Inactivity Detection (Proactive Assistance)

Unlike a standard support chatbot that waits passively to be contacted, EcomBot embeds an Inactivity Detection and Silent Evaluation engine. This feature turns it into a truly proactive sales agent, capable of saving a sale before the visitor leaves the site.

The Principle of Silent Evaluation

When a visitor is identified as inactive (or about to leave the page), EcomBot does not blindly trigger by sending a generic "Can I help you?". It first performs a Silent Evaluation in the background.

1. Navigation Context Analysis

Upon triggering, the AI analyzes the exact URL where the customer is located.

• If the customer is on a Product Page, the AI loads the attributes, price, stock status, and product description.
• If the customer is on a Category Page, the AI analyzes the type of products displayed.

2. Decision Making (Anti-Spam Filter)

Before displaying any notification bubble, the AI asks itself a strict question: "Do I have any added value to offer at this moment?"

• If YES: The AI identifies a potential obstacle (e.g., shipping, warranty, technical complexity of the product) and formulates a tailor-made hook for this specific product.
• If NO: (For example, a generic product with no complexity, and no warranty information in the knowledge base), the AI decides to do nothing so as not to spam the visitor (NO_HELP_NEEDED).

Client-Side Display

If the AI decides to intervene, a contextual hook bubble (Proactive Bubble) is displayed above the closed widget.

The "Good Seller" vs "Robot Seller" Approach: EcomBot is trained to never act like a "Robot Seller" (who would coldly recite the technical data sheet or remain silent if the product is too basic). Instead, it acts like a Good Seller: it translates the most technical feature into a concrete customer benefit. For example, instead of saying "It's a 1800W product", it will say "Its 1800W power will save you precious time every day!".

• It adopts a conversion-oriented posture by offering to answer specific obstacles (e.g., "A question about the maintenance of this set or our delivery times?").
• It often includes clickable quick reply buttons to reduce friction.

Micro-Urgency and Auto-Close: To preserve the user experience and not hide your content, the proactive bubble has a discreet 15-second progress bar. If the visitor does not interact with the bubble within this time, it closes automatically (Zero Friction).

If the viewed product is part of your "Best Sellers" (high total_sales), the AI will even use this criterion as a social proof argument ("It's a safe bet, highly appreciated by our customers!").

How to configure Proactive Assistance?

1. Go to your SuperAdmin Dashboard.
2. Go to the General section (Widget Settings).
3. In the Style tab, scroll down to the "Anti-Abandonment Proactive Assistance" section.
4. You can use the On/Off toggle to completely enable or disable this feature.

💡 Tip: Test the rendering on your store

• From the dashboard: You can test the hooks generated by the AI without leaving the administration interface. Go to the Products tab. Next to each product, click on the bubble icon with stars ("Simulate AI") to see exactly how the agent would approach an inactive customer on this specific product page.
• On your live site (Live URL): If you want to see the proactive bubble appear instantly on the product page of your site under real conditions (without waiting for the inactivity delay), simply add the parameter ?test_proactive=1 to the end of your page URL (e.g., https://mystore.com/product/t-shirt?test_proactive=1).