All right, in the natural vs surrogate key religious war, I rather feel on the natural key side (I have nothing against using a surrogate key as a shorthand for a complex primary key, but I have seen too many people adding to a table a system-incremented column and calling it primary key without other analysis).
But I must admit that counters, whether you call them identity or auto-increment columns, or simply sequences, are extremely valuable for relating the techy’s to the bean counter’s vision of activity. In most applications, there is at least one such counter that represents fairly acurately the business activity - whether it is an increasing order number, invoice number, transaction identifier or whatever. Rather than execute complex queries with conditions on date columns to collect those so precious “key performance indicators”, it may be easier to check the data dictionary, look for sequences or identity columns, and take a daily (or anything) snapshot of the current highest values, together with statistics about the number of executed statements and whatever your DBMS has to offer.
With a good collection of counters, computing differences between successive snapshots will give you a simple, yet fair idea of what your system produced during this interval. You will be able to show awful charts, compute trends, ratios to compare the efficiency (amount of work to generate comparable results) between successive releases of an application … What more do you want?