A closer look at predictive lead scoring in Dynamics 365 Sales Insights

Imagine this: you have thousands of leads, and only so many sellers. A seller can only work through a set number of leads. Your sellers might not get through all of the leads on their list. You risk wasting time on the low quality leads, and never get to the high quality leads. If sales time is wasted with low quality leads, your company might suffer. Pursuing high quality leads can increase sales productivity and guarantee a better return for your business.

Traditional lead scoring

Historically companies has made many attempts at creating a successful (manual) lead scoring models, using historical data and intuition. Traditional (manual) lead scoring has several potential downsides to it; in traditional lead scoring, you have to update your scoring system manually to properly weight user interactions, not to mention the fact that intuition can prove wrong.

Predictive lead scoring

Predictive lead scoring uses a predictive machine learning model to calculate a score for open leads based on historical data. The score helps sellers prioritize leads and achieve higher lead qualification rates, and reduces the time it takes to qualify a lead.

For example, say you have two leads, Lead A and Lead B, in your pipeline. The lead scoring model calculates a score of 80 for Lead A and 50 for Lead B. Based on the scores, you can predict that Lead A has a greater chance of being converted into an opportunity. Further, you can review the top influencing factors to analyze why Lead B’s score is low and decide whether to improve it.

The following image shows an example of a lead scoring widget

Prerequisites

To use predictive lead scoring in Dynamics 365 Sales, you must enable Advanced Sales Insights features.

Open the App Settings area in Sales, and then in the sitemap navigate to Get started under Digital sales to open the Get started with digital sales page – see more here

On the Get started with digital sales page scroll to the bottom of the page where you will find the Lead and opportunity scoring section

In the Lead and opportunity scoring section note the Quick setup button  

Click the Quick setup button to open the Lead and opportunity scoring quick setup task pane.

The Lead and opportunity scoring quick setup task pane will display an alert if the app is unable to train your model – meaning you don’t have at least 40 qualified and 40 disqualified leads from the past two years

If that indeed is the case, make sure you have at least 40 qualified and 40 disqualified leads (e.g., import a bunch of leads and qualify/disqualify them.) I would recommend you ensure that each lead contains data for lead source, industry, job title, no. of employees, annual revenue, or alike – for the model to have something to work with.

Creating the model

When you have created enough qualified and disqualified leads the Lead and Opportunity scoring quick setup task pane will surface a Create and publish button

Click the Create and publish button to start training the model.

The Lead and Opportunity scoring quick setup task pane will display a message saying “Your model training is in progress”

The training of the model takes a while. Return to the Lead and Opportunity scoring quick setup task pane until your model is ready

When the model is trained, the Lead and Opportunity scoring quick setup will display a message saying “Your model training is successfully trained…”

Click the Go to advanced lead settings link to navigate to the Predictive lead scoring screen

On the Predictive lead scoring screen you will see details about the model

• Attributes used
• Training with leads from the past
• (and more)

Note that the Retrain automatically toogle is ON

And the Edit model button is disabled

Edit and retrain the model

For the first run, our model uses auto-feature selection. There is no way to control which attributes is used in the initial training of the model. Most likely the initial training of the model made use of one or attributes you don’t want it to use. But luckily you can edit the model to use the attributes of your choice.

To enable the Edit model button, flip the Retrain automatically button to Off

Click the Edit model button to open the Predictive lead scoring > Edit fields screen

On the Predictive lead scoring > Edit fields screen click Leads to expand the Leads area

In the Leads area you can select the attributes you want the model to consider (the round question mark icons to the left serves as check boxes). In my example, I have selected seven attributes I want to (re-)train the model with.

In the Leads area you can also click an attribute to open the Attributes insights task pane

In the Attributes insights task pane you can learn about the significance of the values of the selected attribute and more

When you have selected the attributes you want the model to include, click the Retrain model button to retrain the model

When the training has completed, navigate to Leads in the Sales area, and then open the My open leads scored view to see Score, Trend, and Grade for each open lead.

Click a lead to see the lead scoring widget (make sure to use the Sales Insights form)

You are now able to pursue your high quality leads, increase sales productivity and realize a better return for your business.

Note

• You can leverage the predictive lead scores in segments in Dynamics 365 Marketing (Real Time)
• You can leverage the predictive lead scores for lead routing purposes in Dynamics 365 Sales (Sales Accelerator)

See also

• Install and administer Sales Insights features – link
• Edit and retrain a lead scoring model – link