One way to think about aging data is that the older it gets, the more it behaves like radioactive material. While the term radioactivity was originally coined to describe the emission of energy from an atom when its nucleus breaks down, it’s now come to symbolize anything that breaks down over time. While the data we use in business every day isn’t susceptible to the same kind of physical decay, it can lose its potency over time calling into question the value of aging data.
Radioactive data is literally information that has lost value. While there are no definitive rules for how quickly data decays, a Deloitte report underscores that the expected life of a data asset directly impacts its value. “As with other assets within a business, in order to understand the value of its data asset, an organization must understand how data impacts enterprise growth, returns, and risks,” the report advises.
It also is important to understand how document automation can mitigate against data decay. In particular, automation helps the business activate unstructured data, making it available earlier in its value cycle.
Aging Data Is Radioactive
It can be difficult to determine when data has reached the end of its shelf life. As a recent article in Forbes notes, “stale data has no smell, but still leaves a bad taste.”
How fast data goes bad often depends on what type of information it is. For example, according to the Sales & Marketing Institute, up to 96% of email addresses and contact data within customer files and CRM systems are at least partially inaccurate. In fact, CRM data degradation is estimated at 2% or more per month. That means any marketing activity that hinges on accurate CRM information is compromised, leading to sub-optimal results.
Maintaining data accuracy is no small task. Businesses close or relocate. People leave their jobs or retire. Phone numbers change. URLs and email addresses are modified. According to a Dun & Bradstreet report, about half a million business addresses change every year. In addition, more than 83,000 CEOs leave their jobs each year, working out to over 227 a day. Many other roles have even higher rates of turnover. Old data being used to target decision-makers who have left the building is, well, useless.
Radioactive data also can be expensive. According to the Gartner Group, poor data quality costs enterprises an average of $8.2 million annually due to squandered resources, operational inefficiencies, missed sales, and unrealized opportunities.
Furthermore, the accelerating pace of data degradation can exacerbate database quality issues. By contrast, ensuring that data is accurate and complete can translate into a gold mine of intelligence for improving business decision-making and strengthening competitiveness.
The massive volume of data generated by companies, large and small, can be overwhelming. Preventing your data from becoming radioactive requires a robust data management strategy, as well as streamlined document automation workflows.
Dun and Bradstreet recommends using a simple formula to help ensure companies “ACT” proactively to keep their data from becoming radioactive. ACT is an acronym for Accuracy, Completeness, and Timeliness. Bottom line, if you can’t ensure these three qualities about your data, it’s probably not all that useful for current business decision-making.
What’s more, personalized, multi-channel marketing is predicated on accurate customer and prospect data. Radioactive data increases the likelihood that digital and physical messages will miss the mark because the target has either changed coordinates or disappeared.
Investing in keeping databases up-to-date can pay dividends. According to Dun and Bradstreet, the ROI from data accuracy includes faster payment of receivables and improved cash flow. Good data also increases up-sell and cross-sell opportunities by as much as 3%, and reduces vendor costs by as much as 20%.
In the same report, one CIO explained managing customer data quality this way: “When I eat I don’t measure the ROI for food. I simply need to eat or I will starve. The same goes for managing customer data quality—we simply need it. If we don’t manage our customer information well, we are starving our company of potential dollars. For some things, I don’t need an ROI calculation. The benefits are obvious.”
Some Old Data Has Value
Old data tends to lose its value over time, but some can still be useful. Artificial intelligence applications are data-hungry. In particular, machine learning algorithms rely on massive datasets to learn.
Of course, the data used to train algorithms needs to be clean and relevant. As we noted in another blog post, machine learning requires companies to pay attention to the old adage of “garbage in, garbage out.” While old data may be useful in training machine learning models, inaccurate data could result in less than optimal results.
Researchers at Columbia University, for example, recently used historic data from patient records to develop treatment protocols for people with pneumonia. The researchers used historic data from clinics to help develop an accurate algorithm to identify high-risk patients.
In the financial services sector, banks and insurance companies use historic consumer data to train machine learning models to assess credit risk and automate portfolio management. At Ocrolus, we use historic data culled from processing more than 500 million financial transactions a year to help integrate robotic process automation with machine learning.
While historical data can be useful for training machine learning models, the financial services industry also depends on having access to real-time information for everything from customer credit histories to loan-to-value ratios.
To sum up, the best way to ensure that your data does not become radioactive is to invest in keeping it up-to-date and accurate. If you want to be data-rich, you need to be smart about how you harness the power of information with document indexing and automation.
Learn more about how Ocrolus’s document automation software can help organizations take better advantage of their data.