;
**Part 6: Measuring and improving your agent · Lesson 2 of 4 · 15 minutes**
## Objective
Read connectivity and goal rate for Ananya's calls, and jump from a figure to the calls behind it.
## The scenario
With three calls the numbers are small, but the reading habit is the same at three thousand: did the calls connect, did Ananya get a promise to pay, and were the retries worth their cost? Analytics answers across all calls.
## Hands-on
On **Monitor → Campaigns**, click the dry run and then **Analytics**. The page opens scoped to the campaign and the date range drops away: a campaign covers all of its calls whenever they happened. For Ananya across every campaign, open **Monitor → Analytics** instead and tick her under **Filters → Assistants**, which selects all her versions.
Eight cards. With a date range, a card can carry a change badge against the previous period of the same length; scoped to a campaign, as here, there is no previous period and no badge. **Total calls**, **Calls connected** and **Connectivity rate** matter most here. Hover **Details** on **Connectivity rate** for **Campaign time exhausted** and **Retry exhausted**: calls stopped by the campaign's end time or its retry count. **Avg handle time** splits by duration on hover; many calls under 10 seconds is the cue to rework the opening.
**Goal achieved** is the share of analysed calls where `payment_promised` came back true, because you marked it **Use as goal** in part 3. **Cost per goal** divides the cost of connected calls by those calls. Both appear only with an agent or a campaign selected, and read `-` if the agent has no goal. **Latency** is Ananya's average response time in milliseconds.
The **Connectivity** chart plots cumulative connectivity per attempt: **Attempt 1**, **Attempt 2** and so on. Hover a point for that attempt's attempted and connected counts. A flat line after attempt 2 means later retries reach no one, and **Retry count** should come down.
Each half-donut under **Cohort analysis** splits connected calls by one Boolean or enum field: `payment_promised` and `objection`. Hover a slice for its connected calls and average handle time. `promised_date` is not charted: only Boolean and enum fields are.
Click the **Calls connected** value. Logs opens with **Call status** **Completed** and the agent, campaign and date filters carried over. A slice of the `objection` donut does the same for one value, with exactly one agent selected.
## Checkpoint
Logs opens filtered to status **Completed** and to the agent or campaign you had selected, listing the connected calls.
## Exercise
Scope the page to one version of Ananya. Under **Filters → Assistants**, untick V2 (from part 5) so only V1 is selected. Expected result: every card recomputes for V1's calls alone; once V2 has taken calls, swapping the ticks shows its figures. **Goal achieved** and **Avg handle time** are the two to compare between versions.
## What you learned
* Analytics is for trends and Logs for one call. Selecting a campaign drops the date range; otherwise the page covers the last month in your browser's timezone.
* **Goal achieved** needs a Boolean Advanced Analysis variable marked **Use as goal** and an agent or campaign selected.
* The connectivity-by-attempt chart tells you whether retries pay off, and clickable figures open Logs with the same filters.
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.