Let’s talk about a fairly complex business here where you are tasked with building some analytics.
Your client is none other than … Santa Claus!
Yes … Really!
He runs a fairly complex operation and it’s not going to be that easy.
Let’s start with business requirements and scoping. Oh yes, we do this with any Data Vault Implementation as well. As Dan states, you need to bring in 100% of the data (in scope) 100% of the time.
So, let’s examine his operations.
He has customers of varying demographics – age, gender, geography, language etc. They send in requests by magic mail or by merely wishing it which somehow transforms into a printed list he can use to optimize inventory. I detect some unstructured data from the scanned letters and semi-structured data in the form of e-mails. Perhaps some massive parallel transformation engine … who knows.
Like all good businesses he has scores attached to the customers, and sometimes when they don’t meet their minimum score of “being nice”, they may be moved down the priority list for later. A side-effect of the best customers moving up as well.
We’ll simply call this his Customer Relationship Management system or CRM.
Does he use an identifier for the customers? You’ll have to find out.
Then he has a toy operation manufacturing in his own facility which requires raw materials, process optimizations, inventory management and other manufacturing related operations.
Let’s call this his MRP or Manufacturing Resource Planning (with subsystems in it for SCM or supplier chain management).
He also needs to source many more toys (and other gifts) from traditional toy manufacturers and suppliers using his Vendor Management systems. Some of it ties back into the SCM as well since they’re doing procurement of stuff whether raw or completed. Hopefully, there’s a chain connecting us to it or we’ll have an entity/identity resolution scenario on our hands as well. Lets just hope they already have their master data identified in an MDM system.
Then there are the delivery systems which work around the clock and deal with different time zones and delivery route optimizations. And the combination feedback and payment system which is traditionally done with the universal currency of
… milk and cookies (All that sugar. No wonder he flies so fast).
Phew … I’m sure I missed something, but this is already getting fairly complex.
So, lets talk about metrics he may want to check:
1. Improvements over prior years and trending – Are they getting more orders or less orders. They’re paid back in smiles which power their operations and if the supply of smiles starts getting low, it directly affects the operations. It also affects the livelihood of hardworking elves by cutting down their candy quotas or worse … leaving them with nothing to do.
2. The standard production related metrics.
3. Analysis of toy popularity and trending as well as predicting which ones will repeat so they can prepare for the next year.
4. Regulatory reporting – Maybe. Don’t know, but I’ve yet to see an operation not requiring it (Like we talked about the US banks yesterday). Do you know if he is required to report to anyone? They’re usually boring and predictable anyway.
5. Insert your own silly metric here (Try it … It’s fun)
Anyway, so after you’ve determined scope and requirements, you need to follow it by identifying the information gaps between what’s needed to be delivered (business wise) to what you have available (sources). After that, you’ll work on “real” business key identification as well as alignmentacross these systems.
After the alignment is done, you can actually start building out the Data Warehouse objects which we’ll of course recommend as a Data Vault (or DV 2.0) using the core constructs of hubs, links and satellite structures. There can be a parallel effort designing the front-end marts as well.
And so on and so forth … till your first set of deliverables are done – ideally in short sprints. Days and/or Weeks.
Santa’s stuff can get fairly complicated with unstructured data in the form of letters from which information has to be extracted, peaks and lows in demand due to seasons and year over year which will require optimizations. You have existing predictable mission-critical systems that just run all the time and all this information needs to be integrated. Lots of systems and lots of data.
If you were to build a Business Intelligence infrastructure for Santa’s organization, I’d recommend you use a Data Vault 2.0 System Of Business Intelligence
Here’s wishing you a Very Merry Christmas and Happy Holidays.
(with Dan Linstedt – Inventor of the Data Vault and DV 2.0)