read

The next big thing in data quality is getting ready for AI

By Elias Rizk

 Artificial intelligence, AI, isn’t just a science fiction concept anymore. This innovation has become the new trend in information technology and it is about to bring massive change to just about every industry. It is projected to account for $36.8 billion in revenue by the end of the year 2025.

While not all industries and functions will change and evolve equally, marketing, sales, and customer relationship management stand to be roles that are affected by the transition to AI. Smart leaders, executives and professionals understand that having good data quality in place at all times will be one of the most influential factors in determining that their company is going to be successful using this technology.

Analytics, Automation, AI: What’s the Difference?

Because marketing and sales AI is so new, many people do not fully understand what it is or what it can do. Getting a clear idea of what this technology entails is critical to understanding why it has so much potential. While AI might be confused with two already-prevalent concepts, analytics and automation, it is significantly different from both.

  • Analytics is the art of using data to answer a question, usually by choosing two or more variables and comparing them to look for meaningful correlations. It’s a process made possible by technology that can produce some amazing insights, but it can only be performed directly by a data analyst.
  • Automation is nearly the opposite of analytics. Its purpose is to make it easier to accomplish the routine tasks like sending emails out at a specific time or tracking a lead’s progress through the funnel. It’s an extension of a professional’s capabilities, and it augments their output rather than making any contributions of its own.
  • AI, on the other hand, is something of a combination of both of these other things. Unlike analytical tools, it can operate mostly independently, but unlike automation, it can also be a direct driver of revenue on its own. Instead of choosing just a few data points to compare among contacts, it can juxtapose them all at once and tease out any significant relationships it finds along the way. It can also help you market to the exact tastes of each specific individual without having to put in any extra effort yourself. The end result is hyper-relevant marketing that produces respectable numbers of sales while easing the burdens on human staff.

Test Your Knowledge: What percentage of UK companies fear they might be wiped out by AI in the next 5 years?

AI’s Real-World Impact

AI might sound great in concept, but you may be wondering if it delivers on that promise in real business settings. For at least some businesses, the answer is a definite yes. Take the e-commerce giant Amazon, for instance. Much of this company’s incredible success has to do with the predictive AI it uses in its consumer-side search features. When you search for a particular item on Amazon while logged into your account, this system decides what it will show you first based on a multitude of individualized insights it has gleaned from your previous browsing and purchase data as well as the contact data they have stored on your record. This ensures that your top results are usually the ones that will appeal most to you, simplifying the shopping process for you and making a purchase more likely. This strategy works so well that 35% of Amazon’s gargantuan sales numbers are now coming directly from the recommendations this AI makes. This is just one example of what a company might do with a marketing AI; there are also potential applications for it in email marketing, social media interactions, product creation, and many other areas.

All of this is impressive enough on its own, but prospective adopters should keep in mind that this may very well only be the beginning of what AI is capable of. This field is making rapid advances and even today’s most generous estimates cannot account for breakthroughs that have yet to be made. What does seem likely, though, is that most organizations will have to use it eventually; in fact, 41% of companies in the UK fear that their current business model may be rendered obsolete by AI in just 5 years. While you should never base major decisions entirely on hypothetical scenarios like this, it’s important to remember that preparing your company for AI is as much about future-proofing your organization as anything else.

The Need For Quality Data

If you’re ready to start taking steps toward AI implementation, the first thing you have to worry about is the state of your internal contact data. A company’s entire set of quality data is currently worth between 5 and 11 times that company’s annual revenue. That’s because it is possibly the single most important factor in a company’s ability to generate income. With enough good data, you can determine everything you would ever want to know about your company and your contacts.

Marketers have already been exploiting the value of quality contact data for years, but the field is only just starting to discover how critically important this same data is to training AI systems. Because AI isn’t sentient, it can’t be taught in the usual manner. Instead, it must learn by absorbing information that is fed to it and breaking that knowledge down into useful patterns. Good data is as essential to this process as a thorough and accurate textbook is to a human student. Without useful data that is clear in its meaning and that provides an accurate picture of your contacts and their needs, the software will not have the context it needs to work its magic.

This is a problem that must be curbed before it ever manifests itself in your operations, or you risk having it go permanently unnoticed. No matter how smart an AI may appear, it still lacks human levels of context awareness and will simply accept whatever data it is given. It has no way of knowing what level of data quality is present nor of identifying any specific errors, so it will be unable to alert anyone to a potential problem if one crops up. You’ll certainly be aware of the lack of progress your AI is making when it fails to have a positive impact on your quarterly financial reports, but you won’t know why this is happening and you won’t know how to correct it.

Data Quality and the Downfall of the Watson Project

To better understand just how far awry an AI project can go without proper data quality standards in place, let’s take a look at a specific example. Even if you don’t follow technology news at all, you’ve probably heard of IBM’s Watson AI. The program was touted as the next big thing in cancer research and treatment and perhaps humanity’s best chance at curing the deadly disease. One of its earliest adopters was MD Anderson, the cancer centre of the University of Texas. This institution believed that Watson could be a big help in their quest to treat and potentially even cure cancer, but the system struggled to produce the results that were promised. After 4 years of attempts, MD Anderson was forced to abandon their work with Watson due to concerns over the viability of the project, wasting over $62 million in funds that had been dedicated to it up to that point.

One of the major problems that impeded Watson’s effectiveness, in this case, is a lack of data quality in the hospital records that were used. Many of the records in this set were not only handwritten and tricky for even humans to read, but also included many diverse shorthand notes and remarks that were difficult to decipher for anyone other than the original author. They also suffered from the kinds of general mistakes (think typos and small copying errors) that you might expect to see in any set of low-quality data. As a result, Watson could not properly interpret the data that was being presented to it. The research team had to spend countless hours catching up on their data quality management tasks before they could make any progress with the affected records. Even then, they never fully succeeded at the task – hence the project’s cancellation. We can never know what impact Watson might have had on MD Anderson’s cancer treatment success rate had everything gone according to plan, but we do know that poor data quality played a major role in its failure.

Staying One Step Ahead of the AI Game

Marketing AI is still a nascent technology with an uncertain future, but the one thing we know about it for certain is that quality data will be an integral factor in its success. No matter how much more powerful any AI system gets, it will always rely on having quality data to get itself started. Contact data quality management is as an investment toward current and future AI initiatives at your company. The relationship between data and artificial intelligence comes down to one very simple fact: if your data quality isn’t where it needs to be, you will never be able to achieve successful results with AI.

Test Your Knowledge: What percentage of all sales lead records is thought to contain bad data?

We’ll have the answer to this question ready for you next time as we continue to explore what data quality can bring to an organization. Subscribers will be the first to know when our next post comes out, so if you want to read it as soon as possible, consider subscribing now.

join your peers

Tags: Sales performance, Contact data quality management, Marketing performance