Skip to main content
Anomalies tells you when your spend changes in a way you didn’t expect. You hear about a jump (or a drop) within days, not at the end of the month. For each anomaly, nOps shows how much it cost, when it started, and what caused it, so you can quickly decide whether to act.

​
Get started

  1. Sign in to nOps.
  2. In the sidebar, open Inform → Anomalies, or go to clara.nops.io/anomalies.
  3. Look at the summary cards, then open the anomalies with the largest impact first.
Anomalies used to be a tab on the Insights page. It has its own page now, and old Insights links still bring you here. For many organizations, Anomalies is also the page nOps opens when you sign in. See Home page.
You need at least one connected cloud account for nOps to detect anomalies. If none is connected, the page shows how to connect one.

​
Where anomalies come from

nOps brings together anomalies from several sources so you can review them in one list:

​
Your anomalies at a glance

  • Anomalies: how many anomalies match your current filters.
  • Impact: the total estimated extra (or reduced) spend.
  • High severity: how many are Critical or High. Click this card to show only those, and click it again to show everything.

​
Filter the list

  • Date range: Last 7 days, Last 30 days (the default), Last 90 days, or All time.
  • Source: show everything, or just one source.
  • Type: focus on one kind of change, such as a Spike (a sudden one-day jump), a Level shift (spend that stays higher or lower), Unit cost (the same usage costing more), or New (first-time spend).
  • Search to find an anomaly by service, account, or other details.

​
Anomalies table

Each row shows:
  • Severity: Critical, High, Medium, or Low, based on how unusual the change is.
  • Type: what kind of change it is.
  • Period: when the anomaly started and how long it lasted.
  • Impact: the estimated extra spend, or the saving if spend dropped.
  • Scope: the service, account, or tool affected.
  • Peak day: the day spend was furthest from normal.
  • Details: expand a row to see what caused the anomaly.

​
What caused it

Expand a row to see a plain-language summary and a ranked list of causes. Each cause shows the account, region, and usage type involved, plus its share of the impact. Click Open in Explorer on any cause to open Explorer with that date range, service, and cause already filtered, so you can look closer.

​
Keep the list focused

​
Ignore anomalies you’ve already explained

Use the Active and Ignored tabs to set aside anomalies you already understand, such as a planned migration or load test.
  • Select rows and click Ignore selected, or click Ignore all to set aside everything in view.
  • Ignored anomalies move to the Ignored tab and no longer count in the summary cards.
  • To bring one back, select it on the Ignored tab and click Un-ignore selected.

​
Snooze or adjust detection

For nOps AI anomalies, admins can use the actions menu on a row to:
  • Snooze 7 days or Snooze 30 days: stop new anomalies for that AI provider, model, or user until the snooze ends. You can also ignore the current anomaly at the same time.
  • Adjust detection: change detection for just that provider, model, or user.
To change nOps AI detection for your whole organization, click the settings icon (Detection settings) at the top of the page:
  • Sensitivity: Low flags only large, clear changes. Medium is recommended for most teams. High flags smaller changes too, so expect more alerts.
  • Ignore changes smaller than: a dollar amount below which changes aren’t flagged.
  • Alert on: Increases and decreases, or Increases only.
  • Turn individual detection types on or off.
Changes take effect on the next daily run. Use Restore defaults to go back to the nOps defaults at any time.
Detection settings apply to nOps AI anomalies. AWS Cost Anomaly Detection is set up in your AWS console.

​
Send and schedule anomaly reports

Use Manage reports to share anomalies with your team:
  • Send now: send the current view to email addresses, Slack channels, or Microsoft Teams channels.
  • Schedule: set up a daily, weekly, or monthly digest so your team gets updates automatically.
  • Reports keep your date range, source, and high severity filters. You can set a minimum impact so only meaningful anomalies are included, and add your own subject and note.

​
FAQs

Anyone in an organization on the Inform plan, as long as their role includes Recommendations. Only admins can snooze anomalies or change detection settings.
Insights was split into two pages: Recommendations and Anomalies. Both are in the Inform section of the sidebar, and old Insights links redirect automatically.
No. It moves to the Ignored tab, and you can bring it back at any time.
Ignoring sets aside an anomaly that has already been found. Snoozing stops new nOps AI anomalies from being flagged for a provider, model, or user for 7 or 30 days.
For a known, explained change, ignore it. For nOps AI anomalies, open Detection settings to lower the sensitivity, raise the minimum change amount, or choose Increases only. To quiet one noisy model or user, use Adjust detection on its row instead.
Every day, as new billing data arrives.