> ## Documentation Index
> Fetch the complete documentation index at: https://docs.allyy.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Model evaluation dashboards

> Assess how your models perform across different business objectives, prediction types, and campaigns.

The model evaluation dashboards answer questions about how well your predictive models are actually working. They are **predictive and forward-looking**: they show you the accuracy, behavior, and business impact of the scores generated by Allyy. (For descriptive historical reporting, see [Analytics dashboards](https://docs.allyy.io/analytics/dashboards).)

> [!NOTE]
>
> Every model evaluation dashboard is automatically updated as new campaign outcomes roll in, letting you see exactly where a model hit the mark or where it needs tuning.

## Business Performance

These dashboards focus on the bottom line — helping you evaluate model quality, predict customer choices, and measure the real-world impact on your business objectives.

> [card] [Classification](https://docs.allyy.io/dashboard-evaluation/classification-model)
>
> **Propensity & Probability.** Track how accurately the model predicts specific actions like email opens, conversions, or subscription churn.

> [card] [Regression](https://docs.allyy.io/dashboard-evaluation/regression-model)
>
> **Value & Amount.** Track how closely the model estimates continuous numbers like expected donation amounts and lifetime value.

## Explainability (SHAP)

These dashboards peel back the "black box" of AI using SHAP values. They show you the exact donor data points, behaviors, and features driving the model's predictions.

> [card] [Classification SHAP](https://docs.allyy.io/dashboard-evaluation/classification-shap)
>
> **Why they take action.** See the core behavioral features driving propensity scores, segmentation, and churn risks.

> [card] [Regression SHAP](https://docs.allyy.io/dashboard-evaluation/regression-shap)
>
> **Why they give that amount.** See which variables push expected donation amounts higher or depress customer lifetime value.

***

## Available business cases

Depending on your campaign goals, you can dive into dedicated dashboards for specific channels and supporter behaviors across the sections above.

### Classification metrics

Use these to prioritize outreach, optimize campaign budgets, and understand customer risk.

* **DM Propensity** – Likelihood that a customer responds to direct mail.
* **TM Reachability** – Probability that a customer can be reached via telemarketing.
* **TM Conversion** – Likelihood that a telemarketing call results in a monthly donor signup.
* **SMS Conversion** – Probability that an SMS campaign leads to a desired action.
* **Email Open & Click** – Metrics tracking the likelihood of a recipient opening an email or clicking a link.
* **Subscription Churn** – Probability that a subscriber cancels their recurring support.
* **Donor Segments** – Models predicting **Middle donor** status, **Active donor** status, or **Subscription Acquisition**.

### Regression metrics

Use these to estimate expected revenue and long-term value to optimize your ask strings.

* **Expected Campaign Amounts** – Individual tracking for expected revenue from direct mail (**DM**), telemarketing (**TM**), and **SMS** campaigns.
* **Lifetime Value** – Estimated long-term financial value of a customer over their entire lifecycle.

> [!TIP]
>
> Evaluating a new campaign's success? Start with **Business Performance → Classification** to see your lift charts, then jump into **Explainability → Classification SHAP** to see *why* the model selected those donors.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.