BOM Change Analyzer — CSV, Excel & Cost Analysis

Compare Engineering BOM Revisions, Track Component Changes and Analyze Cost Differences.

Compare two flat bills of materials from uploaded CSV/XLSX files or JSON arrays. Find added, removed and modified components; review quantity and unit-price effects; download JSON, CSV and Excel reports. No AI service, ERP connection or external API key is required.

Quick Start

1. Upload Old BOM file and New BOM file in Input. CSV and XLSX may be mixed.
2. Leave the alternative JSON and pasted CSV fields unset. Select an Excel sheet if the workbook has multiple sheets.
3. If needed, map column names and set a common currency such as EUR.
4. Run the Actor. Download Full report (Excel), Parts report (CSV), Summary (CSV) or the complete JSON report from Output.

For the JSON API, keep sending oldBom and newBom. No migration is required:

{
  "oldBom": [
    {"part_number": "A-100", "quantity": 2, "unit_price": "10.00"},
    {"part_number": "B-200", "quantity": 1, "unit_price": "5.00"}
  ],
  "newBom": [
    {"part_number": "A-100", "quantity": 3, "unit_price": "12.00"},
    {"part_number": "C-300", "quantity": 3, "unit_price": "2.50"}
  ],
  "currency": "EUR"
}

Expected: one added, one removed, one modified part. Old cost 25.00 EUR, new cost 43.50 EUR, increase 18.50 EUR (74%). A-100 contributes +16 EUR: +10 from quantity and +6 from price. C-300 adds 7.50 EUR and removing B-200 saves 5 EUR.

Input sources

Supply exactly one source per side:

Source	Old field	New field	Format
Native uploaded file	oldFile	newFile	File URL string from the Apify upload widget; a one-element URL array is also accepted.
Existing JSON API	oldBom	newBom	Parts array; empty arrays supported.
Pasted CSV	oldCsv	newCsv	UTF-8 CSV text including a header.

Source requirements are validated by the Actor, because JSON arrays and uploaded files are alternatives. The Input schema intentionally does not mark both JSON arrays as required. Missing or conflicting sources fail with a specific error. Start fees apply to runtime validation failures.

Uploaded files must be accessible HTTPS Apify key-value-store records or S3 objects, optionally with a signed URL. Redirects, credentials in URLs and arbitrary external hosts are not accepted. No account token is attached to downloads. File formats are detected automatically, or use oldFormat / newFormat (auto, csv, xlsx). Legacy .xls is unsupported.

Columns and validation

Each part requires a nonempty text part_number. Matching trims surrounding whitespace and ignores Unicode case. Duplicates after normalization are rejected. Numeric Excel identifiers must be converted to text before upload to preserve leading zeroes.

Optional fields: quantity, unit_price, description, currency, unit. Quantity and price accept nonnegative finite numbers or decimal numeric strings. Booleans, negative values and non-finite numbers are rejected. Empty CSV/Excel cells become missing values, never zero. Formulas, Excel error cells, macros and external workbook links are rejected: upload calculated values only.

Common headers are detected, including Part Number, Part No, PN, MPN, SKU, Quantity, Qty, Unit Price, Unit Cost, Price, Description, Desc, Opis, Currency, UOM and Unit. Case, spaces and punctuation do not affect header matching. If multiple columns match a field, specify a mapping; the Actor does not guess.

{
  "oldColumnMapping": {
    "part_number": "Component ID",
    "quantity": "Count",
    "unit_price": "EUR/unit"
  }
}

Use separate oldColumnMapping and newColumnMapping objects when layouts differ. Map to exact header text. Other columns are ignored on file import. JSON extra fields remain preserved on added/removed records and do not trigger changes.

CSV accepts UTF-8 with or without BOM; comma, semicolon, tab and pipe delimiters. Set csvDelimiter if needed. Decimal separator is a dot; thousands separators and currency symbols are not automatically removed. XLSX requires a single worksheet or an explicit oldSheet / newSheet name.

Financial analysis

Line cost is quantity × unit price. Decimal arithmetic is used; financial results are returned as decimal strings, preserving precision without binary floating-point rounding. Percentage change is rounded to six decimal places using half-even rounding.

