Skip to main content

Lead Scoring Analysis

Introduction to Lead Scoring

Lead scoring is a methodology used to rank prospects against a scale representing the perceived value each lead brings to the organization. The ultimate goal is to prioritize the marketing and sales team's focus on leads with the highest potential for conversion.

Using advanced analytics tools, you can predict which leads are most likely to convert into paying customers.

Example Using Customer Leads Dataset

In the following example, we draw from a Lead Scoring Dataset on Kaggle.

For the purpose of this analysis, we will name the dataset customer_leads. This dataset comprises various features such as lead origin, tags, and last notable activity, among others.

You can download the corresponding raw dataset here.

SQL-inf Query for Lead Scoring

SELECT * FROM Customer_Leads PREDICT(Converted, ignore=Prospect_ID)

The query selects all columns, creating a predictive model targeting the Converted column. The model learns from existing lead conversion data while ignoring the Prospect_ID, which doesn't contribute to the prediction.

Key Findings

Our results can be summarized in four key categories:

  • Model Accuracy: The model has a stellar accuracy of 96%, making it an exceptionally reliable tool for predicting lead conversions.

  • Feature Importance: The top three features responsible for 50% of the prediction accuracy are Lead Origin, Tags, and Last Notable Activity.

  • Highlights:

    1. Lead Origin: The majority of leads originate from Landing Page Submission (53%) and API (39%). Lead Add Form, although accounting for just 8% of leads, has the highest median conversion rate.
    2. Tags: The tag "Will revert after reading the email" leads with a conversion rate of 35.2%, emphasizing the effectiveness of follow-up emails. Conversely, tags like "Interested in other courses" and "Already a student" have low conversion rates.
    3. Last Notable Activity: 'SMS Sent' has the highest median conversion rate, even though the most common last activities are 'Modified' and 'Email Opened'.

Strategic Conclusions and Recommendations

  • Lead Origin Efficacy: While Landing Page Submission and API generate the most leads, Lead Add Form and Lead Import demonstrate high conversion rates. Hence, these methods should be further explored and possibly prioritized.

  • Tagging Strategy: Tags like "Will revert after reading the email" and "Ringing" are especially effective in lead conversion. Marketing initiatives, such as follow-up emails and calls, should be tailored around these effective tags.

  • Communication Channels: The 'SMS Sent' category shows a strong conversion rate, suggesting that SMS might be a more effective communication channel than emails or other methods for this particular dataset. This could inform future engagement strategies.

  • Addressing Low Conversion: Tags indicating interest in other courses or prior student status have low conversion rates. Customized nurturing strategies could be developed to better serve these particular segments.

Implementing these insights can help refine your lead scoring process, allowing for more effective resource allocation, better-targeted marketing campaigns, and ultimately, higher conversion rates.

Advanced View Creation for Lead Scoring

Here we will give examples of views that could be used lead scoring for different scenarios.

Conversion Prediction Based on Web Activity and Profile Data

Assumes tables: leads, web_activity, profiles

SELECT l.lead_id,
       w.page_views,
       w.click_through_rate,
       p.industry,
       p.job_role,
       l.converted
FROM leads l
JOIN web_activity w ON l.lead_id = w.lead_id
JOIN profiles p ON l.lead_id = p.lead_id;

Conversion Prediction Based on Email Engagement

Assumes tables: leads, email_activity

SELECT l.lead_id,
       e.emails_opened,
       e.clicks,
       l.converted
FROM leads l
JOIN email_activity e ON l.lead_id = e.lead_id;

Conversion Prediction Based on Previous Purchases

Assumes tables: leads, previous_purchases

SELECT l.lead_id,
       p.purchase_count,
       p.total_spent,
       l.converted
FROM leads l
JOIN previous_purchases p ON l.lead_id = p.lead_id;

Conversion Prediction Based on Event Attendance

Assumes tables: leads, events

SELECT l.lead_id,
       e.events_attended,
       e.interaction_score,
       l.converted
FROM leads l
JOIN events e ON l.lead_id = e.lead_id;

Conversion Prediction Based on Customer Service Interactions

Assumes tables: leads, customer_service_interactions

SELECT l.lead_id,
       c.tickets_raised,
       c.issue_resolution_time,
       l.converted
FROM leads l
JOIN customer_service_interactions c ON l.lead_id = c.lead_id;

Each of these SQL queries aims to join different aspects of lead interaction and behavior with the 'Converted' metric, providing a holistic view for predictive analytics.