Skip to main content

Confidence Scoring

Every Execution in frugally.app receives a confidence score — a value from 0 to 100 that indicates how reliably the Execution is expected to achieve its goal. Higher scores mean greater confidence that the right resources will be acted on successfully.


What the confidence score represents

The confidence score is a composite metric that considers multiple factors about the Target, its resources, and historical performance. It answers the question: "How confident is frugally.app that this Execution will do what you expect?"

Score rangeLabelMeaning
80–100HighStrong resource match, good historical success, well-configured Target
50–79MediumLikely to succeed but some factors could be improved
0–49LowSignificant risk of unexpected results — review before scheduling

Scoring factors

The confidence score is calculated from the following factors:

Resource match quality

How well the Target's filters match the intended resources.

SignalPositive impactNegative impact
Tag coverageAll matched resources have the expected tagsResources matched by broad filters (e.g. no tags, region-only)
Resource count stabilityConsistent count across recent scansCount fluctuates significantly between scans
Service type matchResources are the expected type and sizeMixed resource types that may behave differently

Historical success rate

How often previous Executions against this Target have succeeded.

SignalPositive impactNegative impact
Past success rate90%+ Executions succeeded fullyFrequent partial failures or full failures
ConsistencyResults are predictable across runsInconsistent outcomes (some runs succeed, others fail)
RecencyRecent Executions succeededLast few Executions had issues

Configuration completeness

How well the Target and its associated resources are configured.

SignalPositive impactNegative impact
IAM permissionsAll required permissions are in placeMissing or outdated IAM policies
Connection healthAll linked Connections are healthyOne or more Connections have issues
Tag specificityFilters use specific key:value pairsFilters are overly broad (key-only or no tags)

Where confidence scores are displayed

LocationDetail shown
Execution detail pageFull score with per-factor breakdown
Executions listScore badge (High / Medium / Low)
Onboarding WizardPer-recommendation score at the review step
Slack notificationsScore included in Execution result messages
Dry run resultsScore for the simulated Execution

[SCREENSHOT: confidence-score-breakdown.png -- Execution detail showing confidence score with factor breakdown]


How to improve a low score

FactorAction
Low tag coverageAdd specific tags to your AWS resources and update the Target's tag filters
Fluctuating resource countInvestigate why resources are being created/destroyed — consider more specific filters
Low historical success rateReview failed Executions for error patterns — often an IAM or connectivity issue
Broad filtersNarrow your Target's filters with additional tag key:value pairs
Unhealthy ConnectionCheck the Connection status and update the IAM role if needed
tip

A confidence score below 50 does not prevent an Execution from running. It is an advisory signal to help you prioritise which Targets to review and improve. Running a dry run is a safe way to investigate before taking action.


Confidence scores and the Onboarding Wizard

The Onboarding Wizard uses confidence scores to rank its Target recommendations. The Select All High Confidence button selects only recommendations with a score of 80 or above, ensuring your initial setup starts with the most reliable Targets.