I’d automate the data pull first, and spot check what comes in. The quick checks on the numbers can run on every company too. I keep the reading for myself, and I only do it for the few companies that get through those checks.
Why the data needs spot checks
I once caught a data source showing zero debt for about 3,100 companies. Those companies just had no balance sheet from the last twelve months, and the source sent a zero where it should have left a blank. Almost 3,000 of them were listed in India, so you’d barely notice if you mostly screen US names.
The sneaky part is that zero debt looks like good news. A company with a missing number came out looking better than it should, and nothing on the screen told me. And that count had been growing for months.
Bad data like that can make a fine method look broken. Then you start tweaking a method that was working, when all you needed was to fix one column.
What the checks do
I run debt, free cash flow and sales growth on everything. In November 2026, only about 10% of the companies I track passed all three. I don’t bother valuing anything until those check out, so I value just that group and go deep on a few.
Since I only hold a handful, I get to know each one well, and that helps me ride out a bad year. After that it’s mostly waiting, and trying to keep my cool when a stock I own drops.
Where this breaks
A spot check only covers the companies you happen to look at. Blanks like that one can also come back after you fix them, so I don’t get to stop checking.