CSV Quality Scorecard

Compute a data-quality scorecard for CSV rows before they hit Sheets, Airtable, Supabase, or your CRM.

What it does

• Detects missing required fields.
• Validates email columns.
• Validates URL columns.
• Flags duplicate rows.
• Computes completeness % and a 0–100 quality score.
• Emits one summary row per source plus one issue row per problem.

Use cases

• Inbound lead CSV QA before CRM import.
• Vendor catalog health checks for marketplaces.
• E-commerce data audit (missing price/availability/SKU).
• CRM cleanup sprints — score, then route bad rows to a human queue.

Output

• One summary row per source with qualityScore, completeness, duplicates, summary (high|medium|low|critical), and per-row issue list.
• One issue row per problem with source, _row, issues[], filled (non-empty cell count).
• SUMMARY with overall score across sources.

Pricing / cost expectations

Default 256 MB / 1 GB. Streaming CSV parse (no full-file load); runs of 10–100k rows finish in seconds. Pay-per-event scheduled to activate after public launch: 1 charge per csv_scored event (per source) + 1 per issue_found event (per problem row).

Related Actors

• ../csv-schema-profiler — per-column type inference + format detection.
• ../csv-dedupe-normalizer — dedupe + header normalization.
• ../csv-header-normalizer — normalize header names to a target convention.
• ../csv-delimiter-autodetect — figure out the delimiter + quote char + encoding.
• ../csv-diff — diff two CSVs on a key column.
• ../jsonl-stream-validator — same idea for JSONL/NDJSON.

FAQ

Does it fetch URLs? Yes, via csvUrls[]. Public URLs only, no auth.

Does it work on Excel? Pre-convert to CSV. For programmatic XLSX/XLS extraction use ../xlsx-sheet-extractor.

Does it call out to email/URL validators? No. Bad email/URL detection is regex-based. For deep SMTP/MX checks, run this Actor first, then a custom step.

Safety

Public data only. No login, no external lookups, no personal data enrichment.