Quality scores

EngineRoom's Quality Score is a Predictive Machine Learning model that assigns a score from 1-10 to each incoming lead for your business, extensively trained on a historical dataset with various User Engagement metrics including but not limited to:

• Number of Pages Viewed

• Session Duration

• Number of Sessions

• Number of Clicks

• Scroll Depth Percentage

• Session Attribution (Source/Medium)

Thirty-five input variables were tested for correlations to whether a “sales conversion” occurred.

The top ~5 variables were selected based on discriminatory power, and each bucketed into intuitive, distinctive segments and converted into binary variables. This was done via a process called One Hot Encoding that converts categorical variables into numeric variables so they can be used in Machine Learning models. This method is also used with numeric categories (i.e. age groups) to capture any non-linear relationships that might exist.

This approach ensures our model is finely tuned to predict lead conversion with high precision, utilising historical conversion data to pinpoint the most promising leads based on similar patterns observed in past successful conversions.

Note: This is only available to customer's with active sales cycle data.

How the Score is Calculated

• Each of the selected model variables above is assigned a unique coefficient that represents the correlation to a sales conversion

• For a specific lead, the coefficients and intercepts are entered into the below equation, to produce a linear combination and then transformed to calculate a predicted probability of conversion:

$$p = \frac{1}{1 + e^{-\displaystyle(\beta_0 + \beta_1 x_1 + \beta_2 x_2 + \ldots + \beta_n x_n)}}$$

• The output “p” is a probability between 0 and 1, with higher probabilities corresponding to leads of higher quality

• This probability is then converted into a lead quality score that lies between 1 and 10

• The lowest quality leads receive a score of 1, while the highest quality leads receive a score of 10

• The remaining leads are normalised accordingly

Results

The model has been applied on all leads over the past year, demonstrating exceptional ability to differentiate between leads that convert into sales vs those that do not.

Lead Quality	Number of Leads	Number of Sales Conversions	Sales Conversion Rate
Low (Score of 1-3)	5,050	200	3.9%
Medium (Score of 4-6)	2,900	170	5.7%
High (Score of 7-10)	2,270	300	13.3%

• The top 25% of leads identified by the Lead Quality model captures almost 50% of all sales conversions

• Remarkably, 13.3% of the leads receiving a quality score greater than 8 resulted in sales conversions

• This contrasts with the 3.9% conversion rate for leads scored between 1 and 3