Use approved tools, verified business information, and de-identified data where possible. The prompts assist work; they do not grant permissions or replace technical access controls. Review customer-facing claims and consequential actions. Replace all bracketed inputs before use.
1. Product-page audit
Act as a customer-experience reviewer. Evaluate the supplied product page against this buying task: [task]. Use only the page content and verified product facts provided. Identify unanswered purchase questions, conflicting information, and functional issues visible in the evidence. Separate confirmed observations from hypotheses. Return a prioritized table with issue, evidence, customer consequence, proposed fix, confidence, and how to verify. Do not invent analytics or claim a guaranteed conversion lift.
2. Factual product description
Using only [verified specifications], draft a product description for [customer/use]. Include a short overview, factual benefits, specifications, included items, limitations, and FAQs. Mark missing facts separately. Do not invent safety, sustainability, health, performance, warranty, review, or delivery claims. Preserve variant distinctions. Return a fact-check table mapping claims to input facts.
3. Customer-feedback analysis
Analyze [de-identified records] for [product]. Group buying motivations, unanswered questions, fit problems, usage issues, and defects. Give source IDs and distinct-record counts for each theme. Flag duplicates. Separate observations from explanations. Do not invent quotes or infer sensitive traits. Recommend what to investigate before changing the page or product.
4. Welcome-flow draft
Draft a three-message welcome sequence for [store/customer]. The promised resource is [resource]. Use the verified product and offer facts below. For each message provide purpose, subject, preview, body, CTA, needed data, and exit conditions. Do not invent deadlines or discounts. Explain what changes after a purchase and where marketing eligibility must be checked.
5. Post-purchase education
Map helpful messages for [product] from purchase through successful use. Inputs: [delivery events available], [verified instructions], [common customer questions]. Distinguish essential order updates from marketing. Use event-based timing where supported; label assumptions. Include suppression for unresolved delivery issues, refunds, and completed next purchases. Do not claim delivery without reliable data.
6. Support draft
Draft a response to [de-identified case] using only [approved policy and product knowledge]. Cite the source for policy statements. Do not approve refunds, change orders, invent delivery dates, or make safety claims. If information is missing or conflicting, ask one focused question or recommend escalation. Provide the draft and a reviewer checklist.
7. Weekly performance review
Review [exports] for [periods/timezone/currency] using [metric definitions]. Identify missing values, duplicates, incompatible scopes, and zero denominators first. Show calculations for changes. Separate facts from hypotheses. Return three investigations and at most three actions with evidence and guardrails. Do not invent benchmarks or causal conclusions. Label forecasts and estimates.
8. Product opportunity research
Evaluate [product idea] for [customer/channel/market] using only [research evidence]. Separate demand, differentiation, contribution, operations, supplier, and marketing fit. List unknowns and any failed requirements. Do not invent market size, search volume, supplier certifications, or success probabilities. Propose the smallest honest test of the most important uncertainty.
9. Product video brief
Write a demonstration script answering [one buying question] for [product]. Use [verified facts]. Include hook, shot list, spoken copy, captions, and next action. State which variant appears. Avoid unsupported claims and misleading edits. List assets and permissions needed. Do not present generated scenes as real customer evidence.
10. Catalog quality review
Compare [catalog records] with [required-field schema]. Report missing fields, inconsistent units, duplicate SKUs, variant conflicts, and unsupported claims. Do not change records. Return proposed corrections only where the supplied authoritative data supports them; otherwise flag a question. Include row IDs and a reversible review plan.
11. Automation design
Map [manual process] into trigger, eligibility, inputs, action, confirmation, duplicate prevention, failure handling, logging, and owner. Prefer deterministic rules where sufficient. Identify where AI classification would need validation. Keep financial commitments and customer-impacting changes behind appropriate approvals. Propose a limited test and rollback plan; do not execute actions.
12. Promotion review
Review [offer] using [revenue basis and cost inputs]. Calculate baseline and promoted contribution, required volume to match contribution, and risks from discount stacking, returns, and acquisition costs. Do not double count refunds or overhead. Flag missing data. Review whether the landing page and terms match the offer. Do not invent urgency or claim incremental impact without evidence.
13. Search-content brief
Using [customer research and verified catalog], propose a buying guide for [question]. Explain the intent, useful comparisons, facts needed, internal links, and customer next step. Avoid invented keyword volumes and ranking promises. Identify what requires original observation or demonstration. Keep the guide distinct from existing supplied pages.
14. Compare platform requirements
Compare [candidate platforms] against [must-have workflows, budget, team skills]. Use only current official documentation supplied or retrieved with source dates. Separate native, extension-dependent, and custom work. Include ongoing ownership and total cost categories. Do not invent current prices or assume every plan includes a feature.
15. Prioritize next actions
Given [store observations, metric definitions, resource limits], recommend at most three next actions. Prioritize confirmed purchase or fulfillment failures, then unresolved economics or measurement, then evidence-backed growth opportunities. Explain the evidence, uncertainty, expected customer benefit, effort, and verification for each. Do not infer a universal bad conversion rate from a small sample.
Want help applying this to your store?
Take the free store assessment for a suggested next step, or book a 15-minute call to talk it through.