Anomaly detection · Planned
Anomaly detection is planned to mark values that depart clearly from their own history, with a reason: which comparison period, which data basis, how large the departure. It is meant to tell data problems apart from real changes. It does not exist yet. Today the pivot shows you the largest value, and empty, null and missing stay distinct.
Today: check data qualityPlanned, not available. In the preview: largest value with its share, visible gaps in the data and marked cut-outs for very large sources.
01 / Example value
Which number would make you nervous if it suddenly doubled?
Receipts: expenses by counterparty and month · Mark (planned): departs from its own history · “Check”: receipt entered twice?
Change the example value. The finding follows the value.
Fictional example · no real data
Two sample questions
And who would notice today?
Are marketplace refunds unusual this week?
Receipts: expenses by counterparty and month · Mark (planned): departs from its own history · “Check”: receipt entered twice?
Orders: refunds by marketplace and week · Mark with comparison period (planned) · Today: largest value with its share
Honest about the status
01 / What is settled today
This capability is specified and not built yet. There is no date and no separate price.
02 / Your control
Businesses and numbers on this page are made up and marked as examples. None of it is a measurement or a customer result.
Compare both evaluations and their reasons.
03 / The clear limit
Peak Analytics is available as a preview: pivot evaluations over orders, receipts, tasks, goals and files, CSV export and sharing.
Your next step
The data quality guide shows how to spot gaps, duplicates and cut-outs. Without that, any anomaly detection gets noisy instead of helpful.
Today: check data qualityPlanned, not available. In the preview: largest value with its share, visible gaps in the data and marked cut-outs for very large sources.
No. This capability is planned and not built yet; there is no launch date. Available today is the Peak Analytics preview: pivot evaluations over orders, receipts, tasks, goals and files with CSV export and sharing.
The plan is to compare it with the measure's own history, with a visible comparison period, not with invented industry figures.
Empty, null and missing stay distinct in Peak Analytics, and very large sources are marked as a cut-out.
The plan is the same path as “Check” today: the finding becomes a task in Peak Task, not just a notification.
