Dynamics 365 lead scoring: predictive scores in Sales, scoring models in Customer Insights - Journeys
Dynamics 365 scores leads in two apps. Here is what each one needs, how to set it up step by step, and what changed when outbound marketing was removed.
Dynamics 365 lead scoring comes in two forms. Dynamics 365 Sales has predictive lead scoring, a model trained on your past qualified and disqualified leads that gives each open lead a score from 0 to 100 and a grade from A to D. Dynamics 365 Customer Insights - Journeys has lead scoring models, where you set the conditions and the points yourself.
This guide covers both, plus a simple rules-based score built with Dataverse columns and Power Automate. Every product detail comes from Microsoft Learn, linked where it is used.
Dynamics 365 lead scoring at a glance
| Question | Predictive lead scoring (Sales) | Lead scoring models (Customer Insights - Journeys) |
|---|---|---|
| Where is it? | Sales Insights settings, Predictive models, Lead scoring | Lead Management, Scoring Models |
| Who sets the logic? | A model trained on your closed leads | You: conditions and points |
| License | Sales Premium, Sales Insights, or Sales Enterprise (1,500 scored records per month) | Not stated on the pages used here. Check your environment |
| What it scores | Open leads that match the model’s filter | Leads, optionally counting parent contact interactions |
| Rule types | None. You choose which attributes the model may use | Attributes (fit) and interactions (engagement), positive or negative points |
| Data needed | 40 qualified and 40 disqualified leads | None |
| Output | Score 0 to 100, grade A to D, score trend | Points, plus grades you define |
| Decay | Not applicable | None documented. Interactions take a timeframe |
| Refresh | New leads in near real time, updated leads every 24 hours | Every two hours |
| How reps see it | Lead score widget, My Open Leads Scored view | Scores tab on the model. For the lead form, check your environment |
Use the predictive score when you have steady lead volume and reps who close leads consistently. Use a scoring model when you want to own the criteria or are under the 40 and 40 minimum. The two can run together.
Before you start: plan the model
Deciding what to score is the slow part, and it is easier on paper. Write down:
- Three to six fit factors, such as company size, industry, job level, and country, taken from what your won deals have in common. Your ideal customer profile is the source, and our five-step framework for building an ICP scoring model covers the method.
- A short list of engagement signals, such as form submissions, event registrations, and email clicks. Go easy on email opens: Litmus reports that over 50% of email opens come from devices with Apple Mail Privacy Protection.
- Disqualifiers that subtract, such as competitors, students, and personal email domains.
- Cutoffs. Decide what score sends a lead to sales before the builder asks.
Then check your data. An empty attribute earns no points.
How to set up a scoring model in Customer Insights - Journeys
These steps follow Microsoft’s page on creating lead scoring models.
Step 1: Turn the feature on
An administrator goes to Settings, Overview, Feature switches, enables Lead Management, and selects Save. Two items then appear in the navigation: Scoring models and Qualification.
Step 2: Create the model
Go to Lead Management, then Scoring Models, select New, name the model, and select Create.
Step 3: Build conditions and assign points
The right side of the builder lists two kinds of data:
- Attributes are demographic information from the lead, or from related tables such as contacts and accounts.
- Interactions are engagements or activities. Microsoft’s example is Emails opened.
Select one and it becomes a condition. Set the points, which Microsoft says “can be negative or positive”. For an interaction, select Edit to set a volume (at least five emails) and a timeframe (in the last 28 days). Select Create new group to add more. Microsoft does not spell out how several conditions inside one group combine, so test a lead you know.
Step 4: Add grades (optional)
The second option on the right-hand bar opens grades. Each grade has a name, an operator (less than, more than, or between), and a score range. Microsoft notes that “you can publish the model without defining any grades.”
Step 5: Review the settings
The third option opens the model settings. A model inside a business unit scores only that unit’s leads. Select Count parent contact interactions if the model should count activity on the lead’s parent contact as well as the lead.
Step 6: Publish and check the results
Select Publish. After publishing you can view the conditions but not modify them, so test first. Microsoft’s page on visualizing results describes two tabs: Insights, with leads scored, the score for the top 10% of leads, the distribution, and leads per grade, and Scores, which lists every scored lead.
Step 7: Set qualification criteria
Per Microsoft’s page on qualifying leads, go to Lead management, Qualification, Qualification criteria. Add a live scoring model and the score that qualifies a lead. You can add several models and require all of them or any one. Then choose the action: set Status Reason to Marketing Qualified, or Sales-Ready to Yes. Select Publish.
