> ## Documentation Index
> Fetch the complete documentation index at: https://help.statisfy.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Health Settings
> Configure the default grace period for manual health overrides, require a reason on every override, and choose which custom objects feed into AI Health.
The admin **Health** settings page controls the **health scale** your team sees, the **grace period** that holds a manual override before AI takes back over, the **reason** users must provide on every override, the **custom objects** the AI considers when it calculates a customer's health, and the **health instructions** that guide how the AI scores accounts.
### Accessing Health Settings
1. Navigate to **Admin → Settings → Health**. You need admin permissions to access this page.
2. The Health section exposes four related controls:
* **Custom Health Scale** — Choose the labels and colors your team uses for account health
* **Default Grace Period (days)** — How long a manual override is honored before the AI recalculates health
* **Override Reason Required** — Always on; users cannot save an override without a reason
* **Additional Custom Objects for Health** — Pick custom objects whose records should flow into AI Health alongside accounts, activities, and opportunities
* **Health Instructions** — Edit the written guidance the AI follows when scoring health, and preview the effect of your edits on a real account before saving
### Custom Health Scale
Out of the box, Statisfy shows account health on its standard scale. If your team thinks in different terms — a simple red/yellow/green, or a five-step scale — you can rename the whole scale to match:
1. Turn on the custom health scale toggle.
2. Start from a preset — **3-point** (Positive / Neutral / Negative) or **5-point** (Excellent / Good / Fair / Poor / Critical) — or build your own.
3. For each step, set the **label** your team should see and pick a **color**. You need at least two steps, and you can add or remove steps freely.
4. Map each of Statisfy's underlying health values onto one of your steps, so AI-computed health always lands on the right label.
5. Save.
Your labels and colors then appear everywhere health is shown — account pages, list views, filters, and reports.
Switching scales doesn't rewrite history by itself. After changing the scale, use **Run backfill** to restate existing health records on the new scale; the page shows the backfill's progress and when it finishes.
### Default Grace Period
The grace period is the number of days a manual health override is locked in before AI-calculated health resumes. Statisfy ships with a default of **30 days**.
**Available options:**
| Days | When to use |
| ------ | ----------------------------------------------------------------------------------- |
| **7** | Fast-moving accounts where you want the AI to revisit quickly |
| **14** | Two-week feedback cycles |
| **30** | Balanced default — enough room to act on the override, short enough to stay current |
| **60** | Slower-moving renewal cycles |
| **90** | Quarterly review cadences |
Pick the value that matches your team's review cadence. Once configured, this becomes the **default** that pre-fills the grace-period selector on every override modal — individual users can still pick a different grace period per override.
Changing the default does not retroactively change the grace period on overrides that have already been saved. The new default applies to overrides created from that point forward.
### Mandatory Override Reason
Every manual health override now requires a non-empty reason. Whitespace-only reasons are rejected. This captures the rationale behind each override and feeds into Statisfy's [Health Memory](/admin/customer_insights_settings) (the **Memory** tab in Settings) — the AI learns from your team's collective context over time.
**What makes a good override reason:**
* Be specific about the signal you're acting on (e.g., "Executive sponsor left; new buyer not yet onboarded")
* Reference the timeframe (e.g., "Through end of Q2 — renewal review")
* Avoid generic placeholders ("test", "checking", "n/a")
Override reasons surface on the Health Memory page (the **Memory** tab in Settings) over time. Concise, specific reasons make the AI's calibration to your business more accurate.
### How Overrides Work
1. From an account detail page, click the **Health** badge and choose **Override Health**.
2. Pick the new health value and an optional grace period (defaults to your tenant configuration).
3. Enter an **Override Reason** — required, non-empty.
4. Save. The override holds for the grace period, after which the AI-calculated health resumes.
### Additional Custom Objects for Health
By default the AI considers accounts, activities, and opportunities when it scores health. If you've modeled additional business state as **[Custom Objects](/admin/object_manager)** — onboarding milestones, support tier records, escalation logs, expansion plays — you can tell AI Health to read those too.
1. Scroll to **Additional Custom Objects for Health** on the Health settings page.
2. Click **Add Custom Object** and pick a custom object type from the dropdown. Only custom objects with `account` as their parent entity appear in the list, because each record needs to attach to a specific account for health to use it.
3. Repeat for every custom object you want included. Each one shows up as a removable chip; click the **×** on a chip to drop it.
The next health calculation reads each linked custom-object record for the account and surfaces it to the AI alongside standard signals. Records appear in the **Health rationale** Stella generates for the account, so reviewers can see exactly which custom-object rows contributed.
Choose objects that genuinely move the health needle — onboarding-stage rows, support escalations, executive-sponsor changes. Adding low-signal objects (e.g., logging tables, audit trails) makes the AI noisier without improving accuracy.
You can change the list at any time. New selections take effect on the next scheduled health recompute for each account, or immediately if you trigger a manual recalculation.
### Health Instructions
**Health Instructions** are the written guidance that tells Statisfy's AI how to weigh signals when it scores an account's health. You can edit any instruction inline and, before committing, **Preview** how your changes would land on a real account.
1. Find the **Health Instructions** section on the Health settings page. Each instruction has its own text box and **Save** button — edit and save them individually, the same way you would any other setting.
2. To check the effect of unsaved edits, click **Preview** at the top of the section. Preview becomes available once at least one instruction differs from what's currently saved.
3. In the **Preview health instructions** dialog, pick an **Account** and click **Compute health**. Statisfy scores that account's health two ways — once with the **saved** instructions and once with your **current (unsaved)** edits — and shows them side by side so you can compare before committing.
4. Previewing never saves your changes. Close the dialog and use the inline **Save** buttons to keep any edits you want.
Preview against a few representative accounts — a healthy one and an at-risk one — to confirm an instruction change moves health the way you expect before you save it.
### Need Help?
Contact [support@statisfy.com](mailto:support@statisfy.com) if you need a grace-period option outside the standard 7 / 14 / 30 / 60 / 90 day choices, or if you have questions about how override data feeds into Health Memory.