Classification basis
Trendier data is classified using text in each channel's PDP (Product Detail Page).
Product name, subtitle, description, marketing keywords, ingredient list, benefit claims, and other PDP text are all analyzed
Since this is keyword-based matching, anything appearing in PDP text gets included in classification
When unrelated products show up
When a PDP shows other product lines from the same brand, or when marketing copy includes keywords, unintended matches can occur. Example: searching for "niacinamide" might pull in another product line from the same brand because that line is shown in the recommendation section of the PDP — even though the actual product doesn't contain niacinamide.
How to get more accurate results
Tool | Accuracy | Best use |
Generator | Broad keyword-based matching | Big-picture category trends |
AI Chat | Context-based precise matching | Best for specific case-level analysis |
AI Chat narrows results based on natural language context, making it better for accuracy-critical validation.
Auto-filter improvements
We're continuously improving auto-filtering logic to reduce noise from PDP recommendation sections and same-brand line exposures.
When you need accuracy verification
For verification of specific cases, email [email protected] with:
Feature used (Generator / AI Chat)
Input keywords / conditions
URL of the misclassified product
Common follow-up questions
Does Trendier hallucinate?
Trendier only analyzes its own e-commerce data. Unlike general LLM hallucinations, mismatches are classification noise, not fabricated data.
How can I increase reliability of my reports?
Cross-validate across multiple channels and metrics. Focus on trend consistency and direction rather than single matches — that's where reliable insights live.
Related articles
How does Trendier identify trends?
The data count looks low
How do you filter out fake reviews?
How is Trendier different from ChatGPT-style AIs?