Use this guidance when editing your instructions or chatting with the Agent.
⚠️ Important
The easiest way to generate effective instructions for the Analyst Agent is to build your Report using chat in the AI Sidebar. Once you’re satisfied with the output, save it as a Mission. You can then manually edit a Mission’s Instructions at any time to adjust future runs, using this article.
Essential and optional instructions
Your instructions must contain the following:
What data to analyze.
Which analysis steps to follow.
Optionally:
Include examples of how the Report should look.
Specify a tone and style for the Report (e.g. concise, executive-friendly, bullet-pointed).
How the output should be structured.
Define a language for the Report. The instructions and Report can be in any of 50 widely-spoken languages.
Instruction optimization
Be as explicit as possible:
When you know a specific Metric, Dimension or Application Variable is needed, use the Object picker: type @ to open a dropdown of eligible options. This way, even if the object names are updated at a later time, the Mission still runs. You can reference multiple objects in one instruction. To be eligible, Metrics must have AI data access turned on. For more information, see Block settings for Metrics.
To have no charts in your Report, specify “Don’t display charts in the Report.”
Metrics shared from other Applications are eligible.
Directly reference in plain text the Dimension Items you want to filter on.
ℹ️ Note
Agents cannot see formulas.
To improve clarity, paste your instruction into an LLM such as ChatGPT, preceded by guidance such as: “What do the following instructions for a Report mean? Can you reformulate them for better readability?”
Supported capabilities
See examples below to replicate in your instructions.
Identify Drivers of Revenue Underperformance
Unlock a clear view of quarterly revenue performance with this analysis example. It enables you to:
Compare actual revenue against forecasts for the last quarter
Pinpoint the top 3 underperforming country–retailer combinations
Drill down to SKU level to surface the main variance contributors
Summarize findings in a concise executive overview before detailing key drivers
Present results in a formal quarterly business review tone—with emojis for emphasis
It’s built to deliver structured, insightful, and engaging revenue variance analysis.
Analysis
Analyze @Revenue metric, compare @Version Actual vs Forecast for the @Last_Quarter
Identify the top 3 @Countries - @Retailers that have underperformed.
For each flagged combination, drill at @SKU level to identify the top 3 contributors of the variance.
Use the notes in text metric @weekly_business_comments to add context / insights for any anomalous or unexpected revenue data.
Formatting
Start with a short executive summary (2-3 lines).
Then list the key drivers of the variance using bullet points.
Use a formal quarterly business review tone, with emojis.Explain OPEX Variance by General Ledger Account and Department
This generates a structured framework for analyzing operational expenses. It helps you:
Compare OPEX actuals vs. budget at a global level and spot top cost centers driving variances
Flag significant deviations by GL Account × Department where variances exceed ±5k
Drill down further by country and highlight both absolute (€/$) and percentage (%) differences
Present results consistently from run to run, using bullets for short lists or tables for longer ones
Add concise explanations to contextualize each variance for clarity
It’s designed to ensure insightful and well-formatted financial analysis output.
Analysis
Start by analyzing the @OPEX metric, compare @Version Actuals vs Budget at global level for the @Current_Month.
Identify the top 3 @Cost_centers contributing to the total variance.
Breakdown by @GL Account x @Department and flag only combinations where the absolute variance between @Versions Actuals – Budget exceeds ±5000
For each flagged combination:
- Drill down further by @Country
- Show variance in both Absolute terms (€) and Percentage (%)
Formatting
- Use bullet points to list key GLs and their associated variances. Add short explanations where possible.
- If there are more than 5 items, use a table instead.
Output Format Example
- GL 60000 - Salaries: +€15.2k vs Budget (+18%) - Higher training expenses in Marketing (+€8.7k)
- GL 62100 - Travel: –€11.3k vs Budget (–9%) - Fewer offsites held this quarterDetect and Explain Attrition Trends
Uncover key attrition risks and workforce trends with this analysis framework. It allows you to:
Identify the top 3 roles with the highest attrition over the past year
Correlate attrition with performance, tenure, and location to surface deeper insights
Start with a concise executive summary on global attrition patterns
Present results in a structured table (Role | Attrition % | HC Lost | Trend 📈/📉)
Add focused bullet points under each role to explain the drivers behind attrition
It’s designed to provide a clear, insight-rich view of attrition dynamics without relying on charts.
Analysis
For the whole analysis filter on @CurrentYear
1. Identify the Top 3 @Employee_Role with the highest attrition using @Attrition metric
2. For each of these flagged roles, provide additional insights by correlating @Attrition to:
@Attrition_Performance → check if high performers or low performers are leaving.
@Attrition_Tenure → check if attrition is concentrated among junior or senior staff.
@Attrition_Location → see if attrition is localized in specific geographies.
Formatting
Begin with an executive summary (1–2 sentences) on global attrition trend.
Present the Top 3 Roles with attrition in a table: Role | Attrition % | HC Lost | Trend (📈/📉).
Add bullet points under each role summarizing key insights
Don't display chartsGenerate a Customized Multi-Report for Each Function to Explain Budget Variance
Get a structured view of functional performance versus budget. This analysis helps you:
Compare actuals vs. budget at a function level and surface the largest positive and negative variances
Drill down into key cost drivers within each function (GL, vendor, or initiative)
Flag only material deviations based on a defined threshold
Add short contextual explanations from prior commentary metrics
Produce a consistent variance summary formatted for monthly reviews
It’s designed to deliver clear variance diagnostics function-by-function.
Analysis
For @Function = @Current Item:
Compare @Account actuals vs budget for @Current_Month.
Rank @Account by absolute variance and flag the top 5 deviations.
For each flagged @Account:
- Break down variance by @Initiative (if available; otherwise by @GL_Account).
- Show absolute and percentage variance.
Include only rows where |Actual – Budget| > ±@Threshold.Analyze staffing gaps
Gain clear visibility into organizational staffing gaps with this analysis approach. It guides you to:
Compare total headcount demand versus supply at a global level, highlighting absolute and percentage gaps
Identify the entity with the largest shortfalls and zoom into the top 3 cost centers driving them
Break down gaps further by job profile within each flagged cost center
Surface any additional critical shortages across the model where gaps exceed 30%
Present results in structured tables and pair findings with actionable recommendations (open positions, transfers, or scope adjustments)
It’s built to deliver a precise and actionable view of workforce capacity gaps for the current month.
Analysis
Compare @Headcount_Demand and @Headcount_Supply to identify the most critical staffing gaps across the organization for the month of @Current_Month.
Global overview to compare total demand vs. supply (absolute and % gap).
- Identify the main @Entity with the largest negative gaps (Demand > Supply).
- Then for each @Entity, zoom into the top 3 @Cost_Center with the largest shortfalls.
- Then, within each flagged @Cost_Center, list all the @Job_Profile with a negative gap.
- Finally, scan the full model to flag additional critical gaps (Gap > 30%) not covered before.
Output format
Use structured tables with the following columns: Entity | Cost Center | Job Profile | Demand | Supply | Gap | Gap %
Recommendations
For largest gaps, suggest potential actions: open positions, internal transfers, or scope adjustments.Limitations
Shared Scenarios are supported, but local Scenarios are not.
Member-based variables are not supported.
Variables with names containing special characters are not supported (for example,
[d]CurrentMonth).The Agent can only access Metrics with data types Text, Number or Integer.
Certain advanced calculations are not supported:
Time series analysis
Correlations
You cannot reference Views in the Instructions.
Scenario Variables that reference a snapshotted Scenario are not yet supported.