Case Study
Tech SaaS
How a B2B SaaS Platform Stopped a Leaky Funnel and Grew Conversions by 22%
Veloxa had an active lead generation engine but a conversion problem they couldn't locate. Lead volume was growing, acquisition costs were rising, and the data to explain why was scattered across a CRM that nobody fully trusted. They needed to understand where deals were dying before they could fix it.
$1.2M
Pipeline recovered
+22%
Conversion lift
-40%
CPL reduction
We're a tech company helping B2B teams extract CRM data, find revenue leaks, and unlock growth. Our approach is simple, combine AI with strategy so you can focus on closing what matters most.
The situation
Veloxa had been growing fast for two years. Sales team of 12 reps, a HubSpot instance with 40,000+ records, and a lead generation program that was producing volume. The problem was that volume wasn't translating to revenue at the rate leadership expected. Acquisition costs were climbing. Conversion from opportunity to customer had plateaued at 18%. The VP of Revenue suspected there was a funnel problem but couldn't locate it in the dashboard.
What made it harder: each team had a theory. Marketing thought the issue was lead quality. Sales thought it was the product demo. CS thought some deals were closing before they should. Everyone was looking at their own slice and nobody had the cross-functional view.
What we found
We extracted the full deal history from HubSpot and ran a stage-by-stage conversion analysis. The problem wasn't distributed evenly across the funnel. It was concentrated at two specific points: the transition from Discovery to Proposal, where 34% of deals were stalling for more than 21 days, and the transition from Proposal to Negotiation, where 28% of deals were being marked closed-lost with no activity in the preceding two weeks.
The stalls at Discovery-to-Proposal correlated strongly with a single factor: deals where the rep had engaged fewer than three contacts at the account. Multi-threading wasn't a process problem. It was a data pattern that nobody had connected to conversion rates. When we cross-referenced activity data with outcomes, the relationship was unambiguous.
The Proposal-to-Negotiation losses were a different issue. Many of those deals had proposals sent but no follow-up logged within 5 days of sending. The deals weren't lost to competition. They were lost to silence.
On the lead quality question, the data told a different story than marketing expected. CPL was high because the lead sources generating the most volume had the lowest close rates. Two sources with significantly lower volume were producing 60% of closed-won revenue. The budget allocation had never been updated to reflect that.
What changed
The multi-threading finding drove an immediate process change: reps were required to log at least three unique contacts before advancing a deal to Proposal. Within 60 days, average contacts per deal went from 1.8 to 3.1 and the Discovery-to-Proposal stall rate dropped by 40%.
The follow-up gap led to an automated 72-hour task trigger after any proposal was sent. Not a sequence — a task for the rep to make direct contact. The Proposal-to-Negotiation loss rate dropped 22% in the first quarter.
The channel reallocation moved 35% of paid budget from the high-volume, low-conversion sources to the two high-performing ones. CPL went up on paper (fewer cheap leads) but cost per closed deal dropped 40%.
Twelve months later, opportunity-to-customer conversion was 22% versus the 18% baseline. The pipeline that had been sitting stalled was worked through, recovering $1.2M in deals that had been written off.