Traditional "spin" bikes that use a flywheel and friction resistance can't report anything meaningful except cadence and time. Stationary exercise bikes that use magnetic resistance have the potential to provide reasonably accurate power readings, but the achieved accuracy depends on the quality of the calibration. If you know power accurately, you (or the bikes computer) can correlate that back to speed on a road bike under specific conditions. One of the big variables is rider weight, so if the computer doesn't ask you to provide that, everything it's telling you (except time and cadence and watts) is a WAG (wild-ass-guess). Conditions also include rider CdA, road surface, air temperature, and barometric pressure. If you don't know what the speed computation is using for those variables, you can't really make any on-road comparisons.