---
schema_version: "1"
id: 10c58d339310d439a2829bcf9c685389418273ce86d89b18cbf6a38d8b7cdcf8
title: Explore Why a Value Does Better or Worse Than Average
collection: admin
version: current
locale: en-us
portal_url: https://docs.salesforce.com/admin/current/en-us/analytics/264-0-0/bi_edd_insight_why_average.md
canonical_url: https://help.salesforce.com/s/articleView?id=analytics.bi_edd_insight_why_average.htm&release=264.0.0&type=5
source_site: help.salesforce.com
content_type: documentation
content_hash: cd5801daae13bd9ba985b1378a96d300bba5d57750e5e223e2069de0138ed5ec
description: Learn why a category or bucket does better or worse than the global average.
products:
  - CRM Analytics
  - Cross Cloud
release: "264"
---
# Explore Why a Value Does Better or Worse Than Average

Learn why a category or bucket does better or worse than the global average.

![Note](https://sf-zdocs-cdn-prod.zoominsoftware.com/tdta-analytics-bi-264-0-0-production-enus/ec791473-f5b4-4f64-a656-9104f455d966/images/icon_note.png)

Note Einstein Discovery stories are now models. We wish we could snap our fingers to update the name everywhere, but you can expect to see the previous name in a few places until we replace it.

While descriptive insights show what happened in your data, diagnostic insights show why it happened. Diagnostic insights drill deeper and help you understand which variables most significantly drive the business outcome you’re analyzing. Einstein Discovery derives diagnostic insights from the model it produces. Einstein uses waterfall charts to visualize diagnostic insights.

![Note](https://sf-zdocs-cdn-prod.zoominsoftware.com/tdta-analytics-bi-264-0-0-production-enus/ec791473-f5b4-4f64-a656-9104f455d966/images/icon_note.png)

Note The term why refers to a strong data pattern, not necessarily a causal relationship.

1. [Open a Model](https://help.salesforce.com/s/articleView?id=analytics.bi_edd_story_open.htm&release=264.0.0&type=5 "Open a model to explore relationships in your data.").
2. In the Variables panel, click a variable.
3. From the Filter Selector, choose a value.
4. Insights based on your selections are listed in the first card.
    
    ![Industry equals Banking variable selected.](https://sf-zdocs-cdn-prod.zoominsoftware.com/tdta-analytics-bi-264-0-0-production-enus/ec791473-f5b4-4f64-a656-9104f455d966/bi/images/bi/bi_edd_insight_why_average_button.png)
    
    ![Results of the comparison showing three numbered areas](https://sf-zdocs-cdn-prod.zoominsoftware.com/tdta-analytics-bi-264-0-0-production-enus/ec791473-f5b4-4f64-a656-9104f455d966/bi/images/bi/bi_edd_insight_why_average.png)
    
    <table class="slds-table slds-table_bordered slds-max-medium-table_stacked slds-m-bottom_small ">
              <colgroup>
                <col style="width:8.375209380234507%">
                <col style="width:16.750418760469014%">
                <col style="width:74.87437185929649%">
              </colgroup>
              <thead class="slds-text-title_caps ">
                <tr class="">
                  <th scope="col" class="slds-cell-wrap" style="vertical-align:top;" id="d202312e108">#</th>
                  <th scope="col" class="slds-cell-wrap" style="vertical-align:top;" id="d202312e111">Area</th>
                  <th scope="col" class="slds-cell-wrap" style="vertical-align:top;" id="d202312e114">Description</th>
                </tr>
              </thead>
              <tbody>
                <tr class="">
                  <td data-label="#" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e108 ">1</td>

                  <td data-label="Area" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e111 ">Title</td>

                  <td data-label="Description" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e114 ">Drivers of {outcome} when {selected value}.</td>

                </tr>
                <tr class="">
                  <td data-label="#" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e108 ">2</td>

                  <td data-label="Area" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e111 ">Factor Explanations</td>

                  <td data-label="Description" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e114 ">Describe how each factor contributed to the predicted outcome.</td>

                </tr>
                <tr class="">
                  <td data-label="#" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e108 ">3</td>

                  <td data-label="Area" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e111 ">Waterfall Chart</td>

                  <td data-label="Description" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e114 ">
                    <p>Visualization of the factors, how they contributed to the predicted
                      outcome, and in comparison to the global outcome.</p>
                    <ul>
                      <li>X-axis represents the drivers' contribution to the model’s outcome.</li>
                      <li>Y-axis represents the drivers (explanatory variables) of the outcome
                        variable.</li>
                    </ul>
                  </td>

                </tr>
              </tbody>
            </table>
    
    The waterfall chart contains the following elements.
    
