Owning the Pipeline Number: Forecasting and Demand Gen at Scale

How a channel-level forecast model, AI-assisted lead scoring, and tight sales alignment tripled MQL-to-SQL conversion and grew lead quality.

The Challenge

In 2024, I took ownership of Dremio's marketing pipeline number before holding a formal demand generation title. The business was shifting from aggressive spend to efficiency: budgets were tightening, and marketing needed to do more with less. I was handed a quarterly MQL target and asked to build the system to hit it consistently.

The problem wasn't volume. FY24 produced nearly 20,000 MQLs. The problem was that most of them weren't converting. With an MQL-to-SQL rate of 2.2%, marketing was generating a lot of activity but not a lot of pipeline that sales could actually work.

The Approach

I started by working with the finance team to pull historical conversion data across the funnel and build a channel-level forecast model. This gave us a week-over-week view of what we needed from each channel to hit the quarterly target, rather than managing to a single monthly or quarterly number with no visibility into where we stood.

From there, the work became about continuous refinement. Quarterly MQL scoring reviews helped us filter out noise and improve what was reaching the BDR queue. Every six months we did a deeper audit. I worked closely with my BDR manager counterpart to go through lists together, reviewing what had and hadn't converted, and feeding that feedback directly back into the scoring model.

As AI tooling became available, I layered an AI forecast model on top of the manual one. Running them in parallel helped surface leads that the hand model might miss, and the feedback loop from the BDR team improved the AI model's precision over time. The result was a program we called "no lead left behind."

On the channel side, I mapped closed-won deals against prior marketing engagement and reallocated spend toward what was actually driving wins. Email became a priority given it was a free channel with strong influence. We also invested in optimizing the Contact Us CTA across the site, adding it to more pages, testing timing and offers, and adding a "how did you hear about us" field to forms to sharpen attribution.

Results

• 2.2% → 7.9% MQL-to-SQL conversion rate, FY2024 to FY2026

• +40% more SQLs in FY2025 vs FY2024, on less than half the MQL volume

• 89% of closed-won deals had a marketing touch in the prior 12 months

The volume story looks counterintuitive at first. MQLs dropped from 19,988 in FY2024 to 9,049 in FY2025 and 7,329 in FY2026. But FY2024's volume was inflated by low-quality leads that weren't converting. Once scoring tightened, the numbers that mattered moved in the right direction: FY2025 produced more SQLs (619 vs 443) and more Sales Accepted Opportunities (316 vs 238) than FY2024 did on more than double the MQL volume. FY2026 FQ3 became the best single quarter on record at a 10.2% SQL rate.

On the channel side, email grew 435% in pipeline influence and became the top performing channel. Events were the strongest driver in EMEA. Together, these shifts meant the team was generating less noise and more of what sales could actually close.

Key Takeaways

Pipeline ownership is only meaningful if it's paired with a clear system for improving what goes into that pipeline. The forecast model created visibility. The scoring refinement created quality. The sales collaboration created trust. Together they shifted marketing from a lead-delivery function into a genuine revenue partner — with fewer leads, more SQLs, and the numbers to prove the difference.