The famous “for want of a nail” allegory describes how a cascading series of seemingly insignificant events eventually causes the loss of a kingdom, showing that small details matter. This is true for calculating the business cost of bad data, where small mistakes may easily lead to the loss of a sale or loyal customer.
An incorrect email address. An offer for a product the customer recently purchased. A misunderstanding of household dynamics. A customer in a wrong segment. A mis-spelled or incorrect name or address. A late or an irrelevant response to an abandoned shopping cart. The list goes on.
When bad data creeps in, what happens is that marketers and executives are conditioned to accept loss, in one form or another. Fewer conversions. More attrition and/or less retention. Is 2 percent acceptable? Five percent? Whatever it is, there’s a certain number that will be tolerated for one of two reasons. It either comes down to a decision that the annual cost of bad data is less than the cost of fixing the problem, i.e. perfecting data, or it’s more general despair over not knowing which holes to plug, a determination that the problem can never truly be isolated.
Nailing Down the Cost of Bad Data
The problem with both lines of thinking, however, is that whatever the agreed-upon number – 2 percent, 5 percent, etc. – is not the true cost of bad data. If, for example, 2 percent of your emails are wrong, you also need to factor in opportunity cost; who didn’t receive that email that should have? How many of that 2 percent would have taken the next step in the customer journey? Now say you’re running five campaigns a week. Over the course of a year, that initial 2 percent has suddenly ballooned to a significant number of missed opportunities.
Consider the ramifications for a static email offer for, say, 10 percent off a gas grill to an audience of homeowners ages 35-50. Brands without an updated unified customer profile that includes behaviors on all channels run the risk of the offer being irrelevant the moment it’s opened. Any member of the intended audience who happened to visit a store and paid full price for the grill before opening the email are now annoyed when they do open it. Is the offer still available on the product they’ve already purchased? Now instead of being proud of their new purchase and excited to host a backyard barbeque, the product reminds them of the bad experience. How many of those customers never return? Multiply that times the average customer yearly spend.
Also factor in the numerous missed upsell or cross-sell opportunities if the customer profile doesn’t really tell you everything there is to know about a customer. Maybe it has transaction history, some PII, an account number, etc. but it doesn’t contain all relevant data – affinity, preferences, demographics. Conversely, with a unified customer profile that is updated in real-time that tells a brand everything there is to know about a customer, a brand can update content on-the-fly until the moment of interaction.
If you know the customer bought the grill at the store, perhaps you change up the email with a 10 percent offer for a stainless steel set of grilling tools with next-day delivery and a brochure on how to sear the perfect ribeye. Or perhaps a customer browses the website, and because your golden record – the unified customer profile – includes household dynamics you know the visitor is the father, not the son, and so you display images and content for a more expensive line of grills with higher profit margins. Or you know the visitor is a caterer or a contractor browsing the website for a portable commercial grill. Maybe instead of an offer for grilling tools, that customer receives a discount offer for an annual bulk propane subscription. By knowing everything there is to know about a customer, brands can move away from random audience segments that limit a success pool, instead creating far more granular segments that match offers, content and actions to what truly matters to a customer at the moment of interaction.
Leave Irrelevance up in Smoke
The above examples start to get at the real cost of bad data and the repercussions of trying to market to a customer without an accurate, up-to-date unified customer profile that is instantly accessible so that real-time decisions are executed as the customer journey is ongoing. It’s not just a static 1 percent or 2 percent – it’s compounded through each phase of the customer journey, with opportunity cost factored in at every step.
By knowing everything there is to know about a customer – every transaction, online session, preferences, interests, etc. – a brand reduces or eliminates wasted communications, irrelevant messaging and lost opportunities. When data quality processes are completed at data ingestion, including advanced identity resolution with probabilistic and heuristic matching, accepting a certain loss percentage because of bad data becomes a thing of the past.
Perfect data is possible. And with it the cascading series of compounded mistakes that cause the true cost of bad data to pile up are stopped in their tracks.