AI Playbook

Design a Baseline Survey with AI

6 steps · Works with any AI assistant · No signup required

You'll end up with

A field-ready survey instrument with indicator-linked questions, skip logic, sampling frame, and pilot testing protocol.

Map Indicators to Questions

Before drafting survey questions, map each indicator in your results framework to the specific questions that will measure it. This ensures every question earns its place and every indicator has a measurement path. Paste your indicators below the prompt when you run it.

Your indicator list*

The AI will map each indicator to 1-3 survey questions with response formats and validated instrument references.

Prompt for this step

You are a senior M&E survey specialist. Your task is to map each of my indicators to the specific survey questions needed to measure it, using the program brief above as context. I will paste my indicator list below.

For each indicator, produce a labelled section with:

1. **Indicator name and definition** (restate exactly as provided)
2. **Survey questions** (1-3 questions that measure the indicator directly)
   - Exact question wording in plain language
   - Response format for each: single-select, multi-select, Likert scale, numeric, or open-ended
   - If closed-response, list the answer options
3. **Indicator contribution** for each question
   - Which indicator value this question feeds: numerator, denominator, or disaggregator
   - How the question's responses will be aggregated to produce the indicator value
4. **Validated instrument reference** where one exists
   - Name the standardized instrument (JMP water-source ladder, WHO Food Consumption Score, WEAI empowerment modules, MICS child-nutrition modules, Washington Group disability questions, PHQ-9, etc.)
   - Cite the source so the user can locate the full instrument
   - If no validated instrument exists for this indicator, state "No standardized instrument identified; custom wording drafted"
5. **One question-design risk with mitigation**
   - Name the risk type precisely: leading wording, double-barrelled construction, recall bias, social desirability bias, ceiling or floor effect, or sensitive-topic disclosure risk
   - State the specific mitigation applied in your wording

Apply these principles throughout:
- Prefer validated instruments over custom wording wherever they exist
- Match question grain to indicator grain (output-level questions for output indicators, outcome-level for outcome indicators)
- Keep recall windows realistic: 24 hours for dietary intake, 7 days for routine behaviors, 12 months for rare events
- Flag any indicator that cannot be measured through survey methods and suggest an alternative data source

Output as one labelled section per indicator, with clear headings. Use structured lists, not tables. Number the sections sequentially.

My indicators:
[PASTE YOUR INDICATOR LIST HERE]

Step 1 of 6