Mastering Lead Scoring: A Comprehensive Guide to Marketing Qualified Leads (MQL)
- John Brown
- Oct 1
- 3 min read
Updated: Nov 4
When it comes to CRM and marketing automation, one of the biggest challenges is separating curious browsers from conversion-ready prospects. That’s where lead scoring comes in. By defining what makes a Marketing Qualified Lead (MQL) and applying consistent scoring rules, marketing and sales teams can focus on the people most likely to convert.
In this post, we’ll show you how to:
Define a clear MQL rule (valid email + at least one open event).
Write SQL queries (Postgres + BigQuery) to score leads in your database.
Use a Python script to score and aggregate leads directly from CSV exports.
Break down MQL performance by UTM source and lead creation date.
This is a step-by-step, practical guide you can drop into your workflow today.
What Is a Marketing Qualified Lead (MQL)?
An MQL is a lead that’s more likely to become a customer based on engagement signals. While definitions vary across organizations, our simple rule is:
✅ The lead’s email is valid (i.e., not bounced).
✅ The lead has at least one email open event recorded.
This makes the scoring model straightforward, measurable, and easy to implement across CRM systems.
Why Break Down by UTM Source and Creation Date?
If you’re spending money on multiple marketing channels, you need to know:
Which UTM sources (Google Ads, Facebook, Newsletter, etc.) deliver the highest-quality leads.
How lead quality changes over time — for example, did a new campaign last month drive more MQLs than the month before?
By grouping MQL data by UTM source and lead creation date, you gain visibility into campaign effectiveness and ROI.
SQL for Lead Scoring
If you have a relational database, you can identify MQLs and roll up stats with a single query. Below are two versions: one for Postgres, one for BigQuery.
Postgres Example
```sql
SELECT
lead_id,
email,
COUNT(open_event) AS open_events
FROM
leads
WHERE
email IS NOT NULL
GROUP BY
lead_id
HAVING
COUNT(open_event) > 0;
```
BigQuery Example
```sql
SELECT
lead_id,
email,
COUNT(open_event) AS open_events
FROM
`project.dataset.leads`
WHERE
email IS NOT NULL
GROUP BY
lead_id
HAVING
COUNT(open_event) > 0;
```
Python Script for CSV Data
If you export your leads and events as CSVs, a lightweight Python (pandas) script works perfectly.
Run it locally to score each lead and generate two reports:
lead_scores.csv – every lead with its MQL score.
mql_breakdown.csv – grouped by UTM source and created date.
👉 See the full script here: score_leads.py
Example Input & Output
Leads CSV
| lead_id | email |
|---------|----------------|
| 1 | example1@test.com |
| 2 | example2@test.com |
Events CSV
| event_id | lead_id | event_type | timestamp |
|----------|---------|------------|----------------------|
| 1 | 1 | open | 2025-07-01 10:00:00 |
| 2 | 2 | open | 2025-07-01 11:00:00 |
Breakdown Output
utm_source | created_date | leads | mqls | mql_rate | total_score |
2025-07-01 | 1 | 1 | 1.0 | 100 | |
2025-07-01 | 1 | 0 | 0.0 | 0 | |
newsletter | 2025-07-02 | 1 | 1 | 1.0 | 100 |
Why This Approach Works
Simplicity: Only two conditions define MQLs.
Flexibility: You can add more signals later (clicks, form fills, webinar signups).
Attribution-friendly: Breaking down by UTM source tells you which campaigns actually deliver quality leads.
Actionable: You can push these scores into your CRM or marketing automation platform for nurturing workflows.
Next Steps
Add scoring for email clicks, form completions, or demo requests.
Normalize UTM values (e.g., fb → facebook).
Automate nightly scoring with Airflow, dbt, or Prefect.
Build dashboards in Looker, Tableau, or Metabase.
Final Thoughts
Lead scoring doesn’t have to be complicated. By defining a simple MQL rule and tracking performance by UTM source and creation date, you create a scalable foundation for better marketing attribution and smarter sales handoffs.
If you’d like to try it, the full repo is here: GitHub: Lead MQL Scoring
Comments