Skip to workflow playbooks
Report WatchBY MAYD-IT

PRACTICAL PLAYBOOKS · HUMAN SETUP, AUTOMATED CHECKS

Check the report
at the end of the flow.

Monitor final CSV reports from n8n, Make and Apify. Check the file after the joins, filters and field mappings—then keep evidence for the period it belongs to.

Try a file locally Choose a playbook ↓

Setup recipes with fictional examples and an importable manual n8n workflow. Your own connection still needs configuration and acceptance.

CHECK THE FIT FIRST

Small recurring files.
Specific expectations.

A useful fit

  • A daily or weekly CSV with an agreed timezone and deadline.
  • Known required headers, a row range and, optionally, dates in the data.
  • A final file within 1 MiB, 100 columns and 10,000 rows.
  • Someone responsible for investigating a failed or missing report.

Choose another control for

  • Reconciling money, identifying missing source records or checking every field’s type.
  • Proving that an email reached a client’s inbox.
  • Large datasets, sub-minute alerts or a strict delivery-time SLA.
  • A single file that only needs the free browser check.
01

AGENCY CLIENT CSV

Agree on the deliverable,
then check that file.

A client report may combine a spreadsheet, an API and a saved export. Define the expectations for the finished CSV separately from the success of each input step.

Input contract

Authorized source extracts are joined and reduced to one client-facing CSV. Choose an explicit reporting period and decide when zero rows is legitimate.

Output contract

For this example: require report_date, client and orders; accept 1–100 rows. The check receipt belongs to that report’s intended scheduled deadline.

Set it up

  1. Write down the contract. Record required headers, the expected row range, timezone, daily/weekly deadline and grace period. A row range does not prove every client is represented.
  2. Choose dates deliberately. If freshness matters, use a date that describes the underlying data. Every row must meet the age rule. Adding today’s export time would only prove when you generated the file.
  3. Check the final bytes. Place the check after merges, redaction and CSV serialization. If you change the file afterward, that later file has not been checked.
  4. Keep delivery separate. Save the receipt and send only under your own approved workflow rules. Track destination delivery with the sending system.

Download the fictional CSV The timestamp is fixed to October 12, 2026. It is a teaching fixture, not a file guaranteed to pass a freshness check today.

When your current tools are enough: if the report already has reliable scheduled checks, retained evidence and an investigation path, an additional service may add little. For a one-time review, use the free checker. A heartbeat service can cover missing job signals; Healthchecks also supports payload keyword filtering.

02

N8N + MAKE

The HTTP request is a step.
Its result needs a branch.

Use the final CSV produced by your existing workflow. A successful HTTP request can return a failed quality check, so the next step must read the JSON result.

Input contract

One immutable submission envelope: the final CSV, monitor ID, intended deadline and stable submission ID. The report token stays in protected workflow credentials.

Output contract

Require the expected JSON receipt and checkId. Branch on status; inspect late separately. HTTP 200 with status: failed is saved failure evidence.

Wire the manual HTTP step

  1. Finish the transformation first. In n8n, use the HTTP Request node; in Make, use the HTTP request module. Send raw UTF-8 CSV to the monitor’s ingestion URL with Content-Type: text/csv.
  2. Map the protected credential and saved headers. Send the report’s bearer token, Idempotency-Key and the intended deadline as 13-digit Unix milliseconds in X-Report-Deadline. Do not add grace to that deadline.
  3. Disable redirects and bound the request. Use a timeout. On an uncertain timeout or transient failure, retry only a limited number of times with the same saved CSV, deadline and ID.
  4. Route the outcome. A failure goes to correction/review. A corrected file uses a new ID and the same original deadline. An identical retry returns the original result; it does not evaluate the file again.

Set up the importable n8n workflow Manual trigger, one request attempt and a receipt classifier. Keep it unpublished while configuring it. The guide records its verification status; no schedule, automatic retry or downstream action is included.

Download the n8n workflow JSON, then select your protected report credential and replace the envelope placeholders. For manual node setup or Make, use the configuration reference. The agent guide includes a bounded Node example.

When native handling is enough: n8n has error workflows and deliberate failure branches; Make has stored incomplete executions and retry handling. Keep them. If your own validation and independent deadline checks already meet the need, you may not need Report Watch. This recipe is not an installed integration or an end-to-end tested customer workflow.

03

APIFY EXPORTS

Check the export after
fields and rows change.

A successful Actor run and a usable downstream file are separate checkpoints. This playbook is for a small CSV after field selection, merging or mapping outside the source dataset.

Input contract

An authorized dataset export, transformed into a deliberately scoped CSV. Preserve the fields the recipient expects and the observation timestamp actually recorded by the source workflow.

Output contract

For this example: require url, title and observed_at, with 1–500 rows. An optional age rule checks every observation date, not the age or completeness of the website’s content.

Choose the right checkpoint

  1. Inspect the export settings. Apify can export datasets as CSV and select or omit fields. Confirm those choices against the final report contract using the official dataset documentation.
  2. Complete downstream mapping. Flatten and rename fields explicitly. Check the resulting CSV after those changes; validating only the original dataset would not test a later mapping.
  3. Keep the scope honest. Make the intended report fit the size limits. Do not silently truncate a larger dataset just to obtain a pass. Use native dataset checks for the larger source.
  4. Submit the final file. Use the same token-scoped submission API, intended deadline and bounded retry rules. Keep the Apify credential separate; Report Watch does not need it or a private dataset URL.

Use the synthetic file below in the free checker with the stated required headers and row range. No date rule is needed to demonstrate this missing-header failure.

Start with Apify’s native option when it fits: the Apify Monitoring Suite already checks run status, dataset schemas, item counts and duplicate keys, with email or Slack notifications. It uses platform compute. Choose it for source-dataset monitoring; consider this final-file check when your report is transformed afterward or combines multiple tools.

ONE REPORT BEFORE A WHOLE PORTFOLIO

Make the acceptance test
part of the setup.

Try a synthetic pass, a known failure, an identical retry and a correction. For monitoring, also verify the intended deadline, saved evidence and alert recipient through your own workflow. Record false alerts and missed known issues before expanding.

Download the blank pilot scorecard. This is a planning worksheet with placeholder rows, not customer results or a Report Watch import file.

Keep the boundaries visible.

Bounded files and history. UTF-8 comma-delimited CSV: 1 MiB, 100 columns, 10,000 data rows; headers up to 100 UTF-16 units and cells up to 16,384. A workspace allows 1,000 new checks per rolling 24 hours and 5,000 saved checks total. Failures count. Waiting does not clear the total-storage cap.

Credentials stay separate. Keep the report token in a protected credential store, outside prompts, source files and exports. Never send owner credentials or forward a token through redirects. Agents submit only to an existing authorized report; they cannot use this API to sign up, pay or configure monitors.

Evidence has a scope. Raw submitted CSV rows are processed in memory; normalized headers, rules and check evidence are retained. A pass does not prove business accuracy, source completeness or inbox delivery. Background checks and email take time; this early-access service has no strict delivery-time SLA. See the workspace privacy policy and free-tool data notice.