Basically the BLS performs two surveys, one of employers (which supplies the job numbers) and one of households (which supplies unemployment rate and number of employed people). Obviously, surveys have margins of errors (which is partly cleaned up via future revisions) and they don't measure the exact same thing. It's not that unusual for them to produce conflicting reports but in this case both numbers -- payrolls and unemployment -- rose much more than all the economists had forecast, right in the middle of a very key period, when the Feb has some very difficult decisions to make regarding interest rates.
If you want to dig deeper, start reading up on the seasonal adjustments and statistical models (specifically, Birth/Death model), there's a lot of ways to break down and to interpret (or re-interpret) the BLS data.
Here's a pretty good explanation: https://www.investopedia.com/hiring-increased-in-may-so-did-...
Basically the BLS performs two surveys, one of employers (which supplies the job numbers) and one of households (which supplies unemployment rate and number of employed people). Obviously, surveys have margins of errors (which is partly cleaned up via future revisions) and they don't measure the exact same thing. It's not that unusual for them to produce conflicting reports but in this case both numbers -- payrolls and unemployment -- rose much more than all the economists had forecast, right in the middle of a very key period, when the Feb has some very difficult decisions to make regarding interest rates.
If you want to dig deeper, start reading up on the seasonal adjustments and statistical models (specifically, Birth/Death model), there's a lot of ways to break down and to interpret (or re-interpret) the BLS data.