ZenotiHelp

Check confidence scores and tiers

ZTZenoti Docs Team Updated October 12, 2026 6 min readGuide

Every forecast in AI Inventory Manager carries a confidence score β€” a continuous value between 0 and 1 that reflects how stable and reliable the AI's prediction is for a given product. The system groups this score into four named tiers (T1 through T4) and displays the tier as a badge on the action grid and inside the explainability panel, so you can immediately see how much weight to place on a suggested order quantity before you act on it.

Audience: Center managers and inventory owners who review AI Inventory recommendations and decide whether to approve, adjust, or defer suggested purchase orders.

Before you begin

Permission: Add or Edit (on the Inventory security object) β€” required to view confidence tiers in the action grid and explainability panel.

  • AI Inventory Manager must be activated for your organization and center. If you do not see the AI Inventory module in your navigation, contact your administrator.
  • You need Add or Edit access on the Inventory security object to interact with the action grid and explainability panel.
  • Confidence tiers are computed nightly. The tier badge you see reflects the most recent overnight refresh β€” intra-day changes to stock or settings do not update the badge until the next nightly run.

Availability: AI Inventory Manager is in Beta. Access requires CSM enablement for your organization.

Steps

Locate the confidence tier badge and its score fields on the action grid

  1. Navigate to AI Inventory in the main navigation.

  2. Select Inventory overview to open the dashboard.

  3. Ensure you are in center view β€” select your center from the center selector if needed.

  4. Select the Action grid tab (or the working grid for your center).

  5. Find the Confidence column in the grid. Each row displays a tier badge for that product:

    Badge Score range What it means
    T1 β€” High β‰₯ 0.75 The forecast is stable and reliable. The suggested quantity is a strong starting point.
    T2 β€” Medium 0.50 – 0.74 The forecast is reasonably reliable. Minor variance is possible; review the suggested quantity before approving.
    T3 β€” Low 0.25 – 0.49 The forecast has notable uncertainty. Treat the suggested quantity as a rough guide and apply your own judgment.
    T4 β€” Very Low < 0.25, or no score available The forecast is unreliable or unavailable. Do not rely on the suggested quantity without manual verification.
  6. To filter the grid to a specific tier, open the Confidence filter and select one or more of T1 β€” High, T2 β€” Medium, T3 β€” Low, or T4 β€” Very Low.

Read the confidence tier in the explainability panel

  1. On the action grid, locate the product row you want to investigate.
  2. Select the row or the Why this? link to open the explainability panel.
  3. In the panel, navigate to the Confidence part. The tier badge appears here alongside a plain-language description of what the score means for this product.
  4. Review the other panel parts β€” Summary, Inputs, and Calculations β€” to understand the full picture before deciding whether to approve or adjust the suggested order quantity.

Note: The Calculations part (reorder level, days of stock cover, target stock, and suggested order quantity) is deterministic arithmetic applied on top of the AI forecast β€” it is not itself AI-computed. Only the predicted daily usage and the confidence score come from the machine-learning model.

Decide how to act based on the confidence tier

Use the tier as a signal for how closely to scrutinize a recommendation before approving it:

  1. T1 β€” High: The AI has a stable demand signal for this product. You can approve the suggested quantity with confidence, though a quick sanity check against recent sales is always worthwhile.
  2. T2 β€” Medium: Review the suggested quantity against your own knowledge of upcoming demand (promotions, seasonal events). Minor adjustments are common at this tier.
  3. T3 β€” Low: Treat the suggestion as a starting point only. Check the Inputs part of the explainability panel to understand what data the model had available, and override the quantity if your operational knowledge differs.
  4. T4 β€” Very Low: The model had insufficient or highly variable data for this product. Manually set the order quantity based on your own records. Consider whether the product has enough purchase history to generate a reliable forecast in future cycles.

Tip: Products that consistently show T3 or T4 may benefit from a manual reorder level override. See Calculate reorder levels and suggested quantities for how to set a manual alert level for a SKU.

Worked example: Your center stocks a professional conditioning treatment (consumable). The action grid shows a T2 β€” Medium badge with a predicted daily usage of 3.2 units and a suggested order quantity of 48 units. You know a promotional event next week will increase demand. You open the explainability panel, confirm the inputs look reasonable, and manually increase the quantity to 60 before adding the line to your purchase order. The T2 badge told you the forecast was a solid baseline β€” but not so certain that you should skip the review.

Tips

  • Filter by tier before your daily review. Start with T1 β€” High and T2 β€” Medium rows to approve straightforward recommendations quickly, then give extra attention to T3 β€” Low and T4 β€” Very Low rows that need manual judgment.
  • T4 does not mean the product is unimportant. A new product, a recently added SKU, or a slow-moving item may show T4 because the model has limited purchase history to work from. The stock status (Critical, Low, or Normal) is a separate signal β€” a T4 product can still be in a Critical stock state and require urgent action.
  • Confidence tiers are product-specific, not category-wide. Two products in the same category can have different tiers if their demand patterns differ. Do not assume that a high-confidence forecast for one product applies to similar products.
  • The tier updates nightly. If you recently started stocking a product or received a large unexpected delivery, the tier may shift at the next overnight refresh as the model incorporates the new data.
  • Use the Confidence filter alongside the Stock status filter. Combining Critical stock status with T3 β€” Low or T4 β€” Very Low confidence highlights the products that need the most urgent manual attention β€” high risk, low forecast reliability.

Troubleshooting

  • The Confidence column shows T4 for a product that has been stocked for years. T4 indicates a low or unavailable confidence score, which reflects high variance or instability in the demand signal β€” not necessarily a short history. Check whether the product has had irregular order patterns, large one-off purchases, or frequent manual adjustments that could introduce variance. Open the explainability panel and review the Inputs part for clues. If the data looks correct, the tier may improve over subsequent nightly cycles as the model stabilizes.
  • The confidence tier badge is missing entirely for some rows. A missing badge is treated as T4 β€” Very Low. This can occur when the nightly forecast run did not produce a score for that product (for example, if the product was added to the system very recently and has no purchase history). The suggested quantity may still appear, but treat it with the same caution as a T4 badge.
  • The tier badge on the action grid does not match what I see in the explainability panel. Both surfaces read from the same nightly forecast. If they appear to differ, try refreshing the page. If the discrepancy persists after a refresh, note the product name and center, and contact your administrator to raise a support request.
  • I filtered by T1 β€” High but the grid shows no results. If your center has few products with stable demand signals β€” for example, a recently onboarded center or one with a small, irregular product mix β€” it is possible that no products currently score at T1. Broaden the filter to include T2 β€” Medium to find the most reliable forecasts available for your center.
  • A product I know sells consistently is showing T3 or T4. Confidence reflects the stability and reliability of the AI's forecast for that product, not sales volume alone. A product that sells consistently but in highly variable quantities (for example, bulk orders every few months rather than steady daily sales) can produce a lower confidence score. Review the demand pattern in the Demand analytics view and consider whether a manual reorder level override is appropriate.
Was this article helpful?
Your feedback helps us improve our documentation.