🔢 DeckMedic · Data Engine · Tool 1 of 2

PPTX Decimal Normalizer
Fix Inconsistent Number Formatting in AI Tables

AI-generated financial tables are inconsistent. Row 1 says "$10.00", Row 2 says "10", Row 3 says "$10.001". Your MD will kill you for it. This tool enforces uniform decimal precision across every table cell in your deck — in seconds, with a before/after audit trail.

🔒

Zero upload

File stays on your device

Instant

No server round-trip

🆓

Free

No sign-up required

📊

Tables only

Text outside tables untouched

Target decimal places

Selected: force 2 decimals on all numbers found in table cells. Currency symbols, commas, and % signs are preserved.

📊

Drop your .pptx here

or click to browse

Runs 100% in your browser — file never leaves your device

What this tool actually does

← drag handle to compare  ·  left = before  ·  right = after →

✕ Before
  • Slide 3, TableRow: $10 / 10.0 / $10.001
  • EBITDA margin12.5 / 12.50 / 12.5000
  • EPS (FY25E)2.3 / 2.30 / 2.3
  • Revenue ($B)4 / 4.0 / 3.99500
✓ After
  • Slide 3, Table$10.00 / $10.00 / $10.00
  • EBITDA margin12.50 / 12.50 / 12.50
  • EPS (FY25E)2.30 / 2.30 / 2.30
  • Revenue ($B)4.00 / 4.00 / 4.00
📖 How this tool works · Tips · Browser Privacy

Decimal Normalizer — How it works

  1. 1

    Unzip + parse

    The PPTX is opened as a ZIP archive. Every slide XML file (ppt/slides/slideN.xml) is read as a UTF-8 string.

  2. 2

    Locate all tables

    The tool finds every <a:tbl> (table) element within each slide. Each table is scanned row by row, cell by cell.

  3. 3

    Identify numeric cells

    Each cell's text content is extracted and tested against a strict pattern: optional currency prefix, numeric value, optional % suffix. Cells containing labels, dates, or mixed text are skipped.

  4. 4

    Apply toFixed()

    Numeric cells are reformatted using JavaScript's toFixed(n) — the currency prefix and % suffix are preserved. The change is recorded in the audit log.

  5. 5

    Rebuild and download

    The patched XML is written back into the ZIP, and a clean PPTX is generated for download with the _decimals_fixed.pptx suffix.

What this tool skips

Text outside tables

Numbers in text boxes, titles, and body text are not touched — only <a:tbl> table cells.

Header rows & labels

If a cell contains any non-numeric text ("Revenue", "Q3", "FY25 E"), it is left unchanged.

Mixed-content cells

"Revenue: 10.5M" fails the pure-number test and is skipped entirely.

4-digit years (1900–2099)

"2024", "2025", "2030" — year values in column headers are detected and preserved as-is.

Tips for best results

  • Run the Decimal Normalizer immediately after AI generation — before peer review or design formatting.
  • Use 2 decimal places for per-share values (EPS, DPS) and currency line items.
  • Use 1 decimal for margin percentages (EBITDA %, revenue growth rates, etc.).
  • Check the audit log before downloading — every changed cell is listed by slide and table row.
  • Only table cells inside <a:tbl> elements are modified — text boxes and title shapes are untouched.

Privacy & data security

🚫

Zero upload

No file ever sent to a server

📴

Works offline

After first page load

🔇

No telemetry

File data never logged

GDPR safe

No data processor needed

Browser compatibility

Chrome 112+Edge 112+Firefox 111+Safari 16.4+iOS Safari 16.4+Chrome Android 112+

Decimal Normalizer vs. alternatives

Method Decimal Normalizer Manual Edit VBA Macro
Processes all slides at once ✅ Yes ❌ One cell at a time ✅ Yes
Before/after audit trail ✅ Full log ❌ None ❌ No log
Preserves currency symbols ✅ Yes ✅ Manual ⚠️ Depends on macro
Zero install ✅ Browser-based ✅ Built into PPT ❌ Requires PowerPoint + VBA
Works on Mac / Linux ✅ Yes ✅ Yes ❌ Windows only
Handles 50-slide deck ✅ Seconds ❌ Hours ✅ Seconds

Who uses the Decimal Normalizer

🏦

Investment Banking Analysts

Pitch books and CIM documents require strict house style: 1 decimal for multiples, 2 for per-share values. AI-generated tables violate this constantly. One run before printing fixes every table in the deck.

📊

Equity Research Associates

Initiation reports and earnings models live and die by consistent decimal formatting. Mixed precision reads as sloppy to institutional clients and portfolio managers reviewing the stack.

🏢

Strategy Consultants

Client-facing board decks, QBR summaries, and market sizing models go through strict QA. A number formatted as "2.5" next to "2.50" next to "3" on the same table is a last-minute fire drill — this eliminates it.

Frequently asked questions

What numbers does the Decimal Normalizer change in a PowerPoint?
The tool scans every <a:t> text node inside <a:tbl> table elements across all slides. It targets cells that contain a pure number string — including integers, decimals, currency-prefixed values ($10, €10), and percentages (10%). It reformats the numeric value to the selected decimal places using toFixed(), preserving the original prefix and suffix.
Why do AI-generated tables have inconsistent decimal formatting?
AI models generate numbers from different data sources within the same prompt — one value from a spreadsheet snippet, another from a conversion, another from direct generation. Each source uses its own rounding. Investment banks and research firms have strict house style rules requiring uniform decimal places — typically 1 or 2 for financial data.
Does this tool modify text that is not a number?
No. The tool applies a strict regex pattern before touching any cell. Only cells whose entire trimmed text matches the pattern of a numeric value (optionally prefixed with a currency symbol, optionally suffixed with %) are modified. Labels, names, mixed-text cells, and standalone 4-digit years (2022–2030) are left completely unchanged.
What decimal precision should I choose for a financial model?
For investment banking and equity research: use 1 decimal for percentage metrics (EBITDA margins, growth rates), and 2 decimals for per-share values (EPS, DPS) and currency amounts. For deal summaries and board presentations, 0 or 1 decimal is standard.
Is my PPTX uploaded to a server?
No. All processing runs in a Web Worker inside your browser using JSZip. The file is read locally, the XML is patched in memory, and the fixed file is downloaded directly. No data leaves your device.

Data Engine · You might also need

🗑️

Ghost Box Buster →

Physically delete every empty text box shape left behind by AI generators and template editors.

🕵️

Metadata Wiper →

Strip author, company, and AI session IDs before delivering the deck to a client.