In my mind the most statistically useful division between residential and business might be whether the account is signed up under a personal entity, or a registered company/charity/etc.
That'd be a fairly coarse filter; there are of course plenty of overlaps, in the home-business market and in the rural sense where connections may be aggregated and have multiple uses, etc.

In other news I don't suppose there's anyone from Stats on NZNOG who could invite a more structured response to the concerns being raised about the validity of the stats?
If ISP's are obliged to take the time to send a return, it seems chronically wasteful if the data can't be trusted.

Mark.

On 16/10/2015 4:10 p.m., Stan Rivett wrote:
Hi Don

Yes, that's why I asked. I scratch my head over the Business / Residential division every time I do the survey. My method of categorising it comes down to something like:

IIf( query.company eq '' , "Residential", "Business")

Not, I suspect, a very scientific method ;-)

Cheers�

Stan Rivett
------------------
Netspeed
PO Box 5691
Dunedin
P: +64 3 481 7245
C: +64 21 323 841
------------------

On 16 October 2015 at 15:14, Don Stokes <don@daedalus.co.nz> wrote:
The stats come from an annual� ISP survey done by the Department of Statistics. Survey returns are compulsory, but the figures are self-reports by ISPs using their own interpretations of the categories. In most cases, ISPs will report "business users" as users of "business" plans, but other divisions may be used.

The individual ISPs are probably in a better position than anyone else to report proportions of business vs residential usage, but you still need to apply appropriate condiments to the numbers. In many cases, an ISP may not know if a customer is operating as a business or not. As you observe, many connections are used for both residential and business purposes, and I can't imagine any ISP putting any serious research into this for the sake of answering one question on one annual survey.

There are other questions that may also suffer an element of this kind of fuzz. Again, check your sodium intake.

-- don


*snip*