Share of bookings and share of revenue do not match, and the platform only sees one of them
The band that booked most often was not the band worth most. Optimise for booking count and the system correctly prefers the cheaper guest.
4 min read
One property, all attributed bookings to date, by guest age.
| Guest age | Bookings | Share of bookings | Share of total booking value | Average booking value, indexed |
|---|---|---|---|---|
| 25 to 34 | 62 | 22.7% | 13.6% | 1.00 |
| 35 to 44 | 86 | 31.5% | 44.1% | 2.34 |
| 45 to 54 | 112 | 41.0% | 37.1% | 1.51 |
How it was measured
Attributed bookings and their value, grouped by the age bracket the platform reports for the reservation, summed across the whole attributed history for the property. Share of bookings is that band’s reservations over all reservations in the table; share of value is that band’s revenue over all revenue in it.
Average booking value is indexed against the youngest band, because the underlying figures are in the currency the property bills in, and a ratio carries the finding without publishing a hotel’s revenue or inviting a currency conversion that would invent a number.
The finding is the mismatch between the last two columns
The 45 to 54 band produced the most reservations of any group. The 35 to 44 band produced fewer, and more revenue.
Optimise toward booking count, which is how nearly every hotel account is configured, and the system correctly prefers the band worth less per reservation. Cost per booking improves the whole time. The drift is therefore invisible on the dashboard anyone would use to detect it.
Two bands are deliberately excluded, and this is the important limit
The 55 to 64 band holds 11 reservations. The 65 and over band holds 2. An earlier version of this analysis reported a smooth rise across all five bands, with the highest value at 65 and over, and that rise turned out to be an artefact of the two smallest samples sitting at the top of the curve.
It is not published. The correction is recorded here rather than removed without a note, because a statistic that failed on its own sample size is worth more as a warning than it ever was as a finding.
Limits
One property, one attribution window, and the platform’s own demographic reporting, which is an estimate rather than a passport. The three bands shown carry 62, 86 and 112 reservations. That supports comparing them with each other. It does not support extrapolating to hotels in general.
What generalises is the mechanism: a platform optimising on conversion count is indifferent to conversion value, and drifts toward the cheaper segment wherever the two diverge.
Reproduce it
Pull share of bookings and share of revenue by whichever segments your platform reports, and put them in one table. Check the reservation count behind every row before you believe any average in it.