Extract figures with page evidence
Extract numerical facts without losing units, scope or page references.
45 minutes estimated—not learner-validated. Original learning materials are free; optional external AI access may require an account, credits or payment.
Start the editable activity →No account needed to begin. Explicit account saving and editable offline export are available in the workspace.
New here? Try the three-activity starter ↗
Why this matters
PDF text can be rearranged, OCR can misread digits, and annual and quarterly totals are not directly comparable. This fixture is a short original report; use it first, then optionally test a public document you are allowed to upload. Every number needs a source location and an uncertainty label.
Before you start
Materials are free. External tools may require an adult account, credits, regional eligibility or a paid plan. Check current terms before using them. Do not paste passwords, API keys, private messages or other people’s personal data. No model runs locally or downloads automatically on this platform.
Your illustrated field guide
Conceptual illustration, not a provider screenshot. Artwork is AI-generated; instructional labels are authored text.
Start with fictional input
Fictional report, page 1: “Technology investment, FY2025: $120,000; FY2024: $95,000.” Page 2: “Marketing spend, Q1 2025: $18,000. Marketing spend, Q2 2025: not disclosed.” Page 3 note: “Figures are USD, before tax; investment excludes recurring software subscriptions.”
Copy and adapt this prompt
Extract technology investment and marketing spend into a table: metric, period, value, unit, page, exact supporting quote, exclusions/uncertainty. Keep not-disclosed values as null/not disclosed. Do not estimate missing quarters, combine periods or convert currencies.
Worked reference—not proof of correctness
Technology investment | FY2025 | 120000 | USD before tax | p1. Marketing spend | Q2 2025 | not disclosed | USD | p2. A yearly marketing total cannot be calculated from a single disclosed quarter.
Five practical steps
Read the fixture first
Build a manual reference table, including the scope note and missing value.
Run extraction
Paste the page-labelled text or upload the original fixture report from the pack where supported.
Verify each cell
Trace every figure to the exact page and quote; compare zeros, commas and period labels.
Introduce an OCR trap
Replace 120,000 with 12O,000 using the letter O. Check whether it flags ambiguity instead of silently asserting a value.
Extend carefully
Try a permitted public report and record the document version, page numbering and extraction limits.
Test your result
- Every numeric row has the correct period, USD unit and page/quote evidence.
- Not disclosed remains missing; no annual marketing total is invented.
- The OCR ambiguity and subscription exclusion are visible in the final notes.
Mistakes to watch for
- Comparing a quarter with a full year as though they share a denominator.
- Confusing the PDF viewer’s page number with the printed page label.
Offline / no-key route
Complete the reference table manually from the supplied page-labelled text; OCR and upload features remain untested.
Reflect
Which number would be most misleading if its period label disappeared?
Keep your notes, tests and self-review →Self-reported practice is not independent assessment or a qualification. Browse all 50 guides · Choose a tool