Asar Digital

The Awesome Predictive Power of Deal Intelligence for SAP Sales Cloud

Sales Reps need the ability to close quickly and hit targets, ensuring an ever-increasing win rate. SAP Sales Cloud’s Deal Intelligence feature will empower your sales team with a toolbox that helps them easily complete these objectives. 

Here we’ll cover Deal Intelligence’s Machine Learning capabilities and offer guidance towards implementation so that you can get your team onboarded to these time-saving features quickly, and keep nurturing opportunities toward close. 

Deal Intelligence: What It Is and What It Does

What It Is

When it comes to opportunity management, modern businesses share common pain points:

 – Sales Managers lack the full picture, and, due to a lack of information, enter at-risk opportunities without a clear sense of what’s going on

  • Less than half of all forecasted opportunities make it to closing 
  • Sales pipelines bloat up with low-propensity opportunities 

Deal Intelligence will automatically rate, score, and list your team’s deals based upon their proximity to close. With this information your team can 

  • close deals more quickly 
  • improve their win rates 
  • accelerate sales and have a higher level of predictability regarding revenue and deal flow 

How It Works

Deal Intelligence migrates historical data from existing opportunities living within SAP Sales Cloud. It uses this data to train the machine learning model. This creates a customer-specific model, and gives the end users (your sales team) a score and ranking on the opportunities, so they have a clear picture of where they stand. 

Upon implementation, the users also see a new pane within their opportunity work centers, giving them status overviews for each relevant opportunity. An additional column; the “score” column, is introduced into the module. This will display the score for each opportunity in their list. 

Based upon authorization roles of each user, they will have viewing access to the pane and the score column. The score field also acts as filter criteria so that opportunities can be selected. 

Scores and other information contained within the side pane automatically update daily based upon changes and modifications made within the system. The updates trigger at midnight in the time zone of your company’s data center. 

Necessary Components for Implementation

If you’re interested to implement Data Intelligence effectively, the following aspects need your consideration: 

  1. An SAP Sales and Service solutions Enterprise License is a mandatory prerequisite 
  2. Data Volume
    1. The greater your archive of historical data, the better. The system will use the prior 12 months of opportunities and their data in order to create its prediction model. The bare minimum for this process is 5000 opportunities. This number greatly depends upon the following: 
  3. Data History
    • An opportunity’s chance to be won or lost is calculated by the model, using historical data and computing how the opportunity has evolved over time. Like with any system, the more accurate and up-to-date the data, the better the predictive model that emerges. 
  4. Data quality
    • The prediction model words by identifying key fields in opportunities, and then considering those for prediction. The fields should be kept relevant via consistent updating with necessary characteristics. In an ideal system, you make the fields mandatory from the onset. 
  5. Data consistency
    • Opportunity fields need to be populated on a consistent basis, or accurate predictive models will not emerge. For instance, with 10,000 opportunities, half without the “probability” field populated, a resulting model will have diminished accuracy.  
  6. Data balance
    • Data balance is critical to creating an accurate, actionable prediction model. Say that your sales reps have only created data for wins in the system, the model will be very inaccurate. Likewise, if won opportunities weren’t set to “won” or “lost” but left open, the model won’t give a clear picture. 
  7. Customization
    • If your system is heavily customized, this can influence your prediction model. Specific developments, extension fields, and other similar customizations need to be individually checked and monitored to ensure they don’t interfere with Deal Intelligence’s machine learning capabilities. 

Implementation Process

Implementation can be segmented into the following process, for most use-cases: 

  1. Precheck Data
    • Your data needs to comply with the above necessary steps and checkpoints. This is currently a manual process. However, special check tools are being built by SAP to support this step for future implementations. 
  2. Activate and Configure Deal Intelligence
    1. Tenant Decision: The order of tenants for implementation needs to be decided upon. Typically, the test tenant will not have enough data, and is not a good candidate for the creation of a prediction model. Another option is to use “productive” data to test the feature in a duplicate of the production tenant. There are considerations if using this method – an additional tenant needs to be created, which could lead to additional costs. Testing the dynamics of the model isn’t as effective in such a duplicate – given that the data will not be constantly changing/updating. Time changes due to duplication/recreation can also lead to different results. 
    2. Activation: Deal Intelligence is activated by SAP. To get started, you need to create an incident in the tenant where it needs to be activated. Within the incident, your request the activation of the Machine Learning scenario: “Deal Intelligence.” SAP is creating a future version wherein activation can be done through scoping. To give authorization for views of the new side panel; Deal Intelligence related data in the opportunity work center, you will need to create a new business role. This role has be specified within your incident request for activation. SAP will then make appropriate changes to the user interface. The role is then assigned to your relevant end-users. In future versions, this process will be replaced whereby authorization can be granted directly from Administrators. 
  3. Train the Machine Learning model
    • A customer-specific, trained model needs to be created prior to being able to use Deal Intelligence. This model is used to predict scores for your opportunities. This can be done by entering the Administration work center, choosing the “Prediction services” view, and selecting “Configure.” Here are the steps:
      1. Choose opportunity scoring 
      2. Add a new model 
      3. View the new model with your chosen description 
      4. Mark the new model, choose “Train” 
      5. Historical opportunity data is gathered, sent to the machine learning component, and the customer-specific model is trained/create based upon your existing historical data 
      6. Status will report “Finished” when the Machine Learning model is complete 
  4. Validate model training results
    • The machine learning component will return an accuracy rating to give you a sense of the usability of your model. Generally, if this rating is below 50, your model is not good enough to provide usable results. This comes with the caveat that individual projects can define their own accuracy level depending upon specific needs. Once the accuracy is satisfactory, activate the model. 
  5. Test
    • Once activated, open opportunities are scored. After the first open day of the model, an overnight scoring occurs via machine learning. Validate the score by checking it against your own expectations for an individual opportunity, as a primary test. Then, move onto another opportunity and see if the updated score matches. 
  6. Go-Live
    • After testing, you can then open up the feature so that your sales team has access to the UI. If you started within the predictive tenant; assign roles accordingly. Steps from phase 3 (Train the Machine Learning model) need to be repeated if you started within the test tenant. 

This automative feature will catapult your sales team to a higher level of deal-closing. Let machine learning organize and rank your opportunities that your team isn’t flying blind, and can work from accurate data to prioritize the right opportunities, and keep your pipeline healthy! 

Visit this article from SAP for a list of FAQs on Deal Intelligence implementation. 

Scroll to Top