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

# Metrics and engagement evidence

> Read counts, rates, time windows, advisor output, and tracking caveats correctly

Signal metrics summarize project outcomes over a selected window.
Use them to find where behavior changed, then inspect the affected messages, domains, contacts, topics, and suppressions.

## Project windows

The dashboard offers 24 hours, 7 days, 30 days, and 90 days.
The 24-hour view uses hourly buckets; longer views can use daily buckets.
Compare equal windows and granularity before calling a difference a trend.

## Counts and rates

Project metrics include sent, delivered, opened, clicked, bounced, complained, failed, and unsubscribed outcomes.
Provider and tracking events arrive asynchronously, so a recent bucket can continue changing.

| Rate               | Contract boundary                                                     |
| ------------------ | --------------------------------------------------------------------- |
| Delivery rate      | Delivered relative to the relevant sent volume                        |
| Open rate          | Tracked opens relative to delivered volume                            |
| Click rate         | Tracked clicks relative to delivered volume                           |
| Click-to-open rate | Tracked clicks relative to tracked opens                              |
| Bounce rate        | Bounces relative to sent volume                                       |
| Complaint rate     | Complaints relative to the denominator defined by the metric response |

The API also exposes topic performance, individual-email performance, read engagement, a project summary, a timeline, and the deliverability advisor.

## Tracking caveats

An open is a tracking-image load.
A click is a tracked-link request.
Image blocking, privacy proxies, security gateways, and link scanners can change those counts without a matching human action.
Use application conversion as a separate measurement when human intent matters.

## Deliverability advisor

The current dashboard bands are **Healthy** at 90 or above, **Good** at 75 or above, **Needs attention** at 50 or above, and **Critical** below 50.
The score summarizes underlying delivery evidence; it is not an independent guarantee of inbox placement.

## Investigation sequence

1. Confirm whether total volume changed.
2. Separate delivery, bounce, complaint, rejection, failure, unsubscribe, and suppression behavior.
3. Narrow the time window to the first changed bucket.
4. Open representative message timelines.
5. Check domain and tracking configuration.
6. Check the audience source and recipient-safety state.
7. Compare the next equivalent window after a correction.

<Warning>
  A percentage without its count, denominator, and time window is incomplete evidence.
</Warning>