How We Helped a B2B Company 3x Their Qualified Leads With AI Marketing

In Q1 of last year, a B2B technology company came to Diztaly with a familiar challenge: their marketing team was generating plenty of leads, but conversion rates were stagnant, sales was complaining about lead quality, and cost-per-acquisition was climbing quarter-over-quarter. Twelve months later, qualified pipeline had tripled, marketing-attributed revenue had grown 180%, and the cost of acquiring an enterprise customer had dropped by 42%. This is the story of how we did it.

The Client and Their Challenge

Our client — a B2B SaaS company offering supply chain management software — had built a solid inbound marketing engine over five years. Their website attracted 40,000 monthly visitors, their content ranked for relevant keywords, and their sales team was actively working 300+ opportunities per quarter. The problem was not top-of-funnel volume. It was conversion efficiency.

Marketing was handing over 450 MQLs per month. Sales was accepting fewer than 100 as SQLs — a 22% acceptance rate. Of those 100 SQLs, only 18 were converting to closed deals. Marketing was spending considerable budget generating leads that sales did not trust and prospects were not ready to buy. Something had to change.

Phase 1: Data Unification and Diagnostics (Month 1–2)

Before deploying any AI capabilities, Diztaly began with a comprehensive data audit. We discovered that the client's customer data was fragmented across four systems: a Salesforce CRM, a Marketo marketing automation platform, a Mixpanel product analytics tool, and a legacy customer support platform. These systems shared no unified customer identifier, meaning that a prospect's full journey — from first website visit to closed deal — could not be tracked in a single view.

We resolved this in six weeks by implementing a Customer Data Platform (CDP) that unified all four data sources using a probabilistic identity resolution model. For the first time, the marketing and sales teams could see a complete picture of every prospect's journey, including which content they consumed, which features they explored in trial, what questions they asked support, and how long they spent on pricing pages.

Phase 2: AI Lead Scoring Implementation (Month 3)

With clean, unified data available, we built a machine learning lead scoring model trained on 18 months of historical closed/won and closed/lost data. The model incorporated 47 behavioural and firmographic signals, including:

• Time spent on ROI calculator pages (strongest positive predictor)
• Number of product pages visited in the first session
• Company size, industry vertical, and technology stack (from enrichment data)
• Email engagement depth (not just open rates, but scroll depth and time spent)
• Frequency and recency of website visits in the 30 days before scoring
• Trial activation depth (for prospects who had started a free trial)

The model assigned each lead a score from 0–100 and predicted probability-of-conversion. We then worked with the sales team to establish a score threshold — leads above 72 would be handed to sales immediately; leads between 40–72 would enter an AI-personalised nurture sequence; leads below 40 would receive long-cycle educational content until their score improved.

Phase 3: AI-Personalised Nurture Sequences (Month 4–5)

For the 60% of leads that fell into the middle-tier nurture category, we built an AI-driven nurture programme that adapted content and cadence based on each prospect's demonstrated interests and engagement patterns. Rather than sending the same sequence to everyone, the system selected from a library of 40+ content assets — choosing the asset most aligned to each prospect's stage and interest signals.

We also deployed an AI email optimisation layer that tested subject lines, send times, and content formats for each individual, learning optimal parameters over the first four to six touchpoints and then applying those learnings to every subsequent communication. Average email open rates in the nurture sequence went from 18% to 34% within 60 days.

Phase 4: Conversational AI for Trial Users (Month 6)

A significant insight from the data unification phase was that prospects who activated at least three features during their free trial converted to paid customers at 4.7x the rate of those who activated fewer features. Yet 62% of trial users were abandoning without reaching the third feature activation.

We deployed a conversational AI assistant within the trial product experience — not a generic chatbot, but an AI that knew each user's activation history and proactively guided them toward the features most correlated with conversion. The assistant surfaced at contextually relevant moments, offered personalised help, and connected users who were clearly high-intent with a human customer success manager.

Third feature activation rates increased from 38% to 71% within 90 days.

The Results: 12 Months Later

Key Lessons From This Engagement

Looking back at the twelve months of this transformation, four lessons stand out. First, data unification was the most valuable work we did — none of the AI capabilities would have functioned without clean, connected data. Second, the biggest gains came not from generating more leads but from converting existing leads more effectively — this is true in most B2B organisations. Third, AI in the product experience (trial AI assistant) produced a higher ROI than AI in any marketing channel — a reminder that customer success and product adoption are marketing problems. Fourth, the sales team's trust in marketing-generated leads changed the entire commercial culture of the business — when sales believes in the leads they receive, they invest more energy in conversion, and the whole engine accelerates.