Build an AI-Assisted Performance Report Workflow

Before using AI, assemble everything you need in one place. This "inputs first" step prevents going back and forth mid-draft.

Checklist of inputs to gather:

• Quantitative metrics: actuals vs. targets for each performance indicator this period
• Milestone status: which milestones were achieved, which were delayed, which are on track
• Narrative notes: any significant events, program changes, or external factors that affected performance
• Challenges: what went wrong or below target, with explanation
• Prior period report: the equivalent section from your last report (for continuity and formatting reference)

Tip: Even rough notes work. "Grant processing time was 22 days vs. 18 day target because of system outage in January" is enough for AI to draft a professional narrative.

Open claude.ai and start a new conversation. Set the context with a single opening message that describes your role, the document type, and the audience:

I am a federal program analyst preparing the Q2 FY2025 performance report for the [Program Name]. This report is submitted to [agency leadership / OMB / Congress — specify]. The audience expects formal federal government prose, factual and specific, with data supporting every claim. I will provide you with raw performance data and notes, and I need your help drafting polished narrative sections.

What you should see: The AI acknowledges the task and may ask clarifying questions. Confirm your inputs are ready and proceed.

Paste your raw metric data and notes. Ask for a specific section draft:

Example prompt:

Here is my raw Q2 performance data. Please draft a 3-paragraph performance summary narrative appropriate for the report. Lead with overall program status (on track / mixed / below target), then summarize key metric results with specific numbers, then address any metrics that missed target with a one-sentence explanation for each.

Raw data:
- Performance Indicator 1: Grants disbursed — Actual: 47 / Target: 45 — Exceeds target
- Performance Indicator 2: Average processing time — Actual: 22 days / Target: 18 days — Below target (January system outage added 4 days avg)
- Performance Indicator 3: Grantee compliance rate — Actual: 94% / Target: 90% — Exceeds target
- Milestone: Annual grantee training completed — Status: Complete (March 15)
- Milestone: FY2026 grant guidelines published — Status: Delayed (targeting May)

What you should see: A polished 3-paragraph narrative that a program official could read directly — factual, specific, and professional.

Paste your notes on challenges and ask for a structured section:

Example prompt:

Draft the "Challenges and Mitigation" section for this performance report. Based on the data I provided, we have two issues to address: (1) the processing time miss due to the system outage, and (2) the delayed publication of FY2026 guidelines. For each: describe the challenge, explain the cause, and describe what we are doing to address it. 2-3 sentences per challenge.

Provide your planned milestones and ask for the forward-looking section:

Example prompt:

Draft a brief "Q3 Outlook" section for this report. In Q3 we plan to: (1) publish the FY2026 grant guidelines (targeting May), (2) complete mid-year site visits to 12 grantees, (3) begin the annual performance data collection process. Write this as 3-4 forward-looking bullet points in formal federal report language.

Copy all the AI-generated sections into your report template. Then:

1. Replace all placeholder names, acronyms, and program-specific terms with the correct terminology
2. Verify every number against your source data — AI might echo numbers back incorrectly if your notes were ambiguous
3. Check that the narrative matches your agency's tone and any required framing (e.g., alignment with strategic goals)
4. Route through your normal review chain before finalizing

Save the prompts that worked well in a Word document or OneNote page. Next quarter, update the data and reuse the prompt structure — your reporting workflow becomes a repeatable template.