Shopify Listing Error Trend Ranker

Upload two or more dated product rejection/error exports from a merchant sales channel. The Actor groups failures by SKU and rule, compares snapshots over time, ranks newly recurring data-quality problems, and outputs action-ready correction rows.

It works only on the data you paste in. It never logs into any account, calls any marketplace API, or accesses the network.

What you get

• Action rows (recordType: action): one per SKU + rule still failing in the latest snapshot, with a status, severity, streak length, first/last seen dates, score, rank, the latest message and a suggested correction.
• Rule trend rows (recordType: rule_trend): per rule, the number of affected SKUs in every snapshot, change versus the previous snapshot, and counts of new, newly recurring, reappeared, chronic and resolved SKUs.
• OUTPUT record in the key-value store: totals, status counts, caveats and limitations.

Statuses

Status	Meaning
new	Failing in the latest snapshot only, never seen before
newly_recurring	No earlier history, then failing in 2 up to newlyRecurringMaxStreak consecutive latest snapshots
reappeared	Failing now, was fixed or absent in the previous snapshot, but seen earlier
chronic	Long streak, or recurring with earlier history

Score = severity weight (error 3, warning 1.5, unknown 1, info 0.5) × status weight (newly recurring 3, new 2, reappeared 2, chronic 1.5) × a small breadth bonus for rules affecting many SKUs. The weights are heuristics, not measured revenue impact.

Input

Copy and edit the prefilled sample. Each snapshot has:

• snapshotDate – YYYY-MM-DD (required)
• channel – optional label; channels are analysed separately
• rows (array of objects) or csv (text with header row), not both

Recognised columns (case-insensitive): SKU (sku, offer_id, item_id, id), rule (rule, error_code, issue, reason), message (message, description), severity (severity, status), field (attribute, field). Rows without a SKU or rule are skipped and reported.

Other options: newlyRecurringMaxStreak, maxActions, minRuleSampleSize, includeRuleTrends, maxRowsPerSnapshot.

Limitations and sample-size caveats

• Use at least three evenly spaced snapshots per channel. With two, "newly recurring" only means "present in both".
• Rules affecting fewer SKUs than minRuleSampleSize are flagged as low sample; percentage changes on small counts are noisy.
• SKUs are matched by exact text. Renamed or re-imported SKUs look like new failures.
• Suggested actions are generic keyword guidance. Always verify against your channel's current requirements.
• Limits: 60 snapshots, 300,000 total rows, configurable per-snapshot row cap; large inputs are truncated and reported.
• Rankings describe data quality trends in your files only. They do not estimate sales or guarantee approval.

Billing

The Actor emits a single analysis-report event per successful run that produces results. Runs that fail on invalid input, or that find nothing to rank, are not charged. Pricing is configured separately in Apify Console, if at all.

Development

Tests are offline: python -m unittest discover -s tests -t .