Intent Data Freshness Has Three Clocks, Not One

Fresh intent data requires more than a recent delivery timestamp. You need to know when the behavior occurred, when the record became available, and when the contact or identity information was verified. These three clocks answer different questions.

A newly delivered record can describe old behavior. A recently verified contact can belong to a buyer whose project ended months ago. A current signal can still be unusable if it arrives after the decision window closes.

Separate event time from delivery time

Event time describes when the observed activity happened. Delivery time describes when your team could use it. The difference is latency. For urgent services, a short delay may matter much more than it does for long-term capital planning.

Ask what timestamps the provider supplies and what each represents. A batch creation date should not be mistaken for the original behavior date. If the source provides only a broad observation period, preserve that uncertainty.

Use the time that information became available when evaluating a historical decision. Otherwise, an analysis may credit the team with information it could not actually have used at the time.

Contact freshness is a different problem

Contact verification concerns whether a field remains usable or accurate under the provider’s definition. It does not establish that the person still has the same buying need, role, authority, or preferred contact method.

Similarly, an identity match does not guarantee purchase readiness. Keep contact quality, match confidence, and intent relevance separate in the review. Combining them into one freshness claim makes it harder to diagnose failure.

Ask what verification covers. A valid address, deliverable email, or persistent identifier may each have a different process. Avoid assuming that one kind of check proves every field is current.

Match the signal lifetime to the purchase

An emergency replacement and a planned enterprise implementation have different clocks. For the first, old activity may quickly lose relevance. For the second, earlier research may still be useful context if the project remains open.

Choose a recency policy based on the action. A signal may be too old for an urgent sales queue but still useful for content planning or a later research discussion. Retention for analysis and eligibility for activation are separate decisions.

Do not select an arbitrary decay rule and present it as scientifically established. Treat the rule as a starting hypothesis, then evaluate it with outcomes and the business’s normal buying cycle.

A hypothetical service example

A local contractor receives a relevant signal after the requested service date has passed. The contact information may be correct, but the immediate opportunity may no longer exist. The useful response is to inspect latency and scheduling context rather than blame the entire audience category.

By contrast, a manufacturer researching a new production line may continue evaluating suppliers for months. A strict short recency filter could remove suitable accounts before they are ready for a technical conversation.

These examples show why customer-specific learning matters. The same age can mean stale in one market and still informative in another.

Evaluate freshness without selection bias

Compare outcomes across signal-age groups only after accounting for different offers, channels, and treatment where possible. Recent records may receive faster follow-up simply because they appear at the top of the queue. That can contribute to their better outcomes.

Record response timing and outcome maturity. A newer cohort may look weaker because its opportunities remain open, while an older cohort has had time to close. Use comparable follow-up windows or an appropriate method for delayed outcomes.

Keep the original event timestamp when records are refreshed. Overwriting it with the latest processing date can make old behavior look permanently new.

What to ask Cortex

Ask which clocks are available in the configured service, how source latency is represented, and how recency interacts with your customer-specific score. Discuss the buying cycle before deciding how long a signal should remain actionable.

Cortex’s feedback process can help distinguish temporary interest from patterns that remain relevant over a longer evaluation. The learning still depends on reliable timestamps and clear outcomes.

Book a Cortex consultation with your typical time from first inquiry to decision and examples of opportunities that arrived too early or too late. Those cases can help define a freshness policy that reflects your market.