The short version: we tested three windows across a book of eight accounts. A seven day window returned zero results. Thirty days returned two. Ninety days returned twenty two — eleven times the monthly count, for the same one credit. Leadership changes surface with a lag, so a weekly alert finds nothing and a quarterly review finds everything the weekly one missed.
This page is the test, the reason behind it, and the prompt to run on the cadence the data supports.
Every figure below was measured live via the Lusha connector on August 15, 2026 across a set of eight enterprise software accounts. Contacts are masked to initials and companies replaced by category descriptors.
The test
One account set, one seniority filter, three time windows. Nothing else changed between runs.
| Window | Senior arrivals found | Credits |
| Last 7 days | 0 | 0 |
| Last 30 days | 2 | 1 |
| Last 90 days | 22 | 1 |
The seven day run cost nothing, because a search returning no results is not charged. That is the cheapest possible way to learn that a weekly cadence has nothing to tell you.
Why the short window is empty
People did change jobs at those accounts in the last seven days. The changes had not surfaced yet.
Any dataset that observes employment rather than being notified of it directly carries a detection lag. Someone starts a new role, and it takes time before that shows up anywhere a data provider can see it. We measured the same effect in our data decay research: across a cohort of 140,964 US Sales leaders, a three month window returned roughly half the monthly change rate that twelve and twenty four month windows both returned. The rate does not slow down in recent months. The observation does.
The practical consequence for account monitoring is direct. A weekly leadership change alert is mostly measuring your data provider’s detection speed, not your accounts’ actual turnover. It will look quiet, and you will conclude nothing is happening, and you will be wrong.
What cadence to run instead
Quarterly, as the default. The ninety day window returned eleven times what thirty days did, for the same single credit. Cost is the same either way because search bills per batch of records rather than per time window, so a longer lookback is free coverage.
Monthly if the book is large. Eight accounts produced two results in thirty days. Two hundred accounts would produce enough to act on, and a monthly rhythm fits how territory reviews usually run.
Never weekly. On this evidence a weekly digest returns an empty inbox that reads as “nothing changed” rather than “we cannot see it yet.” That is worse than not running it, because it builds false confidence in a stale account list.
Ad hoc before any account touch. The exception to all of the above. Before a QBR, a renewal, or a big meeting, check that specific account regardless of when the last sweep ran. A single account check is one credit and catches what a scheduled sweep has not yet seen.
