Configure

Monthly Performance Report

On the 1st of each month, Quincer AI emails an executive recap of what the AI actually did — volume, resolution, customer experience, and an estimated support cost saved, with every assumption shown. It's written for the budget owner, not the daily operator.

Reports are per brand. If your account has several brands, each one gets its own report — a blended org-wide average would hide which brand is carrying the load and which needs work, so we never send one.

What's in it

• The bottom line — one sentence: conversations handled, resolution rate, cost saved, CX score.
• Headline metrics — conversations, resolution rate, median first response, and CX score, each with a month-over-month delta (the first report is a baseline, so deltas start the month after).
• Coverage — how many sessions became real conversations, broken down by channel.
• Customer experience — sentiment (CX score) and CSAT survey results — average rating, self-reported resolution, response rate, and sample size. This section always appears: if you haven't turned on sentiment scoring or the CSAT survey for a brand (or none was captured that month), it simply says “no data captured this month” and points you to where to switch it on.
• Trajectory — your resolution rate over the last 6 months, so the trend is obvious at a glance.
• ROI — the estimated support cost avoided, as a range, with the full calculation shown (see below).
• Top actions & gaps — what visitors asked the AI to do, the content gaps worth closing, and 2–4 recommended next moves.

How the ROI is modelled

The dollar figure is a modelled estimate, not booked accounting — and it is deliberately conservative, so it survives a finance review. We start from the conversations the AI resolved with no human reply, then apply a discount for the ones that would have self-served or been abandoned anyway:

• AI-resolved conversations × escalation likelihood = the conversations a human would actually have handled.
• × average handle time = agent hours saved.
• × loaded agent cost per hour = support cost avoided (shown as a low–high range).

Every assumption used is printed in the report itself, and the figure is rounded to the nearest $100 — no false precision. Below 50 resolved conversations in a month we suppress the dollar estimate entirely and show coverage/resolution only, because a small sample makes a dollar figure unreliable.

Tune your assumptions

Open Settings → Reports to adjust the cost model to your real support economics. These assumptions are set once for your account and apply to every brand's report:

• Loaded agent cost per hour — wage + benefits + overhead (default $34.80).
• Average handle time — minutes a human would spend per resolved conversation (default 7).
• Escalation likelihood — the conservative discount: the share of AI-resolved chats that would otherwise have reached a human (default 65%, allowed 20–95%).

From that same screen you can preview last month's report for the brand you currently have selected in the top bar, rendered with your current assumptions, before it ever sends. Switch brands to preview another.

Who gets it & opting out

Each brand's report goes to that brand's admins on the 1st, covering the month that just ended — the people who own the brand (its owning account's members, plus anyone invited to the brand as an owner or admin). Invited agents don't receive it. We don't send a report for a brand with no activity. Anyone can turn it off under Settings → Notifications → Monthly performance report (it's a separate switch from product updates), