Monthly "jobs created" is a wobbly signal. Estimate total "jobs" by compiling error-laden mis-timed survey data, then seasonally adjust. Subtract last month's rough estimate from this month's, for a delta ~1/1,000th of total jobs. Longer-term measures are a bit more robust.