    ![Larger view of the waterfall chart in the results that shows seven numbered elements](https://sf-zdocs-cdn-prod.zoominsoftware.com/tdta-analytics-bi-264-0-0-production-enus/ec791473-f5b4-4f64-a656-9104f455d966/bi/images/bi/bi_edd_insight_why_average_chart.png)
    
    ![Note](https://sf-zdocs-cdn-prod.zoominsoftware.com/tdta-analytics-bi-264-0-0-production-enus/ec791473-f5b4-4f64-a656-9104f455d966/images/icon_note.png)
    
    Note In the following table, **Impact** and **Contribution to Outcome** mean the same thing. The value represents an estimate of the contribution of a single term to the predictions in aggregate, across all predictions. It compares the subset to the global dataset.
    
    **Methodology**: To compute impact, Einstein Discovery uses an aggregation of SHAP (SHapley Additive exPlanations) values. To learn more, see the original paper, [A Unified Approach to Interpreting Model Predictions](https://arxiv.org/abs/1705.07874), by Scott Lundberg and Su-In Lee.
    
    <table class="slds-table slds-table_bordered slds-max-medium-table_stacked slds-m-bottom_small ">
              <colgroup>
                <col style="width:8.375209380234507%">
                <col style="width:16.750418760469014%">
                <col style="width:74.87437185929649%">
              </colgroup>
              <thead class="slds-text-title_caps ">
                <tr class="">
                  <th scope="col" class="slds-cell-wrap" style="vertical-align:top;" id="d202312e212">Number</th>
                  <th scope="col" class="slds-cell-wrap" style="vertical-align:top;" id="d202312e215">Driver / Term</th>
                  <th scope="col" class="slds-cell-wrap" style="vertical-align:top;" id="d202312e218">Description</th>
                </tr>
              </thead>
              <tbody>
                <tr class="">
                  <td data-label="Number" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e212 ">1</td>

                  <td data-label="Driver / Term" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e215 ">Global Average</td>

                  <td data-label="Description" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e218 ">
                    <p>The gray bar represents the global average of all variables. Hover-over details:</p>
                    <ul>
                      <li><strong class="uicontrol">Drivers of <span class="keyword">Outcome</span>: Global
                          Average</strong></li>
                      <li><strong class="uicontrol">Global Mean</strong>: Mean <span class="keyword">Outcome</span> for
                        all variables in the dataset.</li>
                      <li><strong class="uicontrol">Global Count</strong>: Number of observations for all
                        variables in the dataset.</li>
                      <li><strong class="uicontrol">Contribution to Outcome</strong>: Same as the
                        <strong class="uicontrol">Global Mean</strong>.
                      </li>
                    </ul>
                  </td>

                </tr>
                <tr class="">
                  <td data-label="Number" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e212 ">2</td>

                  <td data-label="Driver / Term" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e215 ">Driver (First Impact)</td>

