solix-backend

Lead intent and MQL methodology

How the website works out what each visitor is interested in, tags every lead to a product line, and decides when a lead is an MQL. The code lives in scoring.py (the model) and intent.py (tracking and lead rollup).

1. What we capture

Only after the visitor clicks Accept analytics in the cookie banner. Visitors who choose Essential only, or whose browser sends Global Privacy Control / Do Not Track, are never tracked. Their forms still work and still create leads, just without browsing history. Nobody’s chat text is stored as a tracking event, only which product a question was about.

Signal	When it fires	Points per product
Product page view	Opening /products/<product> (and /platform, /services-support)	3
Engaged reading	45+ seconds of active reading on a topic page	5
Deep scroll	Reading 75%+ of a topic page	2
Resource opened	Opening a blog, white paper, datasheet, etc. tagged with products	6
Gated download	Unlocking or downloading gated material	15
Concierge question	Asking Sol about a product	6
Pricing question	Asking about price or cost	10
CTA click	Clicking demo / trial / contact from a topic page	8
Resource search	Searching Resources for a product	2

Rules that keep the numbers honest:

• The browser never sends points. It reports what happened; the server applies the weights above.
• A signal scores once per visit per page/product, so reloads and repeat clicks don’t inflate intent.
• Points decay with a half-life (default 30 days): interest from last quarter counts for much less than interest from this week.
• Campaign data (UTM tags, referrer, landing page) is captured from the first page of the visit.

2. From visitor to lead

A visitor becomes a lead the moment they identify themselves through any form: demo, contact, gated download, newsletter, partner application, Solix ECS trial sign-up, or a demo booked in the concierge chat. (Job applications never create leads.)

• There is one lead per email address. Every form, visit and device the person uses is merged into it.
• Their earlier anonymous browsing is attached, and later browsing keeps adding to the same lead.
• Form submissions add points to the product the person chose or was reading about:

Form	Points
Demo request	30
Trial sign-up	25
Contact sales	15
Gated download	15
Newsletter / partner	5

3. Tagging to a product line

Product scores roll up into six product lines. The line with the most intent is the lead’s primary product line, and the top product within it is the primary product.

Product line	Products
Enterprise Edition & Common Data Platform	Enterprise Edition, Common Data Platform, Enterprise Data Lake
Archiving & Application Retirement	Enterprise Archiving, Application Retirement, Data Preservation, SAP, Oracle/OEBS, Mainframe, Email, File and Database Archiving, Active Archiving for Compliance
Data Governance, Privacy & eDiscovery	Enterprise Data Governance, Consumer Data Privacy, eDiscovery
Enterprise AI	Enterprise AI, Data Sense, Data Ask, Application Knowledge Graph, AI Warehouse, Agentic, AI Governance, AI Healthcare, EAI Pharma
Enterprise Content Services (ECS)	Enterprise Content Services
Professional & Managed Services	Services & Support

Content published from the admin carries product tags, so every new blog or datasheet feeds this model automatically.

4. Fit (who they are), max 40

Signal	Points
Business email (not Gmail, Outlook, etc.)	10
Director, Head, VP, C-level, founder	15
Manager, lead, architect	8
Company size 1,000+ / 200+ / 50+	10 / 6 / 3
Phone number given	5

Lead score = intent in the primary product line + fit.

5. Stages

Stage	Set by	Meaning
Lead	System	Known contact, not yet qualified
MQL	System	Asked for a demo, trial or contact (hand-raiser), or lead score reached the MQL threshold (default 45)
SAL	Sales	A rep accepted it
SQL	Sales	Budget, need and timing confirmed
Opportunity	Sales	Active deal
Won / Lost / Disqualified	Sales	Outcome

The system only ever moves leads from Lead to MQL. Stages set by people are never overwritten. Every change is recorded in the lead’s history with who made it and why.

The date a lead became an MQL, and the product line it had at that moment, are kept, so “MQLs by product line” in reports never shifts after the fact.

6. Tuning

In Admin → Settings → Lead scoring an admin can change:

• the MQL threshold;
• the interest half-life.

Saving rescores every lead immediately. Scores also decay automatically every 6 hours. Event and form weights live in scoring.py, so changing them is a code change and gets reviewed.

7. Reporting and access

• Dashboard: the leadership view. It shows visitors, leads, MQLs, SQLs and conversion rates, a daily trend, results by product line, the funnel, lead sources (first-touch channel), product interest, owners, top content, and country/industry. It filters by date range and product line, and every number links to the matching leads.
• Leads: filter by product line, product, stage, owner, source, country, industry, score, date or recent activity. Save a filter set as a shared view. Assign owners or change stages in bulk. Export to CSV or Excel; the Excel file includes a per-product-line summary sheet.
• Roles:

Role	Can do
Admin	Everything
Sales	Work leads
Content editor	Publish content
Leadership (viewer)	Read-only dashboard, leads and exports

8. Data retention

Raw behaviour events are deleted automatically after 180 days, which keeps the free MongoDB tier small. Lead records, their scores and stage history are kept.