In the latest episode of the RevOpsAF podcast, Drew Smith from Attributa warns that artificial intelligence will only be as reliable as the data it's fed. Smith argues that many revenue operations teams overestimate the quality of their campaign data, which leads to AI producing inaccurate or fabricated insights.
The episode outlines three foundational categories of campaign data that must be captured: time series data, category-level data, and method of engagement. Smith emphasizes that time series data—when a prospect interacts with a campaign—is the single most important data point for accurate attribution. Category-level data, such as campaign type and channel, must be consistent and normalized across the organization to avoid confusion.
Smith also highlights the need for unique campaign identifiers for multi-channel campaigns, allowing teams to track performance across different touchpoints. He points out a common language barrier between marketing and sales at the account level, where each department uses different terms for the same actions, leading to misaligned reporting.
Without a solid data dictionary and disciplined data capture, Smith warns, AI tools will produce "garbage out" results—hallucinating correlations or inventing insights that don't exist. The podcast serves as a practical guide for RevOps leaders looking to prepare their data infrastructure for AI-driven analytics.