Institutional investors routinely use both search data and sentiment data. The two are often lumped into "alternative data," but they answer different questions. Search data reflects what people are doing: what they look for, where they click, what they intend to buy. Sentiment data reflects what is being said: tone of news, social posts, and filings. That difference drives when each signal leads revenue and price, and why the best outcomes usually come from using both.
This post summarizes how search and sentiment differ, when each tends to lead, and how to combine them without double-counting or adding noise.
What each one actually measures
Search data is behavioral. It captures demand and intent: Google Search volume for a brand or product, Amazon search trends, YouTube search for a topic, Google Shopping interest. When search volume for a company or product rises, it usually means more people are actively considering or seeking that brand. The signal is "people are doing this." It tends to lead reported revenue because the sequence is awareness and search, then purchase, then revenue, then earnings. Academic and industry work consistently finds that search volume leads revenue by roughly 4 to 12 weeks for consumer-facing names, depending on sector and data type.
Sentiment data is narrative. It captures how a company, sector, or theme is being written and talked about: news sentiment scores, earnings-call tone, social sentiment, volume of positive vs negative coverage. The signal is "this is what is being said." Sentiment can move ahead of price when narrative shifts before the market fully reprices, and it can lag when it simply reflects news that is already in the tape. Research shows sentiment is more predictive in some regimes (e.g. short-term, event-driven) and less so when used alone over long horizons. It is also highly methodology-dependent: different vendors and models produce different series.
So: search answers "Is demand moving?" Sentiment answers "How is the story changing?" Neither replaces the other.
When search tends to lead
Search data is strongest when:
- You care about demand and revenue. For consumer, retail, and digitally exposed names, search volume for brands and products is a leading indicator of revenue. The lead time is typically weeks to a full quarter. See Leading Indicators for Revenue for detail.
- You need a stable, backtestable signal. Search series are usually consistent over time (same geography, same methodology), so they are easier to backtest and to combine with fundamentals.
- You want to avoid reflexivity. Search reflects behavior, not just opinion. It is harder for the market to "talk up" search volume in the same way it can move on sentiment.
Search is weaker when the company is not search-sensitive (e.g. some B2B or infrastructure names) or when the relevant behavior does not show up in search (e.g. in-app or offline behavior). In those cases, other behavioral data (traffic, app usage, transactions) or sentiment may add more.
Stay up to date on our best ideas
When sentiment tends to lead
Sentiment data is most useful when:
- You care about narrative and event risk. Sentiment can flag when a name or theme is getting unusually positive or negative attention, so you can dig deeper or adjust exposure. It is often used for monitoring and risk, not only for alpha.
- You trade around events. Around earnings, M&A, or regulatory news, sentiment can shift before the full price reaction. In fast-moving, story-driven markets, sentiment can lead over short horizons.
- You combine it with behavior. Sentiment alone is often a weak standalone signal (see News Sentiment and Alternative Data). When paired with search or traffic, it helps distinguish a one-off news spike from a sustained shift in demand or narrative.
Sentiment is weaker when it is used in isolation, when the methodology is opaque or changes over time, or when it is heavily crowded. Many practitioners use it as one input among several, not as the primary driver of a trade.
Why combining both usually works best
Prediction markets and surveys measure beliefs about outcomes. Search and other behavioral data measure actual behavior. Sentiment sits in between: it reflects how the world is talking about an outcome, which can influence behavior and price but is not the same as demand. So:
- Search (and traffic, app data) answers: "Is interest and demand rising or falling?"
- Sentiment answers: "Is the narrative shifting, and is it consistent with or at odds with that behavior?"
Using both lets you check whether narrative and behavior align. For example: search up and sentiment stable or improving supports a demand story; search up and sentiment turning negative may signal an emerging risk or a disconnect the market has not yet priced. Conversely, sentiment spiking with flat or falling search may indicate a short-lived narrative rather than a durable demand shift.
Platforms that offer search, sentiment, and other sources in one place (e.g. Paradox Intelligence, with search, news sentiment, and news volume mapped to the same tickers) reduce the work of building and maintaining a combined view.
Practical takeaway
- Search data leads revenue for demand-sensitive names; it is behavioral and usually easier to backtest. Use it when you care about demand and revenue and when the company is search-sensitive.
- Sentiment data captures narrative and event risk; it can lead price over short horizons and is useful for monitoring and risk. Use it when you care about story and events, and pair it with behavioral data so you do not trade on noise.
- Combining both gives you behavior and narrative in one process. Define a simple rule (e.g. "demand up and sentiment stable" or "sentiment and search both inflecting") and apply it consistently. Avoid using sentiment alone as the sole trigger for a trade.
For more on multi-signal use, see How to Use Alternative Data Around Earnings and Best Alternative Data Platforms 2026. For long-form research, see Research.
This post is for institutional investors and research professionals. It is not investment advice.
