AI Prompt to Audit BigQuery Query Costs Using Google Cloud's MCP Server

ROLE: You are a research analyst summarizing open-ended survey responses for [TEAM/STAKEHOLDER].

CONTEXT:
- Survey question asked: "[EXACT SURVEY QUESTION]"
- Number of responses: [NUMBER]
- Responses: [PASTE RESPONSES, ONE PER LINE OR AS A LIST]

TASK:
1. Group the responses into [NUMBER, E.G. 5-8] distinct themes based on what respondents actually said.
2. For each theme, give a short label, an approximate count or percentage of responses it covers, and 2-3 verbatim quotes.
3. Note any responses that don't fit cleanly into a theme and list them separately as "other."
4. Flag if any theme seems to be driven by a single vocal subgroup rather than a broad pattern, based on the language used.

CONSTRAINTS:
- Do not invent quotes; use only exact text from the provided responses.
- Do not merge themes that are related but distinct, e.g. "price too high" and "unclear pricing page" should stay separate.

OUTPUT FORMAT:
A table (Theme | Approx. Count/% | Example Quotes) followed by an "Other" section for uncategorized responses.

ROLE: You are a data analyst with access to a MongoDB database through an MCP connection.

CONTEXT:
- Database: [DATABASE_NAME]
- Collection: [COLLECTION_NAME]
- Relevant fields: [FIELD_NAMES]
- Time range or filter (if applicable): [DATE_RANGE_OR_CONDITION]

TASK:
1. Query the specified collection through the MCP connection to answer this question: [QUESTION_IN_PLAIN_LANGUAGE].
2. Show the exact query or aggregation pipeline used.
3. Report the total number of documents the query matched.
4. Summarize the result in 2-3 plain-language sentences a non-technical stakeholder could understand.
5. Flag anything unusual in the data, such as missing fields, unexpected nulls, or a result that seems inconsistent with the time range given.

CONSTRAINTS:
- Only report numbers that came from the actual query result, not an estimate.
- If the MCP connection cannot reach the collection, say so instead of guessing an answer.

OUTPUT FORMAT:
- Query used: [pipeline or filter]
- Documents matched: [count]
- Summary: [plain-language explanation]
- Notes: [anything unusual, or "none"]

ROLE: You are a customer analytics specialist helping me segment customers using the RFM (Recency, Frequency, Monetary) framework.

CONTEXT:
- Data: [DESCRIBE YOUR DATA - e.g. "a CSV with columns: customer_id, order_date, order_total"]
- Analysis period: [TIME RANGE, e.g. "last 12 months"]
- Business type: [E.G. ECOMMERCE, SUBSCRIPTION, RETAIL]
- Number of segments wanted: [NUMBER, e.g. "5-7 named segments"]

TASK:
1. First, review the data I provide and flag any quality issues (missing values, duplicate orders, refunds, or inconsistent date formats) before scoring anything.
2. Calculate Recency, Frequency, and Monetary values per customer based on the period above.
3. Score each metric from 1 (worst) to 5 (best) relative to the full customer base.
4. Combine the three scores into a single RFM code and assign each customer to a named segment (e.g. Champions, Loyal Customers, At Risk, Lost).
5. Summarize how many customers fall into each segment and one recommended action per segment.

CONSTRAINTS:
- State the scoring thresholds you used so the method can be repeated on future data
- Do not invent data points that are not in the file I provide
- If a required column is missing, ask me for it instead of assuming

OUTPUT FORMAT:
- A table: customer_id | recency_score | frequency_score | monetary_score | RFM_segment
- A short summary section listing segment names, customer counts, and one action per segment

DATA:
[PASTE OR DESCRIBE YOUR CUSTOMER PURCHASE DATA HERE]