# Analytics

> Message volume, delivery, team workload, campaign and AI results — and what each number does not tell you.
> Source: https://onpad.in/docs/analytics

Analytics shows what your workspace did over a period, with the previous period beside it so you can see direction rather than just a number.

## Choosing a period

Pick the number of days. Everything on the page uses it, and the comparison is against the equally long period before it — so 30 days is compared with the 30 before that, not with last calendar month.

## Messages

| Metric | What it counts |
|---|---|
| **Received** | Inbound messages from customers |
| **Sent** | Outbound messages from you |
| **Accepted** | Messages Meta accepted for delivery |
| **Delivered** | Messages that reached a phone |
| **Read** | Messages the recipient opened |
| **Failed** | Messages that did not go |
| **Pending** | Sent, no final result yet |

Shown as a daily chart and as totals.

### Reading these honestly

**Read is unreliable.** Customers with read receipts off never report read, however carefully they read. A low read rate may mean nothing at all.

**Delivered is the real outbound number.** It is the one that says the message arrived.

**Failed is the one to act on.** A failure rate that climbs is either bad data or a problem with the number — both get worse if ignored.

**Received is your demand.** If it rises while your replies do not, customers are waiting.

## Team workload

How the conversations were shared out. Useful for noticing that one person is carrying the queue, or that assignment is not happening at all.

> Workload shows **how much** each person handled, not how well or how fast. It does not measure response time, and nothing here should be read as "fastest responder".

## Campaigns

Per-campaign results over the period: audience, sent, delivered, read, replied, failed.

**Replied is the number that matters.** Delivered says it arrived; replied says it worked. A campaign with 98% delivery and 0.2% replies did not work, however good the delivery looks.

## AI and automation

How many replies the AI Agent sent and how many automation runs happened, with successes.

Watch the gap between runs and successes. Runs climbing while successes do not means a flow is breaking — usually a paused template or an Assign node pointing at someone who left.

## Hot leads

Where ONPAD marks contacts as hot leads, it is a **heuristic** — a guess from message frequency and recency, not a verified intent signal. Useful for ordering a call list, not for forecasting revenue.

## What is not here

**Revenue.** ONPAD does not know what a conversation was worth. Tag contacts that convert and compare tags over time if you need this.

**Per-message cost.** Meta bills you directly; its charges are in Meta Business Manager, not here. See [What it costs](https://onpad.in/docs/what-it-costs).

**Response time.** Not currently measured.

## Using it weekly

Four questions worth asking every week:

1. **Is failed climbing?** If yes, look at the reasons before the next campaign.
2. **Is received rising faster than sent?** If yes, customers are waiting.
3. **Which campaign got replies?** Do that again. Delivery is not the goal.
4. **Is one person carrying the inbox?** Fix it before they stop enjoying their job.

## Next

- [Campaigns](https://onpad.in/docs/campaigns) — the per-campaign detail behind these totals
- [Inbox](https://onpad.in/docs/inbox) — where workload comes from