We use a heuristic to find the header row when two or more are present.
We spent a fair amount of effort doing this originally, but few people use multiple header rows, so we have not invested much effort on this since. The following tips might help us find the header. We “score” the first ten rows when looking for the header. The score is based on the contents of the cells in the row. We look for data values that look like header values as described below. The score is adjusted based on the rows location. The probability falls as the row number increased, so the first row is more likely to be a header that the tenth row. The highest scoring header row wins and is declared to be the header.
- Bold only the actual header row when you have two or more header rows. Bold is a strong signal to the heuristic.
- In your header use short text values that are alphabetic. Short to medium length text string count high. Numbers and long text string count low. The later look more like data values.
- We have a fairly long list of English header words we look for. Using these will yield a higher score. For example, we look for “Name”, “First Name”, “Last Name”, “Address”, 'Street", “City”, “ZIP”, “Postal Code”, “Amount”, 'Price", 'Quantity", etc.
- Empty cells are a strong negative signal. Header rows seldom omit header values.
This article covers this https://help.appsheet.com/app-design/spreadsheet-design/effective-use-of-column-headers
LinkedIn
Twitter
Remembered my old times 
