4,333 sessions and one booking, against 1,860 sessions and eighteen
Session volume is the easiest number to put in a deck and among the least informative in it, because it can be raised almost without limit by accepting cheaper inventory.
4 min read
One property’s website, the same month, two campaign clusters.
| Cluster | Sessions | Checkouts | Sessions per checkout |
|---|---|---|---|
| Unmanaged campaign on the property’s own account | 4,333 | 1 | 4,333 |
| Managed cluster | 1,860 | 18 | 103 |
Of the 4,333 sessions, 1,469 came from Audience Network placements. The unmanaged campaign peaked at 3,033 sessions in a single week.
How it was measured
Sessions are recorded by the property’s own analytics rather than reported by the ad platform, so both clusters are counted the same way. Checkouts are booking-engine events on the same property. Both ran in the same month against the same website, the same inventory and the same rates, so the difference is not seasonality, price or product.
What accounts for the gap
The largest single factor is placement. Audience Network fills budget with low-cost inventory inside third-party applications, where intent is close to zero, and it accounted for roughly a third of the unmanaged campaign’s sessions on its own. The managed cluster excluded it.
That is why session volume is the easiest number to present and among the least informative available: it can be raised almost without limit by accepting cheaper inventory, and the resulting figure looks like progress in every report that stops at traffic.
Limits
One property, one month, two clusters. The ratio is specific to this account and should not be read as a benchmark. The mechanism is not specific to it: any campaign left on default placements will accumulate the same class of traffic.
Reproduce it
Break your sessions down by placement rather than by campaign, and compare cost per session against checkouts per session for each. The cheapest placement is usually the one producing nothing.