Two documented behaviors matter: only leads in an open status can be marketing-qualified, and a Marketing Qualified lead stays that way if its score later falls below the threshold.
An example model in Customer Insights - Journeys
Demo data. A B2B software company sells to operations teams at companies with 200 to 1,000 employees. It builds two models so qualification can require both. Interaction names vary, so use the standard ones listed in your builder.
Model 1: Fit (attributes)
| Group | Condition | Points |
|---|---|---|
| Company size | Account has 200 to 1,000 employees | +30 |
| Account has 50 to 199 employees | +15 | |
| Industry | Account is in a core industry | +25 |
| Role | Job title contains Director or VP | +25 |
| Region | Country is in a served market | +20 |
| Disqualifiers | Email domain is a competitor | -100 |
| Email is a personal domain | -15 |
Model 2: Engagement (interactions)
| Group | Condition | Points |
|---|---|---|
| Buying intent | Submitted the demo form in the last 30 days | +40 |
| Events | Registered for an event in the last 60 days | +20 |
| Clicked an email at least twice in the last 28 days | +15 |
Thresholds and plays
| Fit | Engagement | Play |
|---|---|---|
| 70 or more | 40 or more | Qualification sets Sales-Ready to Yes. Assign a seller the same day |
| 70 or more | Under 40 | Right company, quiet. Keep in a journey |
| 40 to 69 | 40 or more | Qualify with a short call |
| Under 40 | Any | Nurture only |
After a week, open the Insights tab and move the cutoffs until the top group is a list sales can work.
Predictive lead scoring in Dynamics 365 Sales
What it needs
| Requirement | What Microsoft documents |
|---|---|
| License | “You must have a Sales Premium, Sales Insights, or a Sales Enterprise license” (source) |
| Sales Enterprise limit | 1,500 scored records per month through quick setup |
| Training data | “at least 40 qualified and 40 disqualified leads”, created and closed in the time frame you choose (source) |
| Training window | Three months to two years. The default is two years |
| Models | Up to 10, published and unpublished together |
Set it up
- In the Sales Hub app, go to Change area and select Sales Insights settings.
- Under Predictive models, select Lead scoring.
- Name the model. Letters, numbers, and underscores only.
- Optionally pick a Business process flow, the State option set with its Qualified Value and Disqualified Value, and a Filter column with Filter values to limit which leads are scored.
- Choose Train with leads from the past, then select Get started.
- Select View Details and the Performance tab. Microsoft labels a model Ready to publish, OK to publish, or Not ready to publish, based on the area under the curve (AUC) metric. You can publish a model that is not ready, but Microsoft says it will perform poorly.
- Set the ranges under Lead score grading, then select Publish.
For a second line of business, select Add model and give it its own filter. Per stage modeling is documented for opportunity scoring, not for lead scoring.
What reps see
From Microsoft’s page on prioritizing leads through predictive scores:
| What you see | What it means |
|---|---|
| Score | 0 to 100. Higher means more likely to convert to an opportunity |
| Grade | A, B, C, or D, in green, purple, yellow, and red. Your administrator sets the ranges |
| Score Trend | Improving, Declining, Steady, or Not enough info |
| Lead score widget | The top five positive and negative reasons, with Details for the full list |
The My Open Leads Scored system view lists leads with score, grade, and trend. New leads are scored within a few minutes of saving or importing. Updated leads refresh every 24 hours.
Tuning and retraining
Select Retrain automatically and the model retrains every 15 days. Microsoft’s page on editing and retraining explains how to choose attributes by hand: from the lead and its related contact and account, custom attributes, and intelligent fields such as email domain validation. Attributes on custom entities and date and time attributes are not supported. Ignore empty values keeps blank fields out of training.
If you have your own model, Microsoft documents how to write scores from a custom scoring model into the same score, grade, and trend fields through the Web API.
A simple rules-based score with Dataverse and Power Automate
Without either scoring feature, you can still build a basic score on the lead table.
Fit, with a calculated or formula column. Dataverse calculated columns use an IF, THEN, ELSE structure, support Whole Number, and can read a column on a parent table. Formula columns do the same job with Power Fx, where a null numeric column counts as 0.
Engagement, with a flow. The Power Automate trigger When a row is added, modified or deleted runs a flow when a Dataverse row changes. Select columns limits it to the columns you name, and Filter rows narrows it further. A flow can then add points to a number column on the lead.
A calculated column cannot span more than two tables, allows at most five chained calculated columns, and cannot trigger workflows or plug-ins. The trigger is evaluated for every update to a row, so a flow that writes to the lead can run more than once.
