# Understanding Your Score

## How the Score Is Calculated

The Efficiency Score is the average of four individual ratings, each derived from a specific efficiency indicator. Every indicator is rated independently as **Good**, **Needs Improvement**, or **Poor** based on the thresholds below.

| Rating            | Numeric Value |
| ----------------- | ------------- |
| Good              | 100           |
| Needs Improvement | 50            |
| Poor              | 0             |

The overall Efficiency Score is the **average** of the four numeric values, giving you a single percentage between 0% and 100%.

***

## Rating Thresholds

Each indicator uses a different scale depending on whether a lower or higher measured value is better.

### Warehouse Idle Time

Measures the percentage of compute time spent idle (no queries running) across all warehouses over the last **21 days**.

| Rating            | Idle Time % |
| ----------------- | ----------- |
| Good              | < 3%        |
| Needs Improvement | 3% – 8%     |
| Poor              | > 8%        |

### Multi-Cluster Idle Time

Measures idle time on non-primary clusters over the last **7 days**. Only applicable to Enterprise edition and above.

| Rating            | Idle Time % |
| ----------------- | ----------- |
| Good              | < 3%        |
| Needs Improvement | 3% – 10%    |
| Poor              | > 10%       |

### Warehouse Sizing Efficiency

Evaluates whether your warehouses are right-sized for their workload based on query history from the last **7 days**. Higher is better.

| Rating            | Sizing Efficiency |
| ----------------- | ----------------- |
| Good              | ≥ 80%             |
| Needs Improvement | 60% – 80%         |
| Poor              | < 60%             |

### Clustering Column Efficiency

Measures how effectively clustering keys are configured and how well partition pruning performs over the last **7 days**. Higher is better.

| Rating            | Clustering Efficiency |
| ----------------- | --------------------- |
| Good              | > 90%                 |
| Needs Improvement | 70% – 90%             |
| Poor              | < 70%                 |

***

## Estimated Annual Savings

The savings estimate is calculated per indicator and then summed to produce a total. For each indicator:

1. The app measures the inefficiency over the lookback window (7–21 days depending on the indicator).
2. Costs are attributed using your **cost-per-credit** rate (auto-detected from `RATE_SHEET_DAILY` when available, or entered manually).
3. A **weighted average** between a lower and upper savings bound is computed (65% weight on the lower bound, 35% on the upper).
4. The result is **annualized** by extrapolating from the lookback window to 365 days.

The savings breakdown shows how much each optimization area contributes:

* **Auto-Shutdown savings** — eliminating idle warehouse compute
* **Auto-Scaler savings** — reducing unnecessary multi-cluster overhead
* **Smart Pulse savings** — right-sizing warehouses for their workloads
* **Clustering savings** — improving or removing ineffective clustering configurations

***

## Organization-Level Projection

If your Snowflake account is part of an organization, the app queries `ORGANIZATION_USAGE.USAGE_IN_CURRENCY_DAILY` to determine costs across all accounts. It then **projects your account-level savings ratio** across the organization to estimate the total optimization opportunity.

> This projection is an estimate based on extrapolating the current account's efficiency patterns. Actual savings across accounts will vary.