Journey Predictions uses a machine learning model to surface the accounts most likely to become an open opportunity/deal. The model is trained using accounts past or present in an open opportunity/deal in your CRM, and analyzes the fit, intent, and engagement signals in the time leading up to becoming an open opportunity to find accounts that look like and are acting like those accounts that became an open opportunity.

The model scores accounts into High and Medium bins; these Prediction Scores are refreshed weekly on Sunday at 12am UTC, providing a reliable view of account potential.

We make Journey Predictions available in the Command Center, Account List Building, Contact List Building, Workflows, and sync the data to Salesforce or HubSpot.

Journey Predictions takes the guesswork out of how to combine various in-market and fit signals so that your marketing and sales team can improve their efficiency by focusing on the accounts most likely to become an open opportunity.

Who can access

Access to Journey Predictions depends on your package.

To find your current package, log in to AdRoll ABM and navigate to Settings > Billing > Plans & Usage.

In addition to having a package that includes Journey Predictions, you must meet the technical and data requirements below. After these requirements are met and the training data set is available, we will generate prediction scores the next weekly sync on Sunday at 12am UTC. 

Technical Requirements

1. AdRoll ABM is integrated with Salesforce or HubSpot:
   • • Salesforce Initial Installation Guide
     • HubSpot Initial Installation Guide
2. You must customize at least one Journey Stage using any field in the Salesforce Opportunity object or HubSpot Deal object. 

Data Requirements

1. A minimum of 3 months of historical opportunity/deal activity data in your CRM. 
2. A minimum of 10 accounts in your CRM that are currently in, or have been in the Open Opportunity/Deal Journey Stage, based on your Journey Stage configuration. This data is needed as your training data set.

How does Journey Predictions work

Our predictive Machine Learning model scores accounts in a few different ways:

1. The training data set is derived from your Journey Stage configuration in the AdRoll ABM platform. We use the first Journey Stage that uses your Salesforce Opportunity or HubSpot Deal objects as a way to infer what you consider an open opportunity. We will consider all accounts in that Journey Stage, and stages further in the funnel.
2. The prediction model looks at the fit, intent, and engagement of the accounts in your training data set prior to those accounts having an open opportunity or deal opened. The prediction model uses the following signals:
   • • Firmographic Company Attributes (Fit):
       • • Industry
         • Company Size
         • Company Revenue
         • Technologies
     • Engagement:
       • • Website Page Views
         • Form Fills
         • Salesforce, HubSpot, and Marketo Activities, including event type and Inbound/Outbound designations
     • Intent:
       • • AdRoll Keyword Intent
         • Bombora Company Surge Intent
     • Other variables:
       • • Count of intent and engagement over a time frame
         • Cumulative intent and engagement
         • Recency of intent and engagement

3. The weighting of each parameter in our prediction model is unique to your website activity and CRM data. The model evaluates all available data sources and determines which parameters have the strongest predictive power for your specific use case. This approach applies to all data sources, ensuring that the predictions are tailored to your unique data and business context. For example:

   • • If the confidence or quality of your CRM data is low, the model will recognize that there is no clear pattern in this data and assign it minimal weight.

     • Conversely, if the model identifies a strong correlation—such as most Open Opportunities coming from a specific industry—it will assign higher scores to accounts within that industry and lower scores to those outside of it.

     • If no significant patterns are detected (e.g., Open Opportunities occur evenly across all industries), the model will assign little weight to this firmographic attribute. 

        

4. We look at accounts that are not yet in the first opportunity stage and assess how closely these accounts look and are behaving like the accounts that became an open opportunity. In order to make sure that the accounts that are receiving Journey Predictions are relevant we assign the prediction scores to only a subset of accounts.
   • • The model will produce Journey Predictions for accounts that meet at least one of the following criteria:
       • • The account is on an Account List.
         • The account has shown intent or engagement activity in the last 90 days.
     • The model will not produce Journey Predictions for accounts that meet any of the following criteria:
       • • The account is currently part of the training data set (have an open Opportunity or Deal).
         • The account has been part of the training data set in the last year.
         • The account is a current customer.
         • The account was a customer but churned in the last year.

Journey Predictions Score: High vs Medium

Journey Predictions will score accounts either as High or Medium:

• High: top 5% of positively predicted accounts – on average, High accounts are 25x more likely to become an open opportunity than an average account
• Medium: top 40% of positively predicted accounts – on average, Medium accounts are 5x more likely to become an open opportunity than an average account

Some accounts will have a blank Journey Prediction score. Click here to learn the minimum conditions that must be met for an account to receive a Journey Prediction score.

