David Nelson, CEO of Traction Complete, moderated a discussion with Abby Gunning, Senior Manager of GTM Strategy & Operations at Medallia, and Riley Smith, Lead Systems Architect at Klaviyo. Together, they revealed practical AI use cases that revenue operations teams can implement today—and the validation strategies that make them trustworthy.
"Account data is at the heart of every B2B business. Everything we touch in revenue operations hinges off that account record—contacts, opportunities, contracts, they all roll up to accounts." - David Nelson
When this foundational data is incorrect, outdated, or incomplete, your entire go-to-market strategy suffers.
The webinar opened with a poll asking attendees what account data gaps they'd most like to fill. Surprisingly, the results were evenly split between firmographics, ICP fit scores, tech stack data, engagement history, and parent-child relationships. This even distribution highlights just how widespread data challenges are across revenue operations teams.
Gunning shared how Medallia began their AI journey with a simple but powerful use case: validating and enriching basic firmographic data. "We had a need to supplement what we were getting from our existing data providers and fill in some gaps," she explained, particularly for private and international companies where publicly available information is scarce.
Using Google's Gemini Pro, Gunning created a reusable "gem" (a standardized prompt) for account enrichment. Rather than typing the same questions repeatedly, the gem returns consistent firmographic data including company annual revenue, headquarters location, and industry classification—all critical for territory assignment and account segmentation.
The key insight? Start with validation, not replacement.
"I do validate against some of our third-party data providers. Sometimes if I find conflicting information, I'll dig further by looking at the company's website or other sources." - Abby Gunning
Smith took AI implementation to enterprise scale at Klaviyo, processing 1.1 million accounts to identify parent company hierarchies. "We completed it in two days thanks to batching," she revealed, noting that she had initially estimated a week for the project.
The challenge wasn't just identifying existing parent-child relationships—it was discovering missing accounts in their Salesforce instance.
"We were flattening our hierarchies by saying this child account reports right to the ultimate parent, and we're losing all the depth in between." - Riley Smith
Their AI solution identified not only parent companies but also suggested URLs for accounts that didn't exist in Salesforce, enabling their operations team to make informed decisions about which accounts to create. This approach prevents the common problem of bloated tech stacks while maintaining data integrity.
Gunning pushed AI capabilities further by extracting five years of historical revenue data for growth trajectory analysis.
"We needed more directional accuracy to understand if a company's revenue is going up, shrinking, or remaining stable." - Abby Gunning
This advanced use case required sophisticated prompt engineering, including:
The result? Account scoring models that help sales teams prioritize accounts based on growth potential rather than just current size.
Both experts emphasized that successful AI implementation requires patience and iteration. Gunning spent weeks refining her revenue analysis prompt, testing it on the same 20 companies repeatedly to ensure consistent outputs.
"I explicitly told Gemini in my prompt what columns to return and that it needed to return the same columns in the same order every time. Sometimes the output would be missing a column or two, and I would get very frustrated with Gemini and ask it why it wasn't following my instructions." - Abby Gunning
This iterative process revealed a crucial insight: AI tools require the same systematic approach as any revenue operations process. Document your prompts, test thoroughly, and refine based on results.
To gain organizational buy-in, especially from skeptical stakeholders, both experts recommend two key strategies:
Smith emphasized the importance of treating AI-enriched data like any other data source.
"Create a dispute process. You don't want end users to feel like they have no recourse if they spot an error. But require proof—don't take people at their word, just like you wouldn't take AI at its word." - Riley Smith
This approach aligns with broader data governance best practices and helps maintain data quality as AI adoption scales.
When asked about securing budget for AI initiatives, Gunning recommended framing AI as a strategy to reduce context switching and tool sprawl.
"Moving away from the need to stitch together data from different disconnected systems toward the synthesis of that data to drive decision making." - Abby Gunning
The business case becomes clearer when you connect AI capabilities to revenue outcomes:
Smith took a different approach, leveraging the inevitability of AI adoption.
"There's some inevitability with AI—not in a scary way, but in a 'this is where we're headed' way. Early adoption is great for any company." - Riley Smith
She also highlighted a practical reality:
"Reps are already using AI on their own to solve enrichment gaps. If you don't have a great AI solution or accurate data, people will go find it themselves, which is a huge waste of time." - Riley Smith
This resonates with findings from recent RevOps salary reports showing that AI-enabled teams are becoming more productive and commanding higher compensation.
Both experts stressed the importance of identifying specific pain points rather than implementing AI for its own sake.
"Focus on a real use case that is hurting your business. Frame it around that specific problem rather than AI as a band-aid solution for every problem at your company." - Riley Smith
This approach aligns with successful RevOps transformation strategies that prioritize business outcomes over technology adoption.
The experts warned against automating AI workflows too quickly.
"Anything that you can't validate multiple times—if you're getting inconsistent results on a particular field—then be cautious." - Riley Smith
"Don't jump too many steps ahead in the information you're asking AI to produce. Start with sourcing foundational data that can inform analysis rather than asking for complex recommendations." - Abby Gunning
Smith's advice for system design proved particularly valuable:
"Even if you're supplementing AI with enrichment vendors, have one field that you stamp with the actual final value. Don't show all the different potential sources—just give users the value you think is correct." -Riley Smith
This approach prevents the confusion and mistrust that can arise when users see conflicting data from multiple sources.
The webinar concluded with a live demonstration of Traction Complete's new Complete Discover tool, which allows teams to test AI enrichment in Google Sheets before implementing it in Salesforce. This "test before you trust" approach represents the future of AI adoption in RevOps—tools that make experimentation safe and validation straightforward.
"You want to find like the source of who is explaining to you what the results were, sources for the industry, annual revenues, et cetera. And you can see that there are some gaps where it didn't find an answer and didn't make it up." - David Nelson
The message from these RevOps leaders is clear: AI isn't just coming to account data management—it's already here. The question isn't whether to adopt AI, but how to implement it thoughtfully, validate it rigorously, and scale it strategically.
As organizations continue to generate more data and face increasing pressure to improve forecasting accuracy and operational efficiency, AI-powered account data management will become a competitive necessity, not just an advantage.
Check out our blog, join our community and subscribe to our YouTube Channel for more insights.
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
