← All work

Client engagement / Six6 Marketing AI Intelligence Layer

From a website visitor to a qualified sales conversation.

A two-stage qualification and audit system that turns a prospect’s answers into an explainable score and a structured CRM lead.

My role

Solo design, full-stack build & CRM integration

Project status

Deployed to client site

Six6 Marketing’s brand card: Digital Authority for Service Businesses
Six6 Marketing · Client brand image; explore the live project for the product interface

The result you can inspect

The completed audit carries the prospect’s answers and score into the sales workflow.

System overview

How the pieces connect.

  1. 01

    Qualify

    Guided questions establish fit.

  2. 02

    Score

    Rules calculate an explainable audit score.

  3. 03

    Hand over

    Answers and score become a CRM lead.

A simplified view of the documented architecture.

The short version

Context before code.

The problem

Six6 needed their own site to do two jobs at once: filter site visitors down to genuine prospects before spending sales time on them, and prove — live, on the spot — the pitch they run in market, that a website with real reasoning intelligence built in beats a static brochure site. A generic contact form does neither; it doesn't qualify, and it doesn't demonstrate anything.

What I built

Two sequential layers rather than one form. Stage one is a qualifier: a short set of guided questions determines whether a visitor is worth taking further. Visitors who qualify move into stage two, a much more detailed audit that interrogates their current business and marketing efforts and produces a structured score with specific recommended fixes. Completing the audit writes the prospect into Six6's CRM and lists them as a lead in the database, ready for the sales team to follow up. Next.js on the front end, FastAPI on the backend, deployed on Vercel and served live on Six6's own domain.

For a closer look

The decisions behind it.

Architecture & data flow

A two-stage funnel: the qualifier is the gate, and only visitors who pass it reach the heavier, more detailed audit — so the expensive part of the flow only runs on visitors worth the effort. The audit's scoring is deterministic, computed from the visitor's answers rather than generated freeform by a model, matching Six6's own public positioning that it's "guided questions, not a generic chatbot." On completion, the scored result and the prospect's answers are written into Six6's CRM as a structured lead, so the sales team works from a qualified, scored record rather than a raw enquiry.

Constraints & tradeoffs

The audit needed to be explainable and consistent, not a black-box model verdict, so scoring runs on deterministic logic rather than free-form generation. Effort had to be gated: the detailed audit is heavier than a form, so the qualifier stage exists specifically to filter out visitors who aren't a fit before they reach it. And because the site's whole positioning rests on being the reasoning-driven alternative to a static site, the flow itself had to hold up as the live proof of that claim, not just describe it in copy.

Tools used on this project

Continue the conversation

Working on something similar?

Explore it with Jay ↗Talk to Justin ↗
Next project FanBrain AI
top