No Such Thing as a Single Version of Truth
By Jeff Jonas, published January 15, 2019
Interested in knowing what peculiarity occurs in aggregated data sets? If not, ignore this post.
When I’m asked which are the single best attributes to use when multiple attributes are known e.g., best name and best address, I respond, “There is no one answer, the truth is in the eye of the beholder.”
I learned this the hard way in the mid-90s. I was building a data warehouse fed daily by more than 4,000 data sources from a handful of widely recognized consumer brands. The goal was to entity resolve the combined data so the company could recognize when the same person was transacting across its different brands. The motivation: precision marketing.
There I was at the company’s headquarters with a number of marketing VP’s, each representing their brand’s interests. Even though everyone worked for the same parent company, when I said we must decide, here and now, how to create enterprise-wide golden records … I got schooled!
My seemingly easy question: If a consumer has transacted with multiple brands using a slightly different name or new address, which name and address should be considered the correct one?
The answer: It depends. I was given scenario after scenario where it was clear that my insistence on one version of truth was insane. For example, if the customer was a hotel loyalty club member, there was no way they were going to unilaterally change their mailing address to a “better” address and potentially send a cheating husband’s loyalty club statement to the address where the unwitting wife lives.
There was no way to determine an enterprise-wide single version of truth. At best, each brand could tell you how to calculate their truth.
Key takeaway: Be prepared to deliver every version of truth — for truth is in the eye of the beholder.
Truth on demand … so to speak.