Install and configure Website Growth Advisor
Open the configuration
Section titled “Open the configuration”- Open AI Actions > Marketplace.
- On Website Growth Advisor, click Install or Configure.
- Choose an analysis model or Custom configuration.
- Review the advanced settings.
- Click Save.
Choosing a model saves a starting configuration. Custom configuration lets you adjust each option yourself.
1. Choose an analysis model
Section titled “1. Choose an analysis model”| Model | Focus | Why choose it |
|---|---|---|
| Generate leads | Contact requests and inbound opportunities | To increase prospects |
| Improve conversion | Funnel, exits and steps before conversion | To reduce exits before an important action |
| Grow SEO | Pages, queries and search journeys | To work on organic visibility |
| Increase organic acquisition | Sources and content attracting unpaid visits | To grow natural traffic |
| Improve engagement | Interactions and interest signals | To understand what keeps visitors engaged |
| Increase e-commerce revenue | Products, carts and observed purchase steps | To examine an online sales journey |
Choose the model that matches the decision you need to make now. Always check the proposed primary KPI.
2. Set the objective, period and depth
Section titled “2. Set the objective, period and depth”| Setting | Available choices | Why it matters |
|---|---|---|
| Primary objective | Leads, conversion, SEO, organic acquisition, engagement or e-commerce revenue | Guides the priorities |
| Period | Last 24 hours, 7, 30 or 90 days, or custom | Short periods show recent change; longer periods show trends |
| Depth | Quick, standard or deep | Use Quick for a spot check, Standard for routine monitoring and Deep for a fuller investigation |
| Representative sessions (maximum) | 3 to 50 | More sessions enable more comparison, but make the result longer to review |
| Primary KPI | Conversions, leads, engagement rate, sessions, conversations or appointments | Gives recommendations a concrete success measure |
| Secondary objectives | The same objective families | Keeps another important outcome visible without replacing the primary objective |
3. Select sources
Section titled “3. Select sources”Enable only sources that can answer your question: Visitor Intelligence, Conversations, Knowledge, Live site crawl, Google Analytics and Search Console. A source marked as not connected cannot provide useful data. Connect it or turn it off, then save.
Configure the live crawl
Section titled “Configure the live crawl”The live crawl reads the published site without adding it to Knowledge.
| Setting | Choices | Why choose it |
|---|---|---|
| Targeted pages | Starting URLs and useful journey pages | To check a homepage, offer or contact page |
| Bounded site | A limited linked area | When several related pages must be compared |
| Maximum pages | 1 to 20 | Keeps unrelated sections out of the analysis |
| Maximum depth | 0 to 3 | 0 stays on starting URLs; higher values follow more links |
| URLs to inspect | Up to 10, one per line | Useful for a precise page-level question |
4. Set priorities and visitors
Section titled “4. Set priorities and visitors”Priority areas
Section titled “Priority areas”Select Conversion, Journey, Engagement, SEO, Acquisition and/or Content. For example, select Conversion and Journey to investigate an abandoned form.
Visitor segments
Section titled “Visitor segments”Choose All visitors, New, Returning, Organic, Paid, Direct or Referral. Start with All visitors, then compare a segment when you have a specific question.
5. Set journey analysis
Section titled “5. Set journey analysis”| Option | Why enable it |
|---|---|
| Analyse journeys | Connects visited steps |
| Conversion paths | Finds sequences associated with conversions |
| Abandonment | Finds where visitors leave before the objective |
| Compare acquisition | Compares journeys by arrival source |
| Engagement patterns | Finds recurring interaction behaviours |
6. Set evidence and recommendations
Section titled “6. Set evidence and recommendations”| Setting | Why it matters |
|---|---|
| Require evidence | Avoids claims without an observed element |
| Require cause analysis | Requests an explanation of the possible link between problem and cause |
| Allow hypotheses | Allows a clearly labelled lead when evidence is incomplete |
| Disclose uncertainty | Makes limitations visible |
| Require measurement plan | Links an action to an indicator |
| Flexible, medium or high evidence | Use High for sensitive decisions and Medium for routine monitoring |
| Diagnosis only | Understand the situation before deciding |
| Actionable | Get concrete recommendations |
| Implementation ready | Get implementation details only where they can be supported |
| Summary, practical or technical detail | Use Practical for a business team; choose Technical only when the relevant team needs it |
7. Set prioritisation
Section titled “7. Set prioritisation”The criteria Impact, Frequency, Severity, Evidence strength and Conversion proximity help rank recommendations. Keep Impact and Evidence strength for a first setup; add other criteria when the result contains many recommendations.
8. Include conversations and Knowledge
Section titled “8. Include conversations and Knowledge”- Analyse conversations connects visitor requests with website friction.
- Detect objections highlights expressed hesitation.
- Use Knowledge compares findings with information already published in ELChat.
- Write with Knowledge allows a content proposal based on that information.
Do not enable content writing with Knowledge when your content is incomplete or out of date.
9. Add visual evidence and comparison
Section titled “9. Add visual evidence and comparison”Use visual evidence can add observations from selected captures. Disabled, Targeted and Deep control the extent of the visual reading. Compare previous period helps interpret change instead of a single value. Use it only when the two periods are comparable.
10. Choose execution
Section titled “10. Choose execution”- Manual: launch each analysis when needed.
- Scheduled: choose Daily, Weekly or Monthly, then a time.
Start manually. Schedule the agent only after you have checked that the result, period, sources and KPI are stable.
Launch an analysis
Section titled “Launch an analysis”- Save the configuration.
- Choose the model period, a quick period or a custom range.
- Click Run analysis.
- Wait for Completed before interpreting the result.
The period selected just before launch applies only to that analysis; it does not replace the saved model period.
One-off period at launch
Section titled “One-off period at launch”Analysis period can override the model for one run: Today, Yesterday, Last 7 days, Last 28 days, Last 30 days, Last 90 days, This month, Last month or Custom range. Use a short period for a recent event and a custom range for a specific campaign.