Where to access Journey Predictions

Journey Predictions is available in the following areas of the AdRoll ABM platform:

• In the Command Center as a filter and a data column.
• In the Enhanced Contact List Builder as a filter.
• As a filter when creating an Account List through the Command Center.
• As a trigger in Workflows which enables custom actions.
• Synced to your CRM Salesforce Custom Objects. 
• Synced to your CRM HubSpot Custom Properties.

How to use Journey Predictions

Apply as a filter in the Command Center

Start by viewing how the high and medium predicted accounts fall across your customized buying journey: accounts with High or Medium Accounts with a Journey Prediction score of Medium or High are highlighted in yellow, while accounts with no predictions are shown in purple.

Filter the accounts distributed across the buying journey using the Journey Prediction filter.  This will surface only High and/or Medium predicted accounts. You can apply additional filters in the Command center if you want to focus only on a subset of accounts with High and/or Medium Journey Prediction scores layered with Account Firmographic Data, ICP, Account Lists, and Intent Data available in the filters.

You can see the detailed Prediction Scores in the main table, sorted from High to Medium while starting with Prediction Scores first.

After applying Journey Prediction filters use the Insight Cards to identify where your marketing and sales teams are missing out on capturing the demand being displayed. and take the next best actions suggested within the Insight Cards, for example, create new Account records in your CRM or create an Account List.

Create an Account List from the Command Center

If you apply a Journey Prediction filter in the Command Center, and take the action to create an Account List, that filter will automatically be passed through as a rule in the Account List Builder.

Additionally, you can add Journey Predictions as a filter to Account Lists that have been built using the Command Center. Please note that Journey Prediction filters are not available for Account Lists that have not been built using the Command Center.

Create an Enhanced Contact List

You can optionally add Journey Prediction rules in to the Enhanced Contact List. This will surface Contacts from your integrated CRM (Salesforce or HubSpot) at High and/or Medium predicted Accounts.

Trigger a Workflow

You can trigger a Workflow using Journey Predictions - this is a great way to identify Contacts a given Sales Rep is responsible for working.

You can use Contact Discovery Workflows to identify Contacts at High and/or Medium predicted accounts that are not present in your CRM, and add those contacts.

Journey Predictions FAQs

Why do some accounts have a blank Journey Prediction?

In order to make sure that the accounts that are receiving Journey Predictions are relevant we assign the prediction scores to only a subset of accounts.

• • The model will produce Journey Predictions for accounts that meet at least one of the following criteria:
    • • The account is on an Account List in AdRoll ABM.
      • The account has shown intent or engagement activity in the last 90 days.
  • Journey Predictions will be blank for accounts that meet any of the following criteria:
    • • The account is currently part of the training data set (have an open Opportunity or Deal).
      • The account has been part of the training data set in the last year.
      • The account is a current customer.
      • The account was a customer but churned in the last year.

What attributes and signals does the model take into account?

The predictions model looks at the fit, intent, and engagement of the accounts in the positive training data set prior to those accounts having an opportunity opened. We then look at accounts that are not yet in the open opportunity stage and assess how closely these accounts look and are behaving like the accounts that became an open opportunity. 

The Data / Features in the predictions model include:

• Company Attributes (Fit): industry, company size, company revenue, technologies, etc.  We consider all account attributes that are part of the ICP Fit Grade Account Scoring model.
• Engagement: Website Page Views, Form Fills, Salesforce, HubSpot, and Marketo Activities including event type, and inbound/outbound designations.
• Intent: AdRoll Keyword Intent and Bombora Company Surge Intent.
• Other Variables: The model considers other variables like the count of intent/engagement over a given time frame, the cumulative intent/engagement, and recency of the intent/engagement.

What is the duration it takes to generate predictions?

Once the eligibility requirements are met, you can expect prediction scores being generated the next weekly sync on Sunday 12am UTC.

How do you determine which accounts were in an open opportunity/deal?

AdRoll ABM uses your customized Journey Stages to determine which accounts are currently in an open opportunity/deal or were previously in an open opportunity/deal.

AdRoll ABM uses the first stage that uses an Opportunity (Salesforce) or Deal (HubSpot) rule to infer which stage represents an open opportunity.

What happens if I change my Journey Stage definitions?

Predictions are versioned to the stage changes, so it’s likely the accounts with a High and Medium Prediction Score will change. After updating your Journey Stage Definitions using Salesforce Opportunity or HubSpot Deals the system will recalculate Journey Predictions the next weekly sync on Sunday at 12am UTC.