Skip to main content

Configuring Echo: Customise Signals, Plaibook Actions, and Assignees

Customise Echo to your CS team's workflows and risk and upsell identification process

Echo is Hook’s AI-powered sentiment intelligence layer. Echo ingests data across your emails, meetings and support tickets and leverages an LLM to surface key insights to your whole business.

This article explains how to customise Echo signal and Plaibook actions: editing the prompts that drive upsell and churn signal detection, adjusting Plaibook actions, and setting assignees for both actions and signals.

👤 Managers and Technical Admins have access to these settings in Hook

Prerequisites

  • Manager or Technical Admin access to Hook

  • Echo must already be enabled for your organisation, including the automation that is required to activate the AI agent. This is set up during the Hook implementation process.

Navigate to Echo configuration

  1. Go to Organisation settings -> Agent configuration in Hook.

  2. You will see configuration options for signal detection, including separate sections for churn signals, upsell signals and plaibook actions.

Edit Signal Detection Settings

In Agent configuration settings, the following can be updated:

Example of Agent configuration settings for Signal Detection

Handling Pre-Existing Risks

Setting

What it does

Auto Dismiss Older Draft Risks

When enabled, Echo will automatically dismiss old detected draft risks and replace them with newer ones upon re-evaluation. Keeps your risk feed current without manual cleanup.

Recommended to enable.

Support Tickets: Signals Involving Technical / Product Issues or Limitations

When Echo detects a customer is concerned about technical issues or product limitations (from support tickets), this is treated differently to other signals as it's a frequently occurring event and will be ignored unless material.

Setting

What it does

Signal Flag for Technical / Product Issues or Limitations

Select the flag that relates to product/tech issues, if one exists in your configuration. This tells Echo which flag to apply when a technical complaint is deemed material enough to surface.

Example: Tech / Product Issues

Signal Reason for Technical / Product Issues or Limitations

Select the reason category that relates to product/tech issues. Note: a reason must be selected if no flag is chosen.

Example: Functionality gaps

Strict Mode for Support Tickets

When enabled, Echo applies stricter analysis criteria when processing support tickets to reduce false positives. Signals relating to technical issues will be discarded unless of very high severity or expressly causing the customer to leave.

Recommended to enable.

Upsell Signals

Setting

What it does

Detect Upsell Signals and Opportunities

When enabled, Echo will detect upsell opportunities and positive signals in addition to risks. Disable this if you only want Echo to focus on risk detection.

Recommended to enable.

Signal Flags for Upsell Opportunities

Select the flags that should be applied when Echo identifies a positive signal or upsell opportunity

Example: Growth opportunity

Signal Reasons for Upsell Opportunities

Select the reason categories that will be applied when Echo detects upsell opportunities. These help categorise the type of positive signal detected Example: ROI/value

Signal Owner

Customise which role associated to an account will be assigned as an owner of the signal.

Setting

What it does

Owner

Defines who should own the signals detected by Echo. Can be set to a specific role (e.g. Account Executive, CSM) or anyone on the account.

Example: CSM

Customise Echo Signals Output

Echo uses a prompt to add extra context for what it looks for when analysing customer interactions. This is additional information to Hook's default system prompt which is configured to act as a top performing CSM.

Editing the Echo prompt changes what Echo treats as evidence of a signal.

  1. In the Extra Rules, Instructions or Business Context for Signal Detection textbox, update the prompt text to reflect the language, criteria, or context that matches your business. For example, you might specify particular product lines, contract terms, or customer behaviours that indicate risk or expansion opportunity for your accounts. Use the Hook Echo Questionnaire (attached at the bottom of this article) to guide writing this prompt.

  2. Save your changes.

Guidance on Editing Echo Prompts

Follow the best practice guidelines below to ensure the updated Echo prompt generates high quality Signals.

  • Write organisation-specific signal logic. Don't repeat what Echo already knows. Echo's system prompt already covers general churn risk detection, severity scoring, deduplication, and ignoring BAU meeting cancellations or discount posturing. Your custom prompt should only add context Echo can't infer on its own: your products, your customer segments, and the specific behaviours that mean something in your business.

  • Reference your products, tiers, and how customers expand. Echo needs to know how customers land and grow - which product they start on, what expansion typically looks like, and what usage patterns precede an upsell conversation. Without this, it can't distinguish a healthy power user from someone about to hit a ceiling.

  • Use XML tags to separate signal types. Structure your prompt using <upsell_signals>, <churn_signals>, and <ignore> sections. This keeps the logic readable and makes it much easier to update one section without affecting the others.

  • Only add plaibook guidance if you have documented CS workflows to draw from. Echo's action recommendations are most useful when they reflect your actual escalation paths, segment-specific plays, or QBR cadence. If you don't have that documented yet, leave this blank - invented guidance is worse than none.

Edit Plaibooks Settings

Plaibook generation prompt configuration controls how Echo generates recommended actions and escalation paths when a signal is identified. The following can be configured specifically for your organisation.

Example of plaibook settings

ARR Thresholds

Setting

What it does

Free Services ARR Threshold

The ARR threshold above which free services (e.g. training, onboarding) may be offered as a retention tactic. Leave blank to disable this recommendation entirely.

Internal Meeting Queue Jump ARR Threshold

The ARR threshold above which customers can be prioritised over others for internal meetings or engineering/technical escalations.

Discounting ARR Threshold

The ARR threshold above which discounting may be considered as a retention strategy. If left blank, the default is that discounting recommendations are permitted for any account to retain the logo.

Executive Involvement ARR Threshold

The ARR threshold above which executive involvement may be recommended as part of retention efforts.

Management Involvement ARR Threshold

The ARR threshold above which management involvement may be recommended as part of retention efforts.

Pricing & Leadership

Setting

What it does

Permit Usage-Based Pricing

When enabled, usage-based pricing options may be offered as part of retention playbooks. Disable if usage-based pricing is not applicable to your business model.

Involve Leadership for Critical Risks

When enabled, if a risk is classified as critical or major, Echo can recommend involving leadership in the decision -regardless of ARR thresholds.

Actions Assignee

Setting

What it does

Actions Assignee

The job role that generated playbook actions should be assigned to (e.g. CSM). This ensures tasks land with the right person in your team.

Customise Plaibook Actions

Each signal type can have associated playbook actions - the steps Echo recommends or triggers when a signal is detected. These can be customised to your business context by following the steps below.

  1. In Agent configuration, navigate to the Plaibook section.

  2. In the Extra Instructions for generating Plaibooks textbox, update the prompt text to reflect the language, criteria, or context that matches your CS team's workflows. For example, you might specify specify an existing playbook for when a champion leaves the business.

  3. Save your changes.

Only add playbook guidance if you have documented CS workflows to draw from. Echo's action recommendations are most useful when they reflect your actual escalation paths, segment-specific plays etc. If you don't have that documented yet, leave the prompt here minimal - invented guidance is worse than none.

Did this answer your question?