GreenLab: headless SaaS site with AI lead qualification

GreenLab provides a molecular modelling SaaS platform used by pharmaceutical research laboratories. The sales cycle is long (8 to 14 months), with an average deal value of €84,000. Their problem: the "Request a demo" form receives 380 leads per month, but 92 of these are PhD students, job applicants in disguise, or competitors fishing for information. The sales team (3 people) spends 22% of its time manually qualifying leads.

The original WordPress corporate website had a Lighthouse score of 31, which was dragging down their Google Ads performance (Quality Score 4/10). Brief: a performance-focused redesign, a smart lead qualification form, and much tighter sales monitoring.

Headless architecture: WordPress 6.9 + ACF Pro for content management (self-managed marketing team), exposed via GraphQL using WPGraphQL, and consumed by a Next.js 15 frontend deployed on Vercel Edge (latency < 80 ms worldwide).

Demo form with a two-step AI qualification process. Step 1: standard fields + a free-text box labelled "Describe your project". Step 2: the text is sent to an OpenAI GPT-4 agent with a system prompt containing: qualifications (lab size, type of molecules studied, urgency, budget mentioned), red flags (student, thesis, competitor comparison), and a 0–100 scoring grid.

Score > 70: lead routed directly to the most relevant sales representative with a pre-written AI brief. Score 40–70: automated HubSpot nurturing. Score < 40: polite automated response + public resources. Full transparency for the sales team regarding the AI’s reasoning (each score explained).

Weekly reporting: AI quality assessed retrospectively (do leads with a score of over 70 turn into opportunities?). The system prompt is adjusted every two weeks based on false positives.