looking through a hole
PHOTO: Dmitry Ratushny

It's hard to get through a week without news of a new data breach. It's also difficult to avoid stories about companies using data in new, and sometimes creepy, ways. GDPR may eventually help stop the spread of these problems, but the odds are your company is keeping too much data, even if you are fully compliant.

Over time, the information your business collects loses its value. The people and companies you are tracking change. If they are ongoing customers, their actions the last six to 12 months will tell you most of what you need to know. Older data has value for analyzing long-term trends, but offers minimal value in predicting the next purchase. You need to make intelligent choices about how you use the data you collect and how long you store it.

A Little Information Can Be a Dangerous Thing

We’ve all heard about the father learning his daughter was pregnant from some very specific ads from Target. It's the go-to discussion point when identifying where the line is between effective marketing and being creepy. There are worse outcomes though.

How about sending those same ads to a woman who just had a stillbirth? It might not be hard to recognize this happened, if your algorithms are looking. If you aren’t looking, this can happen:

If your business specializes in selling baby gear, you may only know about the pregnancy through very specific, and obvious, actions. You may not be able to stop mistakes like this from happening. The question is, how hard should you rely on older data to override the latest information? In this situation, the most recent data clearly took precedence over everything that had come before.

Related Article: Marketers, Data Collection and the E-Word: Ethics

When it Comes to Marketing Campaigns, Old Data Is Bad Data

There is more to timeliness than simply avoiding creepy or painful situations. It can dramatically impact the bottom line.

A few years back, I was helping a company target its messaging. We determined that if a website visitor visited three related pages but hadn’t bought anything, they were primed for an email campaign. The target audience were known entities so there were no ethical issues around having their email addresses. The product wasn’t sensitive, so the odds of causing a problem if someone else saw it wasn’t a concern.

The campaign was pretty successful. Not everyone bought the product, but the company realized a higher than average success rate when compared to previous campaigns.

Fast forward six months. The company wanted to increase sales for a related product. It decided to put together a campaign that targeted anyone who had shown any interest in that product domain in the past three years.

The campaign failed.

Sure, some sales took place, but the number of unsubscribes to the company's emails was much higher than average. This had a directly negative impact on future campaigns. Why did this happen? It was hard to ask the people involved — they had unsubscribed after all.

However, we did notice a strong correlation between the age of the data and the success rate. The older the data, the worse the response. People had moved on from that product. Imagine an ad campaign trying to sell you what you already purchased on Amazon. Oh wait, you probably don’t have to imagine it. It probably happened to you this week.

Related Article: Data Is Getting Very Personal

Track History, Study Trends, Act on Today

Keeping a history of everything an active customer has bought from you is important. Aside from simply having that history, you can track if they’ve made a purchase, like a bike or training, that they probably won't need to buy again soon.

When studying overall trends, depending on your market, you may only need five years of data. You also don’t need any of that pesky personal information associated with those sales. You can extract that data from your sales database and see what people are buying over time. Using this separate, and anonymized data, can assist you in directing your investments.

And if in the course of that analysis, you determine that people who used to buy Product A are now buying Product B, then you can create campaigns to encourage people to purchase Product B from you before they buy from someone else.

But use common sense. If something is potentially personal, you may want to err on the side of caution. Maybe stick to coupons for the product in question mixed in with coupons for other, less sensitive items.

Related Article: 'Good Enough' Data Will Never Be Good Enough

Treat Customer Data With Respect

You need to study. You need to know your market and how they will react to what you do with their data. You need to provide them value without looking like an all-knowing stalker.

And you must protect your data from unauthorized access and usage. A single rogue action can have far-reaching consequences for you, your organization, and your customers.

Understand your data. Keep it clean. Be transparent as to how you use it. Keep your analytics separate from your transactions. And above all, don’t be a creep.