The no-estimation principle
Trendier's data model is built on one rule: every data point traces to direct observation.
Prices — pulled from live PDPs
Rankings — captured from each channel's own bestseller lists
Review counts and text — streamed from each product's review system
Sales volume — captured wherever the channel exposes it (e.g., Shopee, TikTok Shop US)
Consumer profiles — drawn from reviewer self-tagging on the channel
For data points a channel publishes differently, Trendier surfaces proxy signals — review velocity, ranking trajectory, promo frequency — that read market direction from real observed activity, with no synthetic error layer.
What each data type measures
Data type | Source | Measurement |
Prices | Live PDP capture | The price shown to shoppers at time of capture |
Rankings | Channel bestseller lists | The channel's rank position at time of capture |
Review counts | Channel review system | Cumulative count as displayed by the channel |
Sales volume | Channel UI (where exposed) | The channel's published unit (monthly, cumulative, etc.) |
Consumer profiles | Reviewer self-tagging | Skin type, concern, age, and other attributes reviewers select |
Sentiment / keyword extraction | NLP on review text | Trend-level patterns across the review corpus |
How the measurement model works at each layer
Capture and update cadence is structured. Trendier captures data daily; user-facing updates roll up on a weekly or less-frequent cadence. Yesterday's bestseller appears in the next published update.
Channel fidelity is one-to-one. Trendier captures what each channel publishes. The signal you see matches the signal the channel surfaces to its own users.
Sentiment analysis is built for trend reading. NLP on review text is optimized for directional patterns and thematic signal — what's rising, what's shifting, which attributes are gaining attention.
Coverage scope is the measured set. Top 100 is the top 100 — exact, not modeled.
When data looks wrong
Email [email protected] with the channel, data point in question, and comparison URL. See "How is Trendier's data collected?" for the full verification workflow.
Common follow-up questions
How is sales volume reported for channels that don't publish it directly?
See "What data is available for each channel?" for the proxy signal approach.
Where is sentiment / consumer perception analysis strongest?
At the trend layer — directional patterns, theme emergence, attribute movement across the review corpus. This is what it's built to surface, and where the signal is most reliable.
Is the data audited externally?
Trendier's collection pipeline is operated and verified internally. External third-party audits are being considered for future updates.
Related articles
Where does the data come from?
How is Trendier's data collected?
What data is available for each channel?
My results include unrelated products / how accurate is the analysis?