Putting the score to work
- Qualification and journeys. Qualification criteria set Status Reason or Sales-Ready, and Microsoft lists triggering a journey as one use of a qualified lead.
- Assignment and sequences. Dynamics 365 Sales work assignment uses segments and assignment rules to route leads to sellers, and segments can connect to sequences. Check your environment before routing on score or grade.
- Views. Give reps the My Open Leads Scored view, or add the Lead score column to their own.
Outbound marketing lead scoring has been removed
Customer Insights - Journeys was formerly Dynamics 365 Marketing, and it included a second module called outbound marketing with its own lead scoring models. Microsoft’s notes about outbound marketing removal say: “This module was removed in May 2026.” Outbound lead scoring models are on the list of assets that “weren’t removed” but “can’t be used in real-time”, and the remaining tables “aren’t being updated or refreshed”.
If routing or reports still read a score from an old outbound model, treat it as a frozen number. Microsoft’s pages do not describe an automatic migration for scoring models, so plan to rebuild them in Scoring Models.
Common setup mistakes
| Mistake | What happens | Fix |
|---|---|---|
| Reading a legacy outbound score | Routing runs on a number that no longer updates | Rebuild in Scoring Models |
| One model for fit and engagement | Active poor fits pass the same cutoff as good fits | Two models, both required |
| Inconsistent lead closing | The predictive model reflects rep habits | Agree when a lead is Qualified or Disqualified |
| Publishing a “Not ready” model | Reps stop trusting the score | Check the Performance tab first |
| No timeframe on interactions | Old activity keeps earning points | Set one on every interaction condition |
Our piece on where teams get ICP scoring wrong covers the modeling errors in more depth.
Limits to know
| Limit | Where | Workaround |
|---|---|---|
| 40 groups per scoring model | Customer Insights - Journeys | Split fit and engagement into separate models |
| Standard interactions only | Customer Insights - Journeys | Store other signals as lead attributes |
| Merged leads are not scored until you delete the old score or republish | Customer Insights - Journeys | Republish after a cleanup |
| 10 models | Sales | Use filters to cover segments |
| Leads must be open and created in the last two years | Sales (lead FAQ) | Close or recreate stale leads |
Where BoostIdeal fits
This guide is for teams on Dynamics 365, and BoostIdeal is not a Dynamics tool. It works with HubSpot, or a CSV you upload: you build the model and set the weights, and each score, shown out of 10, writes back to HubSpot properties with its reasons. It is not the right fit if your data is in Dynamics 365 and you do not want to work from CSV exports, or if you want a vendor to hand over a number you never question.
The principles carry over. Keep fit and engagement apart, let activity fade, show the reasons on the record, and make sure a person owns the weights.
Frequently asked questions
Does Dynamics 365 have lead scoring?
Yes, in two apps. Dynamics 365 Sales has predictive lead scoring, which scores open leads from 0 to 100 with a grade from A to D. Dynamics 365 Customer Insights - Journeys has lead scoring models that you build from conditions and points.
How many leads does predictive lead scoring in Dynamics 365 need?
Microsoft requires at least 40 qualified and 40 disqualified leads, created and closed within the training period you select. That period ranges from three months to two years.
Which license do I need for predictive lead scoring in Dynamics 365 Sales?
Microsoft’s setup guide says you need a Sales Premium, Sales Insights, or Sales Enterprise license. With Sales Enterprise, quick setup gives you 1,500 scored records per month.
How often do Dynamics 365 lead scores update?
In Sales, new leads are scored in near real time and updated leads refresh every 24 hours. In Customer Insights - Journeys, scoring models refresh every two hours. A predictive model set to retrain automatically retrains every 15 days.
Is outbound marketing lead scoring still available?
No. Microsoft states that the outbound marketing module was removed in May 2026. Outbound lead scoring models were not deleted, but Microsoft says they cannot be used in real-time journeys.
How do I mark a lead as sales ready in Customer Insights - Journeys?
Go to Lead management, Qualification, then Qualification criteria. Add one or more live scoring models with a qualifying score, choose the action that sets Sales-Ready to Yes or Status Reason to Marketing Qualified, and publish.
The takeaway
Dynamics 365 gives you a predictive score in Sales and a rules-based model in Customer Insights - Journeys. Use the rules model to decide when a lead is ready for sales, and the predictive score to order the queue. Check the 40 and 40 minimum first, rebuild anything that still depends on outbound marketing, and keep fit and engagement in separate models so the reason behind every handoff is visible.