📋 Available on: Growth, Enterprise, Agency 2026, Agency 2026 Unlimited, and Custom plans (Starter plans do not support Sentiment Analysis)
Sentiment Analysis classifies the emotional tone of user-generated content captured by Archive. Each post is reviewed by a brand-aware model and assigned a sentiment label that appears in content views, filters, and campaign reports.
How Sentiment Analysis Works
Each piece of UGC is classified into one of five categories:
Positive — Content expressing favorable opinions, enthusiasm, or satisfaction
Neutral — Informational or factual content without evident emotion
Mixed — Content containing both positive and negative signals
Negative — Content expressing criticism, frustration, or dissatisfaction
Not Applicable — Content where emotional sentiment is not relevant or cannot be detected (e.g. an empty caption with no transcript and a poster frame with no textual or emotional cues, such as silent videos or music-only audio)
What the model sees: your brand context, the content's transcript, its caption, and a thumbnail frame.
Classification is brand-aware. The model considers your brand context (a free-form brand description configured per workspace) when scoring — for example, "this workout was brutal" is classified as positive for a fitness brand, rather than negative based on the word alone.
💡 The model scores sentiment toward the content overall — it does not distinguish whether that sentiment is directed at your brand or at an influencer featured in the content.
Where to Find Sentiment Analysis
Individual Content View
When viewing a single post in Social Listening, the sentiment label is shown alongside the content.
Content Filtering
On the Social Listening content page, you can filter by one or more sentiment categories to surface specific subsets of content — for example, only Positive posts for ad selection, or only Negative posts that require review.
Reporting Sentiment Analysis (Reports Page and Campaigns Reports Page)
Sentiment Analysis Rerpots Reports include a sentiment section with two views: a time-series chart showing the daily breakdown of posts by category (positive, neutral, mixed, negative, and not applicable) and a set of totals tiles (total, positive, neutral, mixed, negative, and not applicable) showing aggregated counts.
Clicking any tile opens the underlying content filtered to that sentiment category.
Use Cases
Ad content selection — filter for Positive sentiment to identify UGC suitable for paid campaigns.
Issue detection — track Negative sentiment spikes to surface emerging product or messaging issues quickly.
Brand perception tracking — monitor sentiment distribution over time to measure how launches and campaigns affect audience response.
Creator performance comparison — compare sentiment distribution across creators to identify which ones consistently drive favorable responses.
Limitations
Historical sentiment processing for content captured before the feature was enabled is available on request, not automatic.
Common Questions
Do I need to configure anything to use Sentiment Analysis?
No manual setup on your end. Once the feature is enabled for your workspace by Archive, it runs automatically on all newly captured content.
Why is some of my older content not classified?
Sentiment is applied to content captured after the feature was enabled for your workspace. Historical content can be processed retroactively on request — contact Archive Support.
How accurate is the classification?
The model is brand-aware and considers your brand context when scoring. Accuracy is highest on content with clear text or transcripts — sentiment is only generated after transcription completes, so silent videos or photos with no caption and no transcribable audio are more likely to land in Not Applicable.