The more potentially disruptive the technology, the greater the hype.
While it’s undeniable that generative AI is already changing how businesses engage with the world, sales and marketing leaders are quickly realizing that AI alone isn’t the solution to all their problems. Without a trustworthy, reliable data foundation, even the most sophisticated generative AI systems will not only fall short of expectations — they could actually supercharge routine error, scaling costly mistakes across a company.
According to ZoomInfo founder and CEO Henry Schuck, ensuring data accuracy is one of the greatest challenges faced by companies evaluating generative AI technologies. Speaking at GTM Partners’ Leaders Summit on Generative AI, Schuck advised business leaders to make data integrity a top priority.
“Start thinking about your data as an infrastructural element of your technology stack,” Schuck says. “It’s just as important as any [IT] infrastructure you have, and ensuring data is incredibly accurate and usable by generative AI will be an infrastructural layer in the future.”
Bad Data, Bad Outcomes
The great advantage of next-generation AI tools like ChatGPT is their ability to produce outputs that are tuned to what a human user wants to hear. But they also run the risk of producing results that are remarkably convincing but also wildly inaccurate.
Press reports, in fact, indicate that employees at Google were so worried about rushing an AI chatbot into public hands that they warned the app was “a pathological liar” and “worse than useless.”
Adding those deviations to a business world awash with uneven data quality could quickly create exponential headaches.
Researchers at credit monitoring agency Experian, for example, determined that the costs of bad data can exceed between 15–25% of a company’s total revenue, driven largely by the time and productivity costs of correcting errors and manually verifying data outputs. Experian also found that only half of the companies surveyed feel their customer data is accurate enough to be used in their go-to-market (GTM) motions.
That combination explains why Boston Consulting Group’s recent CEO’s Roadmap on Generative AI identified the potential for falsehoods as so significant that “companies must mandate double-checking all generative AI outputs, and limiting its use to noncritical tasks.”
As ZoomInfo Senior Vice President Ben Salzman said in a recent LinkedIn Live demo, using generative AI without quality data will prove to be “an accelerant to badness.”
“Companies are going to go headlong into this with their existing data infrastructure,” Salzman says. “But if you layer generative AI on top of bad data, it’s going to get you to bad results real fast.”
Automation Requires Greater Oversight
As anyone who works with structured data can attest, data cleansing is both tedious and critical. Thomas Redman, president of Data Quality Solutions, suggests that up to 80% of the work behind creating AI models goes into data cleaning.
“It is time-consuming, fraught, expensive, and simply no fun,” Redman wrote in CDO Magazine. “Worse, with even the best cleaning, errors go undetected and/or uncorrected, with no way to understand the impact on the predictive model.”
At ZoomInfo, we’re watching these developments closely because data quality is core to our business. With a foundation of the highest quality company and contact data, married with cutting-edge signals such as intent and proprietary survey results, our data is the trusted solution for GTM teams of every size — and the fuel for next-generation applications like generative AI.
Too Much Data, Not Enough Insight
Two decades ago, businesses like Salesforce revolutionized how companies went to market by providing insights into valuable customer data. Businesses gained more visibility into their customers and markets, and no longer had to cast ever wider nets in the hope of reaching new audiences.
Today, the problem has been reversed. Businesses in every sector and vertical are overwhelmed by data. Acting upon increasing amounts of data in a timely way to capitalize on emerging opportunities is difficult to impossible.
Chorus by ZoomInfo is helping to close that gap by using generative AI. Chorus records, transcribes, and analyzes video and phone calls to harvest key GTM insights. It can turn a completed call into an actionable after-meeting brief, complete with action items and next steps.
“I used to spend probably an hour or two a day just going back and reviewing the tape from calls that had happened,” Salzman says. “I’m able to see now what happened on every call, including ones that I wasn’t able to attend. This has probably been one of the biggest unlocks, and it’s gone super fast from trying to develop this to actually getting it in-market.”
Next steps for ZoomInfo include baking generative AI call analysis into sales-call coaching. We’re making recommendations for reps more detailed and actionable — offering specific packages to sell depending on the prospect and the stage they’re at in the sales cycle, for example.
Salzman also demonstrated a pre-release feature for SalesOS that pulls data from a contact profile, including the prospect’s tech stack, their location, and contact data, and creates a relevant outreach message from just a few plain-language prompts.
“It’s basically doing exactly what a human would do when they look at what tech they’re integrating with, where they’re based, and more,” Salzman says. “This is going to get even more sophisticated over time.”
Designing the Modern Tech Stack
One of the most exciting developments in AI outlined in a recent McKinsey Digital report is the fundamental shift in which job functions are seeing the greatest ROI from AI technologies. In 2018, manufacturing and risk-management were among the leading industries benefiting from AI.
Today, it’s sales and marketing.
But it’s not enough for businesses to invest in quality, reliable data. It’s vital that companies seeking to harness the power of generative AI design tech stacks that are flexible and modular to allow for easy integration of new technologies. Flexible data architectures also allow even nontechnical employees to design, deploy, and develop campaigns using generative AI.
“A lot of times, when we talk about what generative AI is going to do for salespeople or account managers or marketers, we focus on the ‘last mile,’” Schuck says. “We go, look at what an amazing experience this is going to be, to be able to deliver a customized email that knows everything about the customer, and then automatically prospect or reach out to potential customers for prospecting or upsell or cross-sell.
“When you have this really unique, proprietary dataset, you start thinking about how do I get to that ‘last mile,’ and how accurate does the information in my systems need to be? How do I enhance the information in my systems so that the ‘last mile’ can be really intuitive and contextual?”
Quality Data: The Fuel for Modern GTM
Many sales and marketing professionals are already using generative AI to reach new markets more effectively and improve their engagement with prospective customers.
However, it’s vital for revenue leaders to understand that generative AI is only one part of a sophisticated, modern GTM strategy. Just as the introduction of the CRM two decades ago both solved old challenges and introduced new ones, generative AI is forcing executives to critically examine the strength of their data foundation, and reevaluate how and where they plan to invest.
Soon, the landscape of B2B sales will be split into those who can act upon their data to seize competitive advantages and those who can’t. More than 35,000 companies trust ZoomInfo’s superior data and integrated platform to fuel their growth engines — let us help fuel yours.