What is your AI actually doing?

You installed an AI sales assistant and conversations are happening. But you log into the dashboard and the only thing you see is a credit counter. You don't know if customers are clicking on recommended products. You don't know if they leave conversations satisfied. You don't know which store is burning through 80% of your credits.

It's a bit like running a store with a sales associate you never observe — they're working, but you have no idea how.

We're launching the Reports module — five tabs that show you what your AI is doing and whether it's doing it well.

Five Tabs, One Complete Picture

1. Overview

A bird's-eye view of AI activity. Eight key metrics:

Conversations - total chats conducted in the selected period
Widget impressions - how many times the widget appeared on pages
Product clicks - total clicks on AI recommendations
Add to cart - how many times a customer added a recommended product to cart
Average messages per chat - how many messages are exchanged per conversation
Handoff rate - what percentage of conversations required human intervention
Positive ratings - percentage of conversations rated positively by customers
Credits used - how many AI credits were consumed in the period

Plus a daily conversations chart showing the activity trend over time.

2. Conversion Funnel

The key question: is the AI guiding customers toward purchase?

The funnel visualizes five stages of the customer journey:

Widget impressions → how many times the widget was displayed
Unique sessions → how many unique visitors saw the widget
Conversations started → how many of them initiated a chat
Product clicks → how many clicked on a recommended product
Add to cart → how many added a product to cart

Three conversion rates between stages:

Session → conversation - what percentage of visitors start a chat
Conversation → click - what percentage of chatters click a product
Click → cart - what percentage of clickers add to cart

An additional daily chart breaks these metrics down over time so you can spot trends and react to changes.

3. Quality & Sentiment

If your AI is having bad conversations, nothing else matters. This tab gives you four quality metrics:

Handoff rate - percentage of conversations transferred to an agent (with absolute numbers)
Moderated messages - how many messages were caught by content moderation
Average conversation length - messages per chat
Median duration - how many minutes a typical conversation takes (median, not average - more resistant to outliers)

Three charts:

Sentiment distribution - a pie chart showing proportions of positive, neutral, negative, and moderated messages
Chat ratings - how many conversations were rated positive, neutral, negative, or not rated at all
Daily sentiment trend - daily lines for each sentiment type so you can catch sudden shifts

A spike in negative sentiment after a product description change tells you something went wrong. Steady improvement confirms your AI is performing better over time.

4. Top Products

The AI recommends products hundreds of times a day. Which ones resonate with customers?

This tab shows a chart of the top 10 products and a detailed table with data for each product:

Recommendations - how many times the AI suggested this product
Modal opens - how many times a customer opened the product preview
Product views - how many times a customer clicked to see full details
Add to cart - how many times the product was added to cart
Total clicks - sum of all interactions
CTR (click-through rate) - ratio of clicks to recommendations, color-coded: green (≥20%), blue (≥10%), gray (<10%)

The table is sortable - arrange by recommendation count, clicks, or CTR.

This data is invaluable for assortment management. If the AI keeps recommending Product A but its CTR is below 10%, the description, images, or pricing needs attention.

5. Credit Usage

AI costs money. It's worth knowing exactly where it goes.

Three metrics:

Credits used - total credits consumed in the selected period
Conversations - how many conversations generated those costs
Remaining balance - credits left in your account

Two views:

Daily usage chart - how credit consumption trends over time
Breakdown by store - if you have multiple sites connected to Upsellic, see how many credits each one uses, with percentage share of the total

From Data to Decisions

Every chart in the Reports module is designed to lead to a concrete action:

You see...	You do...
Spike in negative sentiment on specific days	Investigate what changed - maybe a new promotion is creating confusion
High recommendations, low CTR for Product X	Improve the product description, images, or pricing
Low session → conversation rate	Change the widget greeting or its position on the page
Increasing handoff rate	Check if the AI is missing knowledge on a specific topic
One store consuming 80% of credits	Analyze whether that's proportional to its traffic and revenue

Filtering and Date Ranges

Date range - last 7, 30, or 90 days
Store selection - when you have multiple sites, analyze each one separately
Real-time data - no waiting for nightly batch processing

Available Now

The Reports module is available for every Upsellic account. Go to Dashboard → Reports and start exploring.

No additional setup. No data migration. If you've been using Upsellic, your data is already there waiting to be analyzed.

The Reports module is available now, no additional setup needed. If your AI is having conversations, the data is already waiting for you in Dashboard → Reports.

Check your reports