GA4 Ecommerce Event Payload Auditor

What it does

Audit supplied ecommerce event payloads before a tracking release or during instrumentation QA. Find missing item identity, transaction identifiers and currency, wrong numeric types, inconsistent reported values, and repeated purchase identifiers. The report returns exact source field paths and review actions without contacting Google Analytics, Tag Manager or a customer website.

Input

Supply1to500 JSON records in events and explicitly choose format. ga4-events expects each row to have name and params, matching the event objects inside a Measurement Protocol request or an exported gtag call. data-layer expects event and ecommerce. Context pushes without event, including ecommerce:null clears, are informational. The tool audits each resolved event independently and does not simulate dataLayer merging or replay browser state.

Supported event names: view_item, add_to_cart, remove_from_cart, view_cart, begin_checkout, add_shipping_info, add_payment_info, purchase and refund. Other syntactically valid events are marked OUT_OF_SCOPE. Missing or malformed supported payload fields are findings, rather than reasons to silently drop a row. Invalid batch shape and tool limits fail before any report billing emission.

The limits are500 records,5000 aggregate item entries,3MB total input and7MB serialized report. More than200items within one event is a platform-limit finding, while all supplied items inside the tool batch cap are still inspected. maxEvents lowers the batch cap. amountTolerance is an explicit absolute value from0to1, default0.01; no currency-dependent precision is guessed. Arithmetic rounds amounts and item products to six decimal places, with BigInt accumulation to avoid total overflow; this is an audit convention, not a financial ledger.

Output

One dataset report includes findings, exact JSON source paths, counts, and record-level arithmetic status. AUDIT is complete JSON, FINDINGS_CSV is a downloadable correction checklist, and REPORT is Markdown. recordCount includes context pushes; eventCount excludes dataLayer pushes without event. ERROR and REVIEW statuses concern this documented subset, while CHECKED never means Google acceptance.

The reconciliation basis is supplied item price times quantity, excluding separately supplied tax and shipping. Missing quantity defaults to1; explicit zero is flagged for review and is not replaced. Positive values below audit precision are excluded with an explicit precision warning. Fractional numeric quantities are supported. Item price is already discounted, so discount is not subtracted again. Missing price leaves the total incomplete. A zero value is not treated as missing. Numeric strings are identified as type problems.

Repeated purchase identifiers with equal supplied totals produce context notes, because repeats can be retries and GA4 uses transaction identifiers for deduplication. Identity uses the exact supplied nonempty string; surrounding whitespace is flagged separately and never trimmed or merged. Different totals or currencies produce review warnings for all affected rows. Item order alone does not create a conflict. A refund may omit items for a full-transaction refund; a purchase absent from this sample produces an informational note, not a claim that the refund is invalid.

Pricing and cost control

USD0.03 per completed ecommerce-audit-report, covering the declared batch limits. Check the current Store price before running. The cost estimate in PRICING is a planning assumption. No paid API, proxy, model inference, tracking hits or ad spend is required. The report event is emitted only after exports are saved. Structural failures and failed storage do not emit that report event.

Limitations and fair use

This is offline payload QA, not an official Google validation service or an audit of actual revenue, consent, attribution, delivery, deduplication behavior or account configuration. It implements the documented field subset and syntactic uppercase currency codes, not a complete ISO currency database. Other recommended or custom parameters are outside scope. No campaigns, budgets or tags are changed.

Raw payloads, order IDs and unrelated customer fields are not copied into reports. Transaction identifiers are represented by SHA256 hashes, which reduce direct exposure but are not anonymization. Input is still stored under Apify run-retention settings; use authorized, minimized or synthetic samples and omit secrets or personal customer records.

Example

examples/input.json includes a valid purchase, a purchase value including tax and shipping, an event missing currency and a refund whose purchase is outside the sample. It returns one error, one warning and one context note. examples/input-data-layer.json shows a legitimate clear push and resolved purchase. Run npm ci --ignore-scripts, npm test, then the existing factory release.mjs --local-run --write-example. Canonical source stays here; Apifi boty is only a generated view.

References

Official Google event reference: https://developers.google.com/analytics/devguides/collection/ga4/reference/events . Ecommerce implementation guide: https://developers.google.com/analytics/devguides/collection/ga4/ecommerce . Current official text remains authoritative and provider requirements may change.