The moment an AI assistant starts answering your Telegram profile, a new question appears: is it any good? Vibes are not an answer. Scalegram treats the assistant like a member of the team and gives it what a team member gets: a scoreboard, reviewed weekly.
The Numbers on the Board
- Conversations handled, over time: how much of your inbound the assistant absorbed, and what share of all conversations that is.
- Response time: average and distribution, hour by hour. The metric that justifies the assistant's existence, since speed is the thing no human can supply at 4am.
- Qualification rate: how many conversations produced the answers your playbook asks for, filled into the contact record.
- Closes: conversations that reached the goal you defined, a sent sign-up link, a confirmed deposit intent, a booked call.
- Handovers: how often a human was called in, for which triggers, and what happened after.
- Busy hours: when your inbound actually arrives, which regularly surprises operators and occasionally reshapes their ad scheduling.
Everything is charted in the same glass interface as the rest of the workspace, per day, week and month, filterable by connection.
Reading It Like an Operator
The scoreboard's purpose is not admiration; it is targeting the next playbook edit. Low qualification rate: the questions are awkward, rewrite section three. Strong qualification but weak closes: the pitch or the objection answers are soft, sharpen sections four and five. Handovers spiking on one topic: either give the assistant a real answer for it or accept that topic as human territory and route it faster. One metric, one edit, one week: that loop is the whole methodology.
The Assistant Feeds the Pipeline Numbers Too
Because every assistant conversation lands in the pipeline with stage and tags, its work shows up downstream in the funnel: leads it qualified appear in your stage counts, deposits it set up appear in the affiliate layer's revenue view, and campaigns it answered for appear in per-link results. The AI is not a separate report; it is a contributor inside the same business.
Read the transcripts. Numbers say where to look; the conversation mirror says what happened. Ten minutes of reading the assistant's best and worst chats each week teaches you more about your own funnel than any chart, and the next playbook edit usually writes itself.
"If you cannot measure the assistant, you are just hoping. We gave it the same dashboard discipline a sales team gets, and the playbook improves week after week because the numbers say where."
— Alex Onta, Executive Director, SINGUARD
Key Takeaways
- The assistant gets a real scoreboard: volume, response times, qualification, closes, handovers and busy hours.
- Each weak metric maps to a specific playbook section: measure, edit one thing, compare next week.
- Assistant activity flows into the same pipeline and revenue views as human work, not a separate silo.
- Transcript reading in the conversation mirror is the qualitative half of the review.
Frequently Asked Questions
What counts as a close for the assistant?
You define it: a goal action such as sending the sign-up link, receiving a deposit confirmation message, or booking a call. The analytics count whatever your playbook is trying to achieve.
Can I compare periods after changing the playbook?
Yes, the charts run per day, week and month, so a playbook edit on Monday is legible in the following weeks' lines. One edit at a time keeps the comparison honest.
Does the assistant know about ad campaigns?
Leads arriving through tracking links come pre-attributed with their source, the assistant's conversations inherit it, and per-link stats therefore include conversations and outcomes the assistant produced for each campaign.