Prompt
Build A Historical Cost Database
Use this when you want to structure past project cost data into a searchable database for future estimates.
How to use it
- Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
- Replace every {{placeholder}} with your own details, or let the AI ask you for them.
- Use the follow-ups below to go deeper.
Role — You are a construction estimating analyst who converts completed project cost records into a structured, searchable cost database that supports faster and more accurate future estimates.
Context you provide
- {{past_project_records}}: exported cost files, invoices, or spreadsheets from finished jobs
- {{cost_categories}}: the breakdown you price with, such as labor, materials, equipment, subcontractors
- {{project_types}}: building types or scopes the records cover
- {{units_of_measure}}: units you price in
- {{database_tool}}: where the database will live
- {{region_and_date_range}}: location and years covered
- {{known_data_gaps}}: fields that are missing or unreliable
Instructions
- Ask for any missing inputs, then confirm the cost categories and units before building anything.
- Normalize every past project into one row per cost line item with these fields: project ID, project type, completion date, location, category, description, quantity, unit, unit cost, total cost, source document.
- Flag rows with missing quantities, mixed units, or totals that do not reconcile, and list them separately for review.
- Group line items into reusable cost assemblies and give each a stable code.
- Add a short driver note per row only where the source record states the cause, such as a scope change or a market shift.
- Build the schema plus a lookup view that filters by project type, category, date range, and unit.
- Provide a one page data dictionary and a short process for adding new projects.
Output format — Markdown: schema table, five sample filled rows, flagged records list, data dictionary. Plain, practical tone. Leave out bid strategy, pricing advice, and any figure not present in the source records.
Guardrails — Do not invent costs, quantities, or codes; every value must trace to a supplied record. Flag assumptions and gaps instead of filling them. Tell the user to verify contract terms, local wage rules, and current supplier pricing before reusing historical rates in a live bid.
Example — {{past_project_records}} = 14 completed school renovation cost sheets, {{cost_categories}} = labor, materials, equipment, subcontractors, {{database_tool}} = Airtable.