INCIOGet INCIO Reviews1. Prepare a coherent dataset
Do not mix unrelated products, marketplaces, or variations without labeling them. Keep rating, date, verified-purchase status, helpful count, variation, review ID, and body when those fields matter to the question. Inspect completion metadata so you know whether the export reached the requested scope or stopped early.
Keep review IDs and a product label. If several ASINs are present, tell the model never to merge them unless the output is grouped by product. Remove columns that do not serve the question, but retain collection status, variation, and rating when they affect interpretation.
2. Minimize data before uploading
INCIO creates the export locally, but a third-party AI service receives whatever you upload. Remove reviewer display names, URLs, or other fields you do not need. Check the provider's privacy, retention, training, and enterprise-data terms before using customer-feedback data.
| Field | Keep when | Remove when |
|---|---|---|
| author_name | Rarely needed for research. | Default choice for data minimization. |
| review_url | You need direct source navigation. | Review ID is sufficient for the workflow. |
| variation | Models, sizes, colors, or bundles can differ. | Only when the dataset is confirmed single-variation. |
| verified_purchase | The research question explicitly compares the label. | It has no role in the planned analysis. |
| completion metadata | Almost always; it defines the scope. | Only after its meaning is recorded elsewhere. |
3. Ask for auditable output
A strong request defines the task, evidence rules, and output format:
Ask the model to cite review IDs, count rows, distinguish direct statements from interpretation, identify contradictions, and state what the dataset cannot establish. Avoid vague prompts such as "summarize the reviews."
4. Verify important findings
- Open every cited review ID and confirm the claimed meaning.
- Recount high-priority categories in the source file.
- Check whether one variation or date range drives the pattern.
- Look for positive reviews that contradict the proposed conclusion.
- Repeat the analysis with a narrower prompt when categories overlap.
5. Use multiple passes instead of one giant prompt
- Scope pass: ask the model to restate products, variations, row count, filters, and completeness without analyzing themes.
- Codebook pass: propose a small taxonomy with inclusion and exclusion rules.
- Coding pass: assign review IDs to one or more categories and mark uncertain rows.
- Counting pass: count coded rows using the observed dataset as the denominator.
- Contradiction pass: find positive or neutral evidence that challenges each priority theme.
- Brief pass: draft findings with citations, limits, and testable next actions.
Separating the passes makes it easier to locate an error. If a count is wrong, you can inspect the coding table instead of regenerating an opaque end-to-end summary.
Reusable verification prompt
CSV or JSON for AI analysis?
CSV is usually easier to inspect and annotate manually. JSON preserves arrays, null values, warning objects, and separate job metadata. Choose the format the tool can reliably read, then verify that it recognized the row count and field names before asking for analysis.
6. What AI review analysis cannot prove
A model summary does not prove product safety, legal compliance, market demand, financial viability, or a representative defect rate. It may misclassify sarcasm, merge different failure modes, overlook context, or invent support. Use it to accelerate coding and hypothesis generation, not to replace source review, testing, or professional advice.
Warning signs in a model response
- It reports percentages without a visible denominator.
- It quotes language that cannot be found in the source file.
- It merges products or variations while claiming one shared conclusion.
- It calls an observed review share a defect rate or customer rate.
- It recommends a safety, legal, or engineering conclusion without qualified evidence.
- It does not acknowledge partial collection or warnings in the export.