How to Fix Bad Data — and Use It to Feed CX


Bad data can be the bain of a brand’s existence. Learn how to transform this bad data and use it to promote great CX.

Poor data quality is often the cause of negative experiences for customers, leads and employees. Outdated, siloed, unformatted, duplicated and otherwise bad data can be the culprit that’s ruining your customer experience.

Misspelled names, inappropriate product suggestions, undeliverable messages, duplicate communications, inaccurate transactions and customer service histories — all these issues stem from bad data and lead to customer frustration, annoyance and overall negative emotional experiences.

What can brands do about bad data, and what causes it to occur?

What Makes Data Bad?

Consumers produce vast amounts of data every day through their interactions with brands’ websites, apps, service centers and chat servers. According to a 2020 LinkedIn Pulse report, every single person creates 1.7MB of data every second, and humanity produces 2.5 quintillion bytes of data every day.

With so much information being produced, how can brands ensure they’re not collecting bad data?

Keep in mind: bad data is not just a problem for brands interested in improving their customer experience — it also affects return on investment (ROI). In 2017, Gartner estimated that inferior data costs brands $9.7 million per year.

Data is considered to be “bad” if it is unstructured, inaccurate, inconsistent, incomplete or contains duplicate information. All of the data brands collect comes from a variety of channels, many of which are siloed, and much of the data comes in different formats or from different databases. Other data is more random and is not formatted, with no consistency, and must be aggregated in a structured, consistent way for it to be useful.

Michael Goodman, vice president of data, intelligence and automation at NTT DATA Services, told CMSWire that as humans, we store our ideas of the world and everyone we meet as feelings, memories and impressions. For businesses, this worldview exists through data, much of which has yet to be cleansed.

“The only way companies can store that same view of the world and everybody they meet, in this case customers, partners, etc., is as data and the insights and intelligence derived from it,” said Goodman. “The challenge is that raw data is often a fleeting resource and can be very messy.”

To correct bad data and turn it into good data, it must be “cleansed.” Data cleansing is described as the process of fixing unstructured, incomplete, incorrect, duplicate or otherwise erroneous data in a data set and involves identifying errors and updating, fixing or removing them, improving the quality of the data (i.e., making it “good” data).

Related Article: How to Prepare Data for Ingestion and Integration

What’s the Problem With Outdated Data?

Although it may seem negligible, data that is old or outdated is often worse than bad data. Brands that try to use outdated data to inform their decisions will be doing themselves and their customers a disservice.

Consider consumers in 2018 and how they approached brands and shopping, both online and in brick-and-mortar stores. Flash-forward to 2022, and the landscape (especially post-COVID-19) looks quite a bit different.

For instance, many shoppers today purchase their products online and then drive to a store to pick them up curbside — or have them delivered directly to their door.

Additionally, customer demographics change over relatively short amounts of time: people change their name, address, age, marry, have children, switch jobs, get promotions, adjust their income level, education level and, as noted above, change their shopping and spending habits.

On an individual level, using outdated historical customer purchase history may be deceiving or outright incorrect, especially when used to obtain actionable insights.

George Schoenstein, senior vice president of marketing and corporate communications at Fusion Connect, told CMSWire that his solution is to use a multi-pronged approach to data integrity and cleansing. “Our account representatives augment client contact information as part of their daily interactions with the client base.”

Additionally, Schoenstein said that his clients can update their own information within their systems. They also use third-party services to identify changes to client information, such as when someone’s switched jobs.

Is a Customer Data Platform the Solution?

Assad Jarrahian, chief product officer at Unanet, told CMSWire that metrics overload often distracts brands from focusing on what really moves the needle.

“New technology tools are available to capture data and filter out the anomalous, incomplete information,” said Jarrahian. “Automated analytics capabilities provide insight into data that is easy to view and manipulate.”

Obtaining actionable insights is one of the primary goals of utilizing good data, and fortunately, technology enables brands to wade through the mire to obtain them.

“To tune out the noise and hone in on the numbers that truly drive your business, companies should look for tools to help them easily glean actionable insights. These insights can reveal important trends and anomalies,” said Jarrahian, who added that high-level metrics should be simple, measurable and, most importantly, relevant to organizational objectives.

Customer data platforms (CDPs) are often used to tame the data beast. Steve Zisk, senior product marketing manager at Redpoint Global, told CMSWire that according to IBM, the annual cost of poor-quality data is over $3.1 trillion in the US alone. Brands have to run many different campaigns across a number of different digital and traditional channels, Zisk said, and fragmentation occurs because sifting through all that data without assistance can be difficult, and key information is often missed.

“That’s why marketers, especially those who need easy access to large data streams, may need to rethink their current data practices,” said Zisk. “Achieving high-quality data takes a dynamic customer data management platform that will maintain the quality needed for today’s real-time world.”

Zisk emphasized that a CDP should be enterprise-class, able to ingest data from internal and external sources in batches and in real-time, provide quality assurance (e.g., merging, matching and identity resolution), create an identity graph of each customer along with all profile and transactional history and provide automations and workflows that enable brands to work on strategy while the system handles personalized interactions at scale.

“Data pulled from a ‘no-frills’ CDP or stagnant data lake (aka, data swamp) may be fine for a small business, but it creates poor customer experiences for large organizations,” explained Zisk. “Data streamed from a comprehensive, dynamic CDP enables marketers to engage customers instead of exasperating them.”



Source link

We will be happy to hear your thoughts

Leave a reply

Logo
Reset Password
Shopping cart