                  <td data-label="Description" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e218 ">
                    <p>Category for this insight. Represents the focus of this investigation.
                      Hover-over details:</p>
                    <ul>
                      <li><strong class="uicontrol">Drivers of <span class="keyword">Outcome</span></strong> when
                        <span class="keyword">Category</span> occurs.
                      </li>
                      <li><strong class="uicontrol">Impact</strong>: Contribution to all predictions. The
                        amount of change on <span class="keyword">Outcome</span> that the model attributes
                        to this <span class="keyword">Category</span>.</li>
                      <li><strong class="uicontrol">Coefficient</strong>: GLM (generalized linear model) only.
                        Represents the contribution of this term for a single prediction. For
                        numeric outcomes, the coefficient quantifies the change in the mean of the
                        outcome variable when there is a one-unit shift in the explanatory
                        variable, assuming all other variables in the model remain constant. For
                        binary classification outcomes, the coefficient represents log odds (the
                        logarithm of the odds).</li>
                      <li><strong class="uicontrol">Frequency</strong>: How much of the entire dataset this
                        <span class="keyword">Condition</span> represents.
                      </li>
                      <li><strong class="uicontrol">Conditional Frequency</strong>: How much of this subset
                        the <span class="keyword">Category</span> represents.</li>
                      <li><strong class="uicontrol">Precluded Sum</strong>: Aggregate contribution to the
                        outcome of all observations that are included in the Global Average (gray
                        bar) but are excluded from the Category Average (blue bar). For example,
                        if the Global Average includes all months in a year, and the Category
                        Average includes only December, then the Precluded Sum includes January
                        through
                        November.</li>
                      <li><strong class="uicontrol">Precluded Count</strong>: Number of terms that were
                        precluded from this insight (terms that are not this
                        <span class="keyword">Category</span>).
                      </li>
                      <li><strong class="uicontrol">Contribution to Outcome</strong>: Same as
                        <strong class="uicontrol">Impact</strong>.
                      </li>
                    </ul>
                  </td>

                </tr>
                <tr class="">
                  <td data-label="Number" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e212 ">3</td>

                  <td data-label="Driver / Term" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e215 ">Driver (Next 4 Impacts)</td>

                  <td data-label="Description" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e218 ">
                    <p>Terms interacting with <span class="keyword">Category</span>. Hover-over details:</p>
                    <ul>
                      <li><strong class="uicontrol">Drivers of <span class="keyword">Outcome</span></strong> when
                        <span class="keyword">Condition</span> occurs.
                      </li>
                      <li><strong class="uicontrol">Impact</strong>: The amount of change on
                        <span class="keyword">Outcome</span> that the model attributes to this
                        <span class="keyword">Condition</span>.
                      </li>
                      <li><strong class="uicontrol">Coefficient</strong>: GLM (generalized linear model) only.
                        Represents the contribution of this term for a single prediction. For
                        numeric outcomes, the coefficient quantifies the change in the mean of the
                        outcome variable when there is a one-unit shift in the explanatory
                        variable, assuming all other variables in the model remain constant. For
                        binary classification outcomes, the coefficient represents
                        log
                        odds (the logarithm of the odds).</li>
                      <li><strong class="uicontrol">Frequency</strong>: How much of the entire dataset this
                        <span class="keyword">Condition</span> represents.
                      </li>
                      <li><strong class="uicontrol">Conditional Frequency</strong>: How much of this subset
                        the <span class="keyword">Condition</span> represents.</li>
                      <li><strong class="uicontrol">Contribution to Outcome</strong>: Same as
                        <strong class="uicontrol">Impact</strong>.
                      </li>
                    </ul>
                  </td>

                </tr>
                <tr class="">
                  <td data-label="Number" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e212 ">4</td>

                  <td data-label="Driver / Term" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e215 ">Small Terms Related To</td>

                  <td data-label="Description" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e218 ">
                    <p>The aggregated effect of all terms interacting with
                      <span class="keyword">Category</span> that do not appear in the other bars shown
                      previously. There are many factors impacting the outcome for the selected
                      variable. The chart can't possibly contain all of them. It shows only the most
                      impactful ones. The remaining impacts are grouped together in small terms.
                      This is done specifically to differentiate them from the groups below
                      (Unrelated Small Factors and Unexplained).
                    </p>
                    <p>Hover-over details:</p>
                    <ul>
                      <li><strong class="uicontrol">Combined Impact</strong>: Net impact on the
                        <span class="keyword">Outcome</span> of all the combined terms.
                      </li>
                      <li><strong class="uicontrol">Terms Combined</strong>: Number of terms combined.</li>
                      <li><strong class="uicontrol">Contribution to Outcome</strong>: Same as
                        <strong class="uicontrol">Combined Impact</strong>.
                      </li>
                    </ul>
                  </td>

                </tr>
                <tr class="">
                  <td data-label="Number" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e212 ">&nbsp;</td>

                  <td data-label="Driver / Term" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e215 ">Other Impacts</td>

