Glossary

Data freshness

How current a dataset is relative to real time, measured as the lag between an event happening and it being queryable.

Also called: data recency, freshness

Data freshness measures the gap between when something happened in the real world and when that fact is available to query in a dataset or dashboard. A dashboard showing yesterday's sales as of this morning has a freshness of roughly a day; a dashboard reflecting an order placed thirty seconds ago has a freshness of thirty seconds.

Freshness is a direct consequence of pipeline design: a nightly batch processing job produces freshness measured in hours, while a stream processing pipeline can produce freshness measured in seconds. It is distinct from data quality or accuracy, a dataset can be perfectly fresh and still wrong, and it is distinct from uptime, since a pipeline can be running on schedule but still produce data that is stale relative to what a specific use case needs.

Freshness matters because different decisions tolerate different amounts of lag: a monthly board report doesn't need minute-level freshness, but a fraud model or live inventory count does. Teams formalize freshness requirements as a Data SLA and monitor actual lag against it, using workflow orchestration tools to alert when a pipeline falls behind. A common mistake is chasing near-real-time freshness for data nobody acts on quickly, which adds cost and complexity without changing any decision.

Last reviewed September 22, 2026

In the index now

Related terms

Data SLA (service-level agreement) A formal commitment about a dataset's freshness, availability or quality, with defined targets and consequences. Batch processing Processing data in large, scheduled groups rather than continuously as each record arrives. Stream processing Processing data continuously as individual events arrive, rather than waiting to accumulate a batch. Workflow orchestration Scheduling, sequencing and monitoring interdependent data tasks so each step runs only after the ones it depends on succeed.

Related tools

GreyNoise Classifies internet-wide scanning traffic as benign, malicious, or unknown so SOC teams can filter background noise out of alerts.

Related guides

How to choose a data observability tool Pick a data observability tool by how it generates coverage, where it needs to run relative to your data, and whether you're already living inside dbt. How to choose a threat intelligence platform Threat intelligence splits into internet scanning, noise filtering and closed-community collection — pick by which problem you actually have.