Skip to content

Recommendations

Optimization Recommendations is your centralized catalog of savings opportunities across compute, storage, database, networking, and AI workloads — backed by usage evidence, risk ratings, and step-by-step implementation guidance.

When to use it

  • You want to find and prioritize cost-saving actions across your cloud accounts.
  • You need to review a recommendation before implementing it in production.
  • You want to dismiss, track, or route a recommendation to the right team.

Open Recommendations

Go to Optimization > Recommendations from product navigation.

Review and act on recommendations

1. Apply filters

Filter the list by: Provider, Account, Region, Service, Environment, Recommendation Type, Confidence, Risk, or State (open, dismissed, approved).

2. Understand the summary

The top of the page shows total recommendations, potential monthly savings, and a breakdown by category.

3. Select a recommendation

Each recommendation detail shows:

Field What it means
Resource The specific cloud resource (name, ID, account, region)
Current configuration What the resource is today (instance type, size, tier)
Target configuration The recommended change
Monthly savings Projected cost reduction
Confidence How much usage evidence supports the recommendation
Risk Operational impact if the recommendation is applied
Evidence Usage trend graph (CPU, memory, I/O, etc.)
Implementation guidance Steps to apply the change

4. Take action

Depending on available actions and your permissions:

  • Approve — mark the recommendation as accepted and route to implementation
  • Dismiss — remove from the active list (record the reason)
  • Route — send to a Jira ticket, ServiceNow request, or workflow

Recommendation types

Recommendations can cover: compute rightsizing, idle resource cleanup, storage tiering or deletion, database cleanup, log retention reduction, network cost optimization, resource scheduling, commitment purchases, and service-specific opportunities.

Best practices

  • Prioritize by savings × confidence × risk — not savings alone. A high-savings recommendation with Low confidence or High risk needs more investigation.
  • Check environment before acting — confirm production status and resource owner before making a change.
  • Compare current and target config — don't rely solely on the savings estimate.
  • Document dismissals — record why a recommendation was dismissed so you can revisit it later.
  • Review after infrastructure changes — new deployments and billing corrections generate new recommendations.

Troubleshooting

If no recommendations appear, see Missing recommendations and anomalies.