exploring a dataset described as: <SHORT DATA DESCRIPTION> (e.g., "transactions table with columns: id, user_id, amount, ts, product_category"). Provide: 1) 6–8 analytical questions to answer, 2) recommended visualizations and aggregation levels, 3) potential data quality checks, and 4) a short list of KPIs to track. Suggest SQL snippets or pseudocode for key queries.

Example filled prompt: Include a sample schema or paste a small sample of rows when possible to get more precise SQL suggestions.

Quick customization notes: Specify the analysis timeframe and primary business objective (e.g., increase retention, reduce churn) to bias recommendations toward relevant questions.

Test cases / validation:

• Suggested questions map to the stated objective.
• At least two concrete SQL snippets or pseudocode examples are included.
• Data quality checks are actionable (e.g., null rate thresholds, referential integrity checks).

Common pitfalls: Prompts that assume access to enriched joins or fields that don't exist. Attach a schema or sample to avoid this.

Support: Triage & response draft with escalation prompts

When to use: Draft initial support responses, triage categorization, recommended next steps, and an escalation message for engineering or product teams.

Prompt template

You are a customer support specialist preparing a first-response to this ticket: <PASTE CUSTOMER MESSAGE OR DESCRIPTION>. Provide: 1) a short empathic opening (1–2 sentences), 2) an initial troubleshooting checklist tailored to <PRODUCT / OS / BROWSER>, 3) suggested next steps for the customer, 4) triage category and urgency, and 5) an escalation note to engineering including steps to reproduce and suggested logs or screenshots to collect.

Example filled prompt: Paste the real customer message but remove PII when required. For internal escalation drafts, include reproduction steps and expected vs. actual behavior.

Quick customization notes: Configure tone to match company voice (friendly, formal). Provide typical platform versions to improve triage accuracy.

Test cases / validation:

• Response includes an empathic opening and clear next steps.
• Triage category is defensible and maps to existing SLAs.
• Escalation draft contains reproduction steps or a checklist for the engineer.

Common pitfalls: Overly technical replies to non-technical customers or missing SLA classification. Include the customer's technical context and SLAs in the prompt.

Practical tips for prompt adoption and iteration

• Keep one canonical prompt per role and version it when you change it.
• Run A/B comparisons: store outputs for two prompt variants and compare against your test cases.
• Capture a short rubric for quality (e.g., accuracy, completeness, tone) to evaluate outputs consistently.
• Start conservative: prefer clearly constrained outputs (bullet lists, max length) to reduce hallucination risk.

Where this starter pack can grow

Consider collecting role-specific prompt performance (examples that worked, ones that failed) and turning this pack into a team-owned toolkit that includes tested prompt versions, example outputs, and usage guides.

Safety and privacy

Never include PII, secrets, or sensitive customer data in prompts unless you have a compliant, approved environment. Prefer synthetic or redacted examples when sharing prompts externally.

Use these templates as living starting points. Tailor them to your domain, run simple tests, and capture scoring so the team can iterate toward consistently useful results.