CloudAscent Propensity reports available from Partner Center dashboard

Appropriate roles: Executive report viewer | Report viewer

The Partner Center dashboard provides downloadable propensity data from the CloudAscent program. The data shows the customers' likelihood to purchase Microsoft products. This article describes the breakdown of this data, how to use the scoring, and what it means.

Summary definitions

  • SMC Customers– This is the total number of customers in the propensity downloads. Customers are identified by partner of record.
  • Expire Agreements– Within the current fiscal year, we're providing the number of expiring agreements.
  • Open Expiring Revenue– The revenue associated to the open expiring agreements.

Screenshot of Customers Opportunities Summary dashboard.

CloudAscent SMB segmentation

The small to medium business (SMB) segment is divided into three distinct sub segments.

  1. Top Unmanaged includes the largest SMB customers with the most opportunity for Microsoft. Typical Top Unmanaged customers share similar characteristics to Managed accounts, with large number of employees, large IT budgets and spend, and large amounts of potential revenue for Microsoft.

    We define Top Unmanaged in two ways:

    • Top Unmanaged User Based– includes accounts with 300 or more employees. User-Based accounts are great targets for first-time purchase, or expansion of user-based subscription products such as Microsoft 365, Dynamics 365, or Surface.
    • Top Unmanaged Compute Based – includes accounts with Azure potential greater than $10k. Compute based accounts include existing Azure. accounts with significant future year potential and accounts who have yet to purchase Azure yet but have potential for Azure greater than $10k.
  2. Medium Business includes existing customers and prospect accounts with 25 to 300 employees.

  3. Small Business includes businesses with 10-25 employees.

  4. Very Small Business includes businesses with 1-9 employees.

Customer by SMC type.

Top Unmanaged and Medium Business subsegments represent high life-time value (LTV) customers for Microsoft, and Microsoft Partners. Because of this, they're the lead areas of focus for driving growth in this segment. In these two subsegments, we're better positioned to acquire the socket with Microsoft 365, monetize further with D365/Azure line of business (LOB) apps, and realize a high LTV for Microsoft.

Today we have two key areas of opportunity – 1. our customer adds growth; 2. while we do well acquiring cloud sockets leading with Microsoft 365, we have a large opportunity in Dynamics 365 and Azure.

The following screenshot represents the four SMB Subsegments. CloudAscent prioritize the profiling, scoring, and modeling of all Top Unmanaged and Medium Business accounts.

Screenshot of SMB subsegments.

CloudAscent Machine Learning

SMB uses machine learning technology to drive sales and marketing customer predictions within the Top Unmanaged and Medium Business segments. How are signals collected and turned into propensity recommendations?

  • Data Collection: Web crawlers scan and collect billions of customer signals by pinging the company domains, and monitoring blog posts, press releases, social streams, and technical forums. In addition to the collected signals, firmographics information is collected from both internal and external sources such as D&B, Microsoft Internal subscription and transactional data.

  • Machine Learning: The signals are fed into the machine learning model that outputs a structured data set of Sales and Marketing predictions for each customer by cloud product and cluster. Each customer is scored using a look alike model to Microsoft's top SMB that determines the customer's Fit, and machine learning algorithms that integrate the customer's online behavior define as Intent. The scoring is merged into clusters that show a customer's propensity to purchase Microsoft Cloud Products.

  • Optimization: The Machine Learning system optimizes the models by consuming the transaction data monthly and the subscription data quarterly. Using the win/loss data, the Machine Learning adjusts the algorithms and validates that the models are working as expected by comparing cluster recommendations to opportunities acted upon in MSX.

Screenshot of SMB machine learning.

CloudAscent Propensity

How are propensity recommendations created?

Using signals collected via web crawlers and data provided from various sources, we consolidate the firmographics data and customer's social media signals. The scoring uses these signals and data in comparison models for fit and scoring models for Intent.

  1. Customer Account Fit

    • Internal and External data points that define firmographics.

    • Fit scoring uses a look alike model to our best SMB to compare customers and see if they're a potential fit for Microsoft Cloud Products.

    • Fit scoring is updated quarterly

  2. Customer Account Intent

    • Signals related to Social media and a customer's online behavior define Intent.

    • Intent scoring is overlaid on top of fit to define the clusters.

    • Intent scoring is updated monthly.

    CloudAscent SMB predictive models.

  3. Clustering

    The Signals for fit and intent are consolidated into a clustering score. CloudAscent has four clusters:

    • Act Now - sales ready customers
    • Evaluate - marketing ready customers
    • Nurture - drive awareness campaigns
    • Educate - educate and monitor for intent

    The clustering allows users to target specific customers for sales and marketing initiatives based on segment factors, for example: product, geo, industry and vertical.

    The Propensity model tab in the CloudAscent Workbooks shares the propensity and the estimated whitespace revenue. To define the clustering of Fit and Intent, we go through the following steps:

    1. Using ML Models, we first calculate Customer Fit Score and intent Score on a scale of 100. Exact Scores will vary based on ML Models. Example Scores Below:

      Classification Score
      High 75 - 100
      Medium 55 - 74
      Low 30 - 54
      Very Low 0 - 29
    2. Using the rule above, we classify companies to be High, Medium, Low, and Very Low across both Customer Fit and Intent Signals.

    3. We plot customer fit and intent signals on a 2D matrix with each intersection representing the propensity. For Example, High Fit + High Intent = A1, representing the highest propensity.

    4. Finally, these segments group to form clusters. For Example, A1, A2, A3, A4 form the Act Now cluster.

      CloudAscent models.

    For these customers, we recommend targeting Act Now and Evaluate customers.

CloudAscent Products & Models

The following graphic provides a view of each propensity model within CloudAscent:

CloudAscent propensity model.

Whitespace models are composed of predictions for existing Microsoft customers where they don't have a product and/or are net new prospect customers.

Upsell models use transaction data to predict the potential for upsell in Azure and Microsoft 365 SKUs.

These customers will already have both Azure or Microsoft 365 and the upsell model shows that they’re likely to purchase more of their existing SKU.

EOS shares the end of service (EOS) customers for Win 7, Office 2010, SQL Server, and Windows Server. The EOS data is pulled from MS Sales and overlaid with the CloudAscent propensity modeling where available. EOS data lives in the Modern Work and Azure Sales plays.