• Missing quantity or price makes that part unpriced, including a zero quantity with a missing price.
• old_total / new_total are null when the corresponding BOM is incomplete. known_old_subtotal / known_new_subtotal contain only priced rows and are explicitly not full costs.
• difference and percentage_change are null if overall comparison is incomplete or units/currencies are incompatible. Percentage change is also null for a zero old total.
• currency supplies a common three-letter code for missing row currencies; it never overrides an explicit row value. Mixed currencies, or a mix of declared and undeclared currencies without a common currency, disable aggregate costs and cross-currency ranking. No conversion is performed.
• If every currency is omitted, calculations assume one common unspecified currency, and the report includes that assumption. Verify the assumption before making financial decisions.
• unit describes the unit in which quantity and unit price are expressed. A unit mismatch between matched revisions, including declared versus undeclared, prevents a comparable financial impact. No conversion is performed. Different components may have different units.
• Numeric costs are supported up to 100 significant digits and decimal exponents of magnitude 100. Values beyond this range still compare through the legacy JSON comparator but are marked unpriced in financial analysis.

For matching comparable parts:

quantity_effect = (new_quantity - old_quantity) × old_price
price_effect = new_quantity × (new_price - old_price)
net_impact = quantity_effect + price_effect

This convention assigns the price/quantity interaction to the price effect. Added and removed parts have a net impact without artificial price/quantity decomposition. topN (1–100, default 10) ranks nonzero comparable impacts by absolute magnitude. Incomplete rankings are labelled ranking_complete: false.

Reports and API compatibility

OUTPUT retains the existing summary and changes fields and adds report_version, parts and cost_analysis. Existing change types, matching rules, ordering and dataset row types remain unchanged. Integrations should ignore new JSON fields.

The default dataset contains one record_type: summary row and change rows only, as before. OUTPUT.parts and file reports include all parts with added, removed, modified or unchanged status. Only quantity, price and description affect the legacy change status; currency/unit metadata additionally determine financial comparability.

Record	Contents
OUTPUT	Full JSON: legacy summary/changes plus costs and all-part detail.
REPORT-CSV	UTF-8 BOM CSV with all parts, quantities, prices, line costs, differences and effects.
SUMMARY-CSV	Summary counts, cost completeness, totals and warnings.
REPORT-XLSX	Excel workbook with Summary, Parts and Top cost drivers sheets.

Exports escape formula-like text to avoid spreadsheet formula execution. Exact signed decimal values remain numeric text. CSV/JSON preserve long text; Excel follows its 32,767-character cell limit. Reports do not calculate engineering equivalence, assembly hierarchies, supplier substitutions or unit/currency conversion.

Limits, pricing and data handling

• Up to 10,000 parts per input, 10 MiB per file, 128 file columns, 64 MiB expanded XLSX content and 2,000 ZIP entries. XLSX has a header row plus up to 10,000 data rows.
• Default runtime allocation is 256 MiB with a 120-second timeout. Hosted 10,000-part CSV/XLSX benchmarking is recorded in the release evidence. Input size and content affect runtime costs.
• Existing price remains $0.01 per start up to 1 GB RAM, with no per-result charge. Apify's start event is billed once per started GB (minimum one). Failed runtime validation also incurs a start event. Platform usage is included; always check the live Pricing tab.
• Inputs and reports are stored on Apify under your storage/access settings. Uploaded files are retrieved from their supplied storage URLs; no data is sent to AI services. Upload only information you are authorized to process.
• Logs contain counts and safe validation messages, not BOM contents or signed upload URLs. Reports intentionally contain the submitted part details; control access to their storage.

Examples, support and development

The repository examples/ contains JSON input/output, sample old CSV/new XLSX and rendered reports. For support use the Actor Issues tab with a run ID and a minimal synthetic example; do not attach confidential BOMs or credentials.

Run tests with python -m unittest discover -s tests -v after installing requirements.txt. Production releases use a separate candidate build, hosted validation, unchanged pricing and a saved previous build for rollback. See docs/releases/ for audit and measured results.