When to use this prompt
Quarterly. Or before a major positioning or messaging refresh. Customer reviews on G2, Capterra, Trustpilot, and Google contain the language buyers actually use to describe your product. Marketing and product teams that read those reviews directly outperform teams that rely on internal interpretation.
The output is meant to be the input to two downstream artifacts: a sales enablement update and a content roadmap.
The prompt
<role>Voice-of-customer analyst synthesizing reviews into a brief that product, marketing, and sales teams can act on.</role>
<task>Analyze the reviews below. Extract the major themes, sentiment-weighted strengths and weaknesses, persona signals, and 5 specific content opportunities the reviews surface.</task>
<inputs>
<product_or_service>[NAME]</product_or_service>
<category>[CATEGORY]</category>
<reviews>
[PASTE ALL REVIEWS, EACH PREFIXED WITH:
- Source platform (G2, Capterra, Trustpilot, etc.)
- Reviewer role if available (e.g., "VP of Product")
- Star rating
- Date
- Full review text]
</reviews>
</inputs>
<instructions>
1. Identify the 5 to 8 most-mentioned themes across the reviews. A theme must appear in at least 3 reviews to count. Order by frequency.
2. For each theme, classify sentiment: PRAISED, MIXED, CRITICIZED. Note the count of reviews mentioning it.
3. Extract verbatim phrases (5 to 10 total) that capture how customers describe the product. These are the words for marketing and sales to mirror.
4. Identify persona signals: which buyer roles or company types appear most often, and what those personas tend to praise or criticize.
5. Surface 5 specific content opportunities the reviews reveal. A content opportunity is a question or objection that appears in multiple reviews and is not currently addressed in obvious public content. Format each as: "Buyer signal → Content recommendation."
6. Identify the single biggest gap between how customers describe the product and how the marketing copy describes it (if marketing copy is provided in the inputs; otherwise, base on review themes alone).
7. Do not invent themes that are not in the reviews. If a theme is implied but not explicitly stated, do not list it. Stick to what is verifiably in the data.
</instructions>
<output_format>
## Top themes
| # | Theme | Sentiment | # of reviews mentioning |
|---|-------|-----------|-------------------------|
## Verbatim phrases (buyer language to mirror)
- "[exact phrase]" — [persona/role if known]
- ...
## Persona signals
- [Persona] (X reviews): [what they praise / what they criticize]
- ...
## Top 5 content opportunities
1. **[Buyer signal]** → [Content recommendation]
2. ...
## Biggest copy-vs-customer gap
[1 paragraph or "Cannot evaluate; marketing copy not provided in inputs."]
</output_format>How it works
The “must appear in at least 3 reviews” constraint prevents a single passionate review from being mistaken for a theme. Themes are emergent patterns, not anecdotes. This is the most common failure mode of LLM-driven review synthesis: a model surfaces a vivid one-off complaint as if it were category-defining.
The verbatim-phrase requirement is what makes this brief operationally useful. Marketing teams that mirror customer language outperform teams that translate their internal vocabulary. Pulling exact phrases from reviews and using them in service-page headlines, ad copy, and sales decks is one of the highest-leverage moves in B2B positioning.
The “content opportunities” output is the bridge to the editorial roadmap. Reviews are full of unanswered buyer questions, and questions that show up across multiple reviews are exactly the questions your content should answer. The “Buyer signal → Content recommendation” format makes each item directly actionable.
The “do not invent themes” guardrail is essential because frontier models will produce a clean five-theme summary even when the reviews actually have 12 disparate complaints with no clear pattern. Forcing the model to ground every theme in real review counts surfaces that messiness and lets you decide how to handle it.
Example output
Top themes
# Theme Sentiment # of reviews 1 Implementation speed and ease PRAISED 23 2 Pricing transparency CRITICIZED 14 3 Reporting flexibility MIXED 11 Verbatim phrases
- “Up and running in a single afternoon” — VP Product
- “Pricing felt like a black box until we got on the call” — Marketing Director
Top 5 content opportunities
- “Pricing felt opaque” appears in 14 reviews → Publish a transparent pricing-tier breakdown page with example monthly costs by company size.
- “Reporting is powerful but the learning curve is steep” appears in 11 reviews → Create a reporting cookbook with 10 common reports, step-by-step.
Biggest copy-vs-customer gap
Marketing copy emphasizes “advanced analytics,” but customers consistently praise “speed of setup” and criticize “pricing complexity.” Speed-to-value is the differentiator buyers experience; the public messaging undersells it.
Variations
- Single-platform mode: Run on G2 reviews only, then again on Trustpilot only. Different platforms attract different buyer personas; comparing the syntheses surfaces persona-specific patterns.
- Competitor-overlay mode: Run the same prompt on a competitor’s reviews. The themes that appear in competitor praise but not in yours are gaps; themes that appear in competitor criticism but not in yours are quiet wins.
- Sales-enablement mode: Add an instruction to extract specific objections and the rebuttals reviewers themselves give. Useful for refreshing a battle card.