How good is your data chain of custody?

Maybe you got one of the many CSI box sets for Christmas, and are snuggling down in your onesie to hear various suspiciously good-looking scientists intone “The evidence never lies”. One of the themes that also crops up is “preserving the chain of custody”, from corpse to court. The key here is that if a piece of evidence is going to be used in court, then there needs to be a full audit trail which documents all of its movements, processes, and locations all the way back to the crime scene. There are plenty of high-powered defence lawyers on the show ready to cry “Challenge!”, and making sure the “chain of custody” is rock solid is critical to winning the case.

In business analytics, often the chain of custody breaks down under a challenge

Imagine one of your executives looking at a dashboard and seeing that on-time delivery or quality is way down at one of the plants. How does she evaluate the precision and reliability of that number? Who created it, and when? Has it been aggregated or passed through some analyst’s spreadsheet on the way? Are we sure that the supplier is counting “days late” the same way that we do? If you want the title “Data Scientist” (and let’s face it, it impresses people at parties), then you need to start acting like a scientist.

So how reliable is that number?

In most ERP systems, the auditability of numbers is pretty solid. The concept of “drill down” was introduced by SAP in the 1990s to exactly provide this level of control: so that when anyone saw any number, perhaps in a general ledger, they could drill down to the associated transactions to bolster their confidence that the number was accurate. This was fine when most data decisions were made on internal data: but how can we deliver the same level of confidence when much of the critical data has come from outside the company?

Business Networks Provide the External Chain of Custody

This is where business networks come in: they can provide the critical authentication steps so that the same level of confidence can be used for external data. Rather than uploading a spreadsheet from a business partner, the data is hooked up directly to their shipping system, and authenticated along the way. So a business network is able to act as a secure “outside lab” in your diagnosis.

Some of the ways this is done:

• External companies have to go through an authentication process so you know who they are
• All companies must agree to Terms & Conditions regarding data security and other protocols
• Electronic signatures are used for some documents (e.g. invoices)
• Bad data can be rejected via business rules

Although this piece of evidence isn’t being used to send a hoodlum to the slammer, it still might be used to consider switching suppliers, changing plant production, or putting a manager on notice. You want to be able to go into the next quarterly meeting and be able to confidently say, “We follow the evidence, wherever it leads” and not have anyone yelling “Challenge!” and questioning your Chain of Custody.