- Rule 1: the number cannot represent a label. What this means is you numbers can't represent something else, such as your flight number (which represents your flight pattern to/from a location), your home address is also a label etc.
- Rule 2: NO Restrictions on the number. You can't restrict the upper and lower limits of what the numbers represent.
- Rule 3: Small is small and Large is LARGE. Basically a small number like a 1 must represent something small and 1,000,000 should represent something BIG. This doesn't conflict with Rule #2 because there are plenty of numbers out there in our great universe that you can use Benford, which I'll describe below.
- Rule 4: You Must have at LEAST 4,000 transactions you're review to leverage Benford's model. This is my own rule (so please note it's not been proven but rather I have experienced this in practice).
So now that you got the background you must be saying to yourself. WOW! Sonia has a lot of time on her hands. NO! I'm actually quite busy, and really my prospects and clients either see Benford as something "COOL". There I said it and yes, I'm an auditor so maybe using the word "Cool" and "Auditor" is an oxymoron.
Those that say "Gee, that's ok but not interested" I'm just curious why they don't see the value. Most will say "I have great controls so there's no need for data analytics like this and I trust my people" or "What's the cost because if it's more than what we're paying now I don't want it".
Those that say it's COOL usually will have us perform the analysis once or twice then drop off their enthusiasm for performing this type of work. I'll continue the struggle to show this Law in it's truest form and hope that enough folks today will keep the Benford dream alive. I think it'll go somewhere and have performed a few webinars on the topic hoping that some will actually perform the analysis themselves.
For now we'll have to perform it as part of the internal audit package. Currently, I'm starting a grass roots effort to get a special HIT list for internal auditors to ask them, "If you could have your top 10 dream analytics what would they be?". If you have your own list please share as I'd like to continue the teaching that I'm doing at both the local level and the grand ol internet.