AI Sentiment Analysis FAQ
Table of Contents
Q: What is an Analyzer exactly?
A: An Analyzer is an AI Sentiment Analysis widget on a NewsWhip search. It reads every article in that search as it publishes and classifies it against your question and categories, building a trend you can explore over time.
Q: How does AI Sentiment Analysis differ from standard, off-the-shelf sentiment models?
A: Standard sentiment models use fixed, generic definitions to score the tone of the text as positive, negative or neutral. AI Sentiment Analysis reads the full article and judges its effect on how you're perceived, using your target, your categories and your rules. The two often agree, but they differ exactly when it matters most.
Q: Can I define sentiment our own way?
A: Yes. Classification rules let you state what positive, negative and neutral mean for your organisation.
Q: Can I customize the categories for different campaigns, issues or initiatives?
A: Yes. Each Analyzer has its own question, categories and instructions, so you can set one up for each search you track: a campaign, a live issue, or your ongoing brand coverage.
Q: Are there usage limits?
A: Your team shares an allowance of Analyzers, which you can check under More → AI Analysis. Paused Analyzers don't count towards it. For very high-volume searches (more than 1,000 articles a day), results are based on the top 1,000 articles per day by engagement.
Q: How long before I see results?
A: Usually within an hour. The one-month backfill can take up to 24 hours to complete.
Q: What happens to historical data and my existing searches when I add an Analyzer?
A: Your search and its existing widgets don't change. The new Analyzer backfills the last 30 days of coverage and then classifies every new article as it publishes. Coverage from before that 30-day window isn't analysed, so it's worth starting Analyzers on your priority searches early.
Q: Does it work with articles in other languages?
A: Yes. Articles in all languages are analysed. Categories, descriptions and driver summaries are shown in English.
Q: Can the rest of my team see the results?
A: Yes. We recommend one person sets up Analyzers on a shared dashboard and shares it with the team. That keeps results consistent and makes the most of your allowance.
Q: Why are some articles Unclassified?
A: The backfill may still be running, the articles may be older than the one-month window, they may have started matching after a search edit, or the search may be high-volume and the articles fell outside the daily sample. Hover over the Unclassified series to see the reason.
Q: Sentiment looks wrong on articles that barely mention us. What should I do?
A: Usually the target isn't specific enough. Sharpen the target, then add a classification rule describing what counts for you.
Q: I duplicated a dashboard and the AI Analysis widgets are missing. Why?
A: This is expected. Analyzers aren't copied with a dashboard. Share the original dashboard instead.
Q: Will an Analyzer find issues we don't know about?
A: No. It measures what you define. To surface emerging stories you aren't yet tracking, use Trellis Monitoring Agents.
Q: How does NewsWhip ensure accuracy?
A: Every classification is chosen from the categories you defined, so the Analyzer can't invent an answer. Every point on a chart links to the articles behind it, and we run quality checks on classifications every day. We recommend starting with a search your team knows well and comparing the results with how you'd have read the coverage yourselves.
Q: Is this the same as sentiment in Sprout Social Listening?
A: No. Listening measures sentiment across social conversation. AI Sentiment Analysis in NewsWhip reads news and web coverage and measures how it affects your reputation. It doesn't analyse social media posts.
Was this article helpful?
