If you ask Revelio Labs, the US economy lost jobs last month.

I’d suggest the Revelio tally — which, you’re reminded, is derived using individual-level data from more than 100 million online professional profiles — should be deemphasized now that the BLS is functioning normally, but who knows if it is (functioning normally) and if it is, who knows for how long.

Exactly none of the questions which prompted investors to put more emphasis on alternative measures of the US hiring impulse are answered, so those measures can’t be summarily dismissed as wholly subordinate and thereby meaningless. The BLS, I’d argue, has suffered a permanent loss of credibility, even if it’s still the most “actionable” release from traders’ perspective.

With that lengthy lead-in, the new Revelio report, released on Thursday, showed a net 16,647-job loss for last month.

There’s the chart. It plots the series with the BLS figures. All available revisions are included in both measures. As you can see, Revelio’s at least directionally consistent with the government’s tally.

The adjustments accompanying Thursday’s Revelio release transformed a 13,272-job loss for January into a near 27,000-job gain. I’d lampoon that were it not for the fact the BLS revisions are often just as laughable, particularly once you account for the impact of the annual benchmarking process.

By category, food services & accommodation and retail trade were responsible for the negative print. Revelio counted 62,000 lost jobs between them. The biggest job gains came in health care and social assistance.

Unlike Wednesday’s ADP release, the Revelio report suggested professional and business services added jobs last month. According to ADP, that sector lost jobs for a seventh consecutive month in February and a ninth in 10.

Take the Revelio numbers with a grain of salt ahead of Friday’s BLS release. But don’t dismiss them altogether, because if it were possible to quantify credibility deficits, the BLS’s red ink would be just as deep as the US government’s budget shortfall.