Which private measure of total jobs is best?

The Street has this interesting article on two private measures of employment. The TrimTab estimate traditionally seems to be closer than the ADP number.

ADP's estimate compares with economists' consensus of 126,000, according to a survey by Bloomberg. TrimTabs Investment Research is even more cautious. The company said Wednesday that the U.S. economy likely added only 64,000 jobs in November . . .
How could two firms measuring employment come to such disparate figures? The difference between ADP's call of 206,000 and TrimTabs' 64,000 is a staggering 142,000. For some context, the Bureau of Labor statistics said the U.S. economy added only 80,000 jobs in October.
ADP says its estimate for private payrolls growth is derived from actual payroll data. Chances are that your paystub has the ADP logo in the top right corner, which means the firm is measuring jobs in the most direct way it can.
TrimTabs' employment estimates are based on an analysis of daily income-tax deposits to the U.S. Treasury from all salaried U.S. employees. Schnapp says that while the measure isn't perfect, it's a better view than a survey subject to revision. . . .
TrimTabs says estimates using tax deposits are historically more accurate than initial estimates from the Bureau of Labor Statistics. Schnapps says that once the government is through with revisions, TrimTabs' estimates are usually within 10% of the final figure.
So how did Schnapp react when she saw the ADP Employment Report show a substantially greater increase to private payrolls than her firm's estimates?
"It is frustrating because it moves the markets," she says of the ADP report. "One of us is going to be closer to the truth. Everyone looks at how close you are to the BLS. I can only call what the numbers are telling me." . . .
But as the chart below shows, the total employment data from the Bureau of Labor Statistics doesn't always match up perfectly to the private sector employment data ADP collects. . . .