Call Detail Analyzer

Drop your call detail CSV here (Files are not uploaded; all computation happens in your browser.), automatically detects headers and provides: success rate, failure rate, P50/P95 latency, performance per tool, top callers, andWhich callers are anomalous?— abnormally high failure rate, excessive concentration of calls, short-term spikes, and persistent freeloading without payment. Finally outputs an analysis report ready to send directly to teams or customers.

Free · No login required Local parsing · Files are not uploaded Auto-detect headers · Column names do not need to be modified

1Call logs

Paste or upload a CSV. Headers are auto-detected (supports both English and Chinese for Time / Caller / Tool / Status / Latency / Amount / UA).

CSV content

2Anomaly Detection Threshold

Caller Failure Rate Threshold (%)

Call Volume Concentration Threshold (%)

P95 latency alert threshold (ms)

Hourly Surge Multiplier

Anomalous callers fall into four categories: excessively high failure rate / excessively high share / short-term spikes / fully free (many calls but $0 amount).

ResultWhat do these calls look like?

Distributionby tool / by caller / by hour

OutputAnalysis report (deliverable)

0 rows

—

We’ve fallen into this pitfall

1 · Daily aggregation cannot incrementally accumulate.The first version, running once per day, doubled PV (1654 vs. actual 415). Must.Recalculate daily coverage; historical data is derived from the summary table, ensuring idempotence.

2 · Formatting timestamps before applying time-window filtering will cause failures.Date string missing .timestamp(), stringification must occur after all aggregationsAfter。

3 · Regular expressions for UA detection must be complete.Exact match only bot then only the UA contains spider / crawler will be misidentified as "real browser".

4 · Do not reuse other fields when taking the maximum value.We reused the `hours` value extraction function when plotting trend bars, but the `daily` structure lacks that field, resulting in bar height calculated as 25100%.