Earnings season is a focal point for fundamental and quantitative investors. Alternative data is often used to set expectations, validate or challenge consensus, and interpret results. Not every dataset is equally useful around earnings, and misuse can add noise rather than signal. This post outlines how alternative data is used around earnings and what tends to help in practice.
Where alternative data fits in the earnings process
Pre-earnings: setting expectations. Demand signals (search, e-commerce, app or web traffic) can inform revenue or unit expectations for consumer, retail, and digitally exposed names. The value depends on how well the data maps to the company's revenue drivers and on using a consistent methodology so period-over-period comparisons are meaningful. Sentiment and news volume can indicate whether narrative or event risk is building ahead of the print.
Post-earnings: interpreting results. After results are out, alternative data can help explain beats or misses (e.g. demand trends that aligned or diverged from reported numbers) and inform forward views. It can also highlight companies where the data and the stock reaction are out of sync, suggesting potential mispricing or follow-up work.
Screening and prioritization. Funds with broad coverage use alternative data to decide which names deserve deeper pre-earnings work or which earnings calls to prioritize. Sustained moves in demand or sentiment can flag names where the upcoming print may be consequential.
Data types that are commonly used
Search and intent data. Search volume and mix (Google, Amazon, YouTube, etc.) are widely used for demand-sensitive names. They are most useful when mapped to companies or key products and when used over multiple periods to establish a baseline. Platforms that offer normalized, company-mapped series (e.g. Paradox Intelligence) reduce the burden of building and maintaining these links.
News and sentiment. Sentiment and news volume are used to capture narrative shifts and event risk around earnings. As in other contexts, sentiment alone is rarely sufficient; it is more useful when combined with behavioral or demand data and when the methodology is consistent over time.
Web and app traffic. For companies where digital engagement is a revenue or engagement driver, traffic and usage trends can support or challenge growth assumptions. Often used alongside search for the same names.
Transaction and spending data. Where available, aggregated card or receipt data can support revenue or same-store expectations for consumer and retail. Coverage and latency vary; integration with earnings calendars (e.g. Paradox Intelligence earnings) helps align data with the reporting schedule.
Avoiding common pitfalls
Overfitting to one print. A single quarter's correlation between a data series and a beat or miss can be coincidence. Use multiple periods and multiple signals where possible to distinguish signal from noise.
Ignoring timing and lag. Data availability and reporting lag matter. Ensure the data you use is actually available and comparable for the period you are evaluating (e.g. same end date, same geography).
Relying on a single source. Combining two or more data types (e.g. search + sentiment, or traffic + search) usually improves robustness. Define a simple process (e.g. "demand up and sentiment stable" or "traffic and search both inflecting") and apply it consistently.
Skipping the fundamentals. Alternative data should complement valuation, quality, and catalyst analysis, not replace it. Use data to test or refine a view, not as the sole input for a trade.
For more on sentiment and multi-signal use, see News Sentiment and Alternative Data. For long-form research, see Research.
- Find Your Plan
This post is for institutional investors and research professionals. It is not investment advice.
