Key Takeaways
- • ZoomInfo launches GTM Bench, a vendor-run benchmark system evaluating large language models and AI agents on go-to-market tasks including target list building, record enrichment, account scoring, and decision-maker identification across 20+ jobs, 4 systems, and 3 models.
- • GTM.AI significantly outperforms competitors on the index: ZoomInfo's system scored 77 on the GTM Bench Index versus 47 for Apollo, 36 for Exa, and 31 for open-web search, completing 98% of requested work with a cost of $0.79 per task.
- • Data grounding and verification are core metrics: The benchmark measures "Answer" (work completion rate) and "Grounding" (share of returned data traceable to verified sources), with ZoomInfo returning 478 verifiable records per 1,000 versus 7 to 35 for competing systems.
- • ZoomInfo acknowledges benchmark limitations: The company disclosed four categories where its advantage is limited or absent, including copywriting and owned CRM data that external tools cannot access, and noted grounding is measured against ZoomInfo's own verified records rather than independent audit.
- • Stock valuation disconnect persists despite strong fundamentals: Despite 72% share price decline over the past year to $2.94, ZoomInfo maintains 86.75% gross profit margin, $1.25 billion in annual revenue, and profitability, with InvestingPro analysis suggesting current undervaluation.
Summary
ZoomInfo released GTM Bench, a benchmark system designed to evaluate the performance of large language models and AI agents on go-to-market core functions. The initial version covers more than 20 distinct jobs, 4 systems, and 3 models, with methodology, sample tasks, and grading rubrics made publicly available. The benchmark assesses systems across two primary dimensions: Answer, which measures the percentage of requested work completed, and Grounding, which evaluates the share of returned data that can be traced to a real, current source.
In the benchmark's inaugural results, ZoomInfo's GTM.AI achieved a GTM Bench Index score of 77, substantially outpacing three competing systems—Apollo at 47, Exa at 36, and open-web search at 31. ZoomInfo's system completed 98% of the operator's requested work product and returned 478 verifiable records per 1,000 queries, compared to 7 to 35 records for other tested systems. The cost per task executed by ZoomInfo's system was $0.79, positioning it as both more effective and operationally efficient than alternatives evaluated.
Demonstrating transparency regarding its benchmark methodology, ZoomInfo acknowledged that the system is vendor-run and disclosed four categories where its competitive advantage is limited or absent: copywriting, owned CRM data that external tools cannot access, and other specialized applications. The company also noted that grounding measurements are calibrated against ZoomInfo's verified records rather than through independent third-party audit, a material distinction for institutional investors evaluating claimed performance advantages.
ZoomInfo indicated that Version 2 of the benchmark will expand to include agentic multi-step workflows, international coverage, and an owned-data axis. The company further stated that other data providers and AI agent builders can submit systems for testing and inclusion on the public leaderboard, positioning GTM Bench as an evolving competitive assessment tool rather than a closed proprietary framework.
The benchmark release occurs amid significant market headwinds for ZoomInfo. The company's stock has declined 72% over the past year to $2.94 per share, despite maintaining operational strength: gross profit margin of 86.75%, $1.25 billion in annual revenue, and continued profitability. The company is valued at $867 million. InvestingPro analysis indicates the stock is trading below intrinsic value, suggesting potential disconnect between market sentiment and underlying financial performance.
In the benchmark's inaugural results, ZoomInfo's GTM.AI achieved a GTM Bench Index score of 77, substantially outpacing three competing systems—Apollo at 47, Exa at 36, and open-web search at 31. ZoomInfo's system completed 98% of the operator's requested work product and returned 478 verifiable records per 1,000 queries, compared to 7 to 35 records for other tested systems. The cost per task executed by ZoomInfo's system was $0.79, positioning it as both more effective and operationally efficient than alternatives evaluated.
Demonstrating transparency regarding its benchmark methodology, ZoomInfo acknowledged that the system is vendor-run and disclosed four categories where its competitive advantage is limited or absent: copywriting, owned CRM data that external tools cannot access, and other specialized applications. The company also noted that grounding measurements are calibrated against ZoomInfo's verified records rather than through independent third-party audit, a material distinction for institutional investors evaluating claimed performance advantages.
ZoomInfo indicated that Version 2 of the benchmark will expand to include agentic multi-step workflows, international coverage, and an owned-data axis. The company further stated that other data providers and AI agent builders can submit systems for testing and inclusion on the public leaderboard, positioning GTM Bench as an evolving competitive assessment tool rather than a closed proprietary framework.
The benchmark release occurs amid significant market headwinds for ZoomInfo. The company's stock has declined 72% over the past year to $2.94 per share, despite maintaining operational strength: gross profit margin of 86.75%, $1.25 billion in annual revenue, and continued profitability. The company is valued at $867 million. InvestingPro analysis indicates the stock is trading below intrinsic value, suggesting potential disconnect between market sentiment and underlying financial performance.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.