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

# Optimize

> Use unanswered questions, objections, and follow-up insights to improve your AI agent.

## Overview

The **Optimize** page helps you improve Apten with real conversation data. It shows where the AI needs more information, what leads are pushing back on, and which follow-up approaches are getting replies.

Optimize has three sections:

* **Unanswered Questions**: questions Apten could not answer confidently

* **Common Objections**: recurring concerns or barriers raised by leads

* **Follow-ups**: weekly insights into follow-up strategy and message performance

Use the **Profile** filter at the top of the page to review all profiles or focus on one customer profile.

<Note>
  Optimize is most useful when you review it regularly and make small, targeted changes. For example, add one missing FAQ, adjust one customer profile instruction, or apply one follow-up recommendation, then compare future results.
</Note>

## Unanswered Questions

Unanswered Questions shows questions Apten was not confident enough to answer during a conversation.

Similar questions are grouped together so you can fix the knowledge gap once instead of reviewing the same issue multiple times.

### What you can review

Each question group shows:

* A representative question

* The profiles where the question appeared

* How many times similar questions were asked

* Whether the group is active, addressed, or archived

Expanding a group shows the individual lead questions and a link to open the original conversation.

### How to address a question

1. Go to **Optimize > Unanswered Questions**.

2. Open an active question group.

3. Review the example questions and source conversations.

4. Click **Address**.

5. Add the missing answer, FAQ, instruction, or business detail.

6. Click **Save Changes**.

Saving changes updates your Custom Knowledge and marks the group as addressed.

You can also click **Mark addressed without changes** if the question is already handled somewhere else.

### Best practices

* Add Custom Knowledge only when the answer applies across your account.

* If the answer only applies to one customer profile, update that profile in **Build & Test** instead.

* Use **Archive** for old or irrelevant groups you no longer want in the active list.

* Use **Show archived** if you need to review archived groups later.

## Common Objections

Common Objections shows recurring reasons leads hesitate or push back during conversations.

An objection is more than a simple "no." It includes a reason that explains what is blocking the lead from moving forward.

Examples of objections:

* "It's too expensive for us right now."

* "We already have a vendor for this."

* "I need approval from my manager before moving forward."

Examples that are usually not objections:

* "No thanks."

* "I'm driving, call me later."

* "How much does it cost?"

### What you can review

Each objection group shows:

* A representative summary of the concern

* The profiles where it appeared

* How many leads raised a similar concern

* The first and latest times the concern appeared

Expanding a group shows individual examples and links to the original conversations.

### How to use objections

1. Go to **Optimize > Common Objections**.

2. Review the highest-volume objection groups first.

3. Open example conversations to understand the context.

4. Update the relevant customer profile, follow-up strategy, or Custom Knowledge with better handling for the objection.

5. Archive groups that are no longer useful.

Good objection handling usually means giving Apten clearer positioning, stronger qualification logic, or better instructions for how to respond when the concern comes up.

## Follow-ups

Follow-ups shows weekly insights into automated follow-up performance. It helps you understand which follow-up strategies, message patterns, and schedule choices are driving replies.

### What you can review

The Follow-ups report includes:

* **Recommended Actions**: suggested changes you can review, apply, or reject

* **Strategy Performance**: reply metrics for each follow-up strategy

* **Detected Patterns**: recurring message themes in follow-up content

* **Key Observations**: trends, concerns, opportunities, and anomalies

Use the **Period** selector to switch between weekly reports. Use **Sort by** to sort metrics by message volume or reply rate.

### Recommended Actions

Recommended actions can include:

* **Update Strategy**: revise the instructions for an existing follow-up strategy

* **Optimize Schedule**: change follow-up timing, channel, or strategy assignment for a step

* **Delete Strategy**: remove a redundant or underperforming strategy

* **Create Strategy**: add a new strategy for a missing follow-up approach

* **Set Up Strategies**: add follow-up strategies for profiles that do not have them yet

Click **Review** to inspect the rationale, evidence, and exact proposed change. From the review dialog, you can apply the change or reject it with a reason.

<Warning>
  Applying a recommendation changes live follow-up configuration. Review the proposed change and evidence before applying it.
</Warning>

### What applying a recommendation changes

* Applying an instruction update changes the selected follow-up strategy's instructions.

* Applying a strategy deletion removes the selected follow-up strategy.

* Applying a strategy creation adds a new follow-up strategy for the profile.

* Applying a schedule change updates the profile's follow-up configuration for the proposed step.

* Rejecting a recommendation records your decision so Apten can avoid repeating the same recommendation without stronger evidence.

## Strategy Performance

Strategy Performance compares follow-up strategies for the selected week.

Each strategy can show:

* Message count

* Reply count

* Reply rate

* Average response time

* Sample messages

* Links to source conversations

Deleted strategies can still appear in historical reports if they were active when the messages were sent.

## Detected Patterns

Detected Patterns groups follow-up messages by what the message says, not just which strategy generated it.

Each pattern can show:

* A pattern label and keywords

* Message count

* Reply rate

* Strategy breakdown

* Sample messages and reply status

Patterns are useful when strategy-level metrics are too broad. For example, one strategy might perform well overall but struggle with a specific message pattern, or a low-volume pattern might reveal a promising angle to test.

## Interpreting reply rates

Follow-up reply rates depend on where a message appears in the sequence.

A later follow-up usually reaches harder-to-convert leads because those leads already ignored earlier messages. That means a strategy used later in the sequence can have a lower reply rate and still be valuable if it uses a different angle and gets replies from people who ignored earlier follow-ups.

Good follow-up optimization usually means:

* Put the broadest-appeal strategy earlier in the sequence.

* Use meaningfully different approaches across steps.

* Keep niche or experimental approaches for later steps.

* Remove strategies that are redundant or consistently weak after enough volume.

* Create new strategies when the data shows an uncovered message angle.

## FAQ

### Why don't I see any Optimize data?

There may not be enough matching conversation or follow-up activity yet. Follow-up reports appear after the weekly report is available.

### Why did the same question or objection show up under multiple profiles?

Similar questions and objections can be grouped across profiles when the underlying issue is the same. Use the profile badges and profile filter to decide whether the fix should be shared or profile-specific.

### Should every answer go into Custom Knowledge?

No. Use Custom Knowledge for facts and instructions that apply broadly. Use profile edits in **Build & Test** when the answer only applies to one profile.

### Why does a rejected recommendation still appear?

Rejected recommendations remain visible in the weekly report as history. Future recommendations take that decision into account.

### Can I review the original conversation?

Yes. Question examples, objection examples, strategy samples, and pattern samples link back to the source conversation when available.
