Chat Analytics in MARREK: What Your AI Consultant’s Statistics Tell You
Messages, dialogs, AI and manual replies, leads, peak hours, channels and the longest conversations — what each MARREK metric shows and how to use it to sell more in Instagram and Telegram.
An AI consultant that answers customers in Instagram, Telegram and on your website produces something more than replies: data. How many people write, when they write, which channel brings buyers, how much of the work the AI does and how much still lands on managers. MARREK collects all of it in the Analytics section of the dashboard — and this article explains what each chart means and what to do with it.
Why chat statistics matter
Most businesses selling through Direct and messengers have no numbers at all. They know sales went up or down, but not why: whether fewer people wrote, whether they wrote but did not buy, or whether messages simply went unanswered. Chat analytics answers these questions.
- See the real demand. The number of messages and dialogs shows how many people actually reach out — the top of your sales funnel.
- Measure the AI’s contribution. AI replies versus manual replies shows how much work the AI consultant takes off your team.
- Count leads, not chats. How many conversations turned into interested customers and how many into ready buyers.
- Plan your team’s time. Peak hours show when customers write, so managers are online when it matters.
- Invest in the right channel. The channel breakdown shows where your conversations come from.
What MARREK analytics shows
Analytics can be viewed for the last 7, 30 or 90 days. At the top are six key figures:
| Metric | What it counts | Why it is useful |
|---|---|---|
| Messages | All messages in the period — from customers, the AI and managers | Overall activity and load |
| Dialogs | New conversations started in the period | How many people reached out |
| AI replies | Messages sent by the AI consultant | The work the AI took off your team |
| Manual replies | Messages sent by the owner or managers | Where people still had to step in |
| Leads | Customers marked as interested or ready to buy | Conversations that turned into sales opportunities |
| Ready to buy | Customers marked as ready to buy | The warmest leads — the ones to close first |
Messages by day
A bar chart with the total number of messages for each day of the period; hover a bar to see the date and the exact count. It shows how demand changes over time: the effect of a new ad campaign, a Reels that took off, a sale, or a quiet week. Days with a sharp jump are worth opening in the chat section to see what people asked about.
New lead dynamics
A line chart of new leads per day. This is the chart to compare against your marketing: if messages grew but leads did not, the traffic brought curious visitors rather than buyers, or the catalogue and knowledge base do not answer their questions well enough. A lead is counted on the day its status last changed, so past days can shift slightly — the note under the chart says so too.
By channel
Horizontal bars with the number of messages from Telegram, Instagram and the website widget. It tells you where your customers actually are and which channel deserves more attention — more posts, more ads, or a link to the chat in a more visible place.
Message role distribution
A donut chart with the total number of messages in the centre and the split between leads (customers), the AI agent and the manager — each with a count and a percentage. A healthy picture is the AI answering almost every customer message, with managers joining only where a human is needed. If the manager share is growing, look at which questions the AI hands over — often one missing document in the knowledge base fixes it.
Peak hours of incoming messages
A heatmap of messages from leads by day of the week (Monday to Sunday) and hour (0 to 23); the darker the cell, the more messages. Time is shown in your time zone, which is stated under the chart. Most businesses discover that a large share of messages arrives in the evening and at weekends — exactly when nobody is at work. That is where the AI consultant earns its keep, and it is the data to use when planning manager shifts.
Top 5 longest dialogs
The five conversations with the most messages from leads, shown with the customer’s name and the message count. Long dialogs are usually either very interested customers or questions the AI could not answer quickly. Look them up in the chat section: the first kind are worth a personal follow-up, the second kind show what to add to your knowledge base.
The daily overview
Besides full analytics, the dashboard home shows today at a glance: unread dialogs, AI replies today and new leads today, plus the status of your channels and how much of your plan’s AI message limit has been used. It is the screen to open each morning.
How to use analytics to sell more
- Check weekly — the 7-day view is enough to spot changes early.
- Compare messages and leads. Growing messages with flat leads means improving the catalogue, answers or offer.
- Watch the manual reply share. When it rises, find the questions behind it and add answers to the knowledge base.
Frequently asked questions
Which time zone does analytics use?
Your own. Daily charts and the peak hours heatmap are calculated in your time zone, so “Tuesday 20:00” means your Tuesday evening.
Why can lead numbers for past days change?
A lead is counted on the day its status last changed. If a customer who was interested on Monday becomes ready to buy on Thursday, they move to Thursday. This keeps the numbers tied to real progress.
Does analytics show the content of messages?
No. Analytics works with counts only. Conversations themselves are read in the chat section, by the people who have access to it.
The takeaway
An AI auto reply without statistics is a black box. MARREK analytics shows how many customers reach out, how many become leads, when they write and how much of the work the AI does — enough to make decisions about marketing, staff and content with numbers instead of guesses. Read more about the benefits of an AI sales assistant.