                  <td data-label="Description" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e218 ">
                    <p>Terms that are not specific to <span class="keyword">Category</span> but that still
                      occur more or less often. Hover-over details:</p>
                    <ul>
                      <li><strong class="uicontrol">Drivers of <span class="keyword">Outcome</span></strong> when
                        <span class="keyword">Condition</span> occurs.
                      </li>
                      <li><strong class="uicontrol">Impact</strong>: The amount of change on
                        <span class="keyword">Outcome</span> that the model attributes to this
                        <span class="keyword">Condition</span>.
                      </li>
                      <li><strong class="uicontrol">Coefficient</strong>: GLM (generalized linear model) only.
                        Represents the contribution of this term for a single prediction. For
                        numeric outcomes, the coefficient quantifies the change in the mean of the
                        outcome variable when there is a one-unit shift in the explanatory
                        variable, assuming all other variables in the model remain constant. For
                        binary classification outcomes, the coefficient represents log odds (the
                        logarithm of the odds).</li>
                      <li><strong class="uicontrol">Frequency</strong>: How much of the entire dataset this
                        <span class="keyword">Condition</span> represents.
                      </li>
                      <li><strong class="uicontrol">Frequency</strong>: How much of the entire dataset this
                        <span class="keyword">Condition</span> represents.
                      </li>
                      <li><strong class="uicontrol">Conditional Frequency</strong>: How much of this subset
                        the <span class="keyword">Condition</span> represents.</li>
                      <li><strong class="uicontrol">Contribution to Outcome</strong>: Same as
                        <strong class="uicontrol">Impact</strong>.
                      </li>
                    </ul>
                  </td>

                </tr>
                <tr class="">
                  <td data-label="Number" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e212 ">5</td>

                  <td data-label="Driver / Term" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e215 ">Unrelated Small Contributors</td>

                  <td data-label="Description" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e218 ">
                    <p>Unrelated small contributors are not specific to
                      <span class="keyword">Category</span>. This section shows us factors that have
                      positive or negative effects on all <span class="keyword">Outcome</span>s.
                    </p>
                    <p>Hover-over
                      details:</p>
                    <ul>
                      <li><strong class="uicontrol">Combined Impact</strong>: Net impact on the
                        <span class="keyword">Outcome</span> of all the unrelated small contributors.
                      </li>
                      <li><strong class="uicontrol">Terms Combined</strong>: Number of terms combined into the
                        unrelated small contributors.</li>
                      <li><strong class="uicontrol">Contribution to Outcome</strong>: Same as
                        <strong class="uicontrol">Combined Impact</strong>.
                      </li>
                    </ul>
                  </td>

                </tr>
                <tr class="">
                  <td data-label="Number" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e212 ">6</td>

                  <td data-label="Driver / Term" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e215 ">Unexplained</td>

                  <td data-label="Description" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e218 ">Represents the unexplained variables on the <span class="keyword">Outcome</span>.
                    In diagnostic insights, drivers are derived from a model, and no model is
                    perfect. The unexplained bar quantifies the gap that this model hasn't been
                    able to attribute to different drivers.</td>

                </tr>
                <tr class="">
                  <td data-label="Number" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e212 ">7</td>

                  <td data-label="Driver / Term" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e215 ">Category Average</td>

                  <td data-label="Description" class="slds-cell-wrap" style="vertical-align:top;" headers="d202312e218 ">
                    <p>The blue bar represents the average for the selected category. Hover-over details:</p>
                    <ul>
                      <li><strong class="uicontrol">Drivers of <span class="keyword">Outcome</span>: Average when
                          <span class="keyword">Category</span></strong></li>
                      <li><strong class="uicontrol">Global Mean</strong>: Mean <span class="keyword">Outcome</span> for
                        <span class="keyword">Category</span>.
                      </li>
                      <li><strong class="uicontrol">Global Count</strong>: Number of observations for
                        <span class="keyword">Category</span>.
                      </li>
                      <li><strong class="uicontrol">Contribution to Outcome</strong>: Same as
                        <strong class="uicontrol">Global Mean</strong>.
                      </li>
                    </ul>
                  </td>

                </tr>
              </tbody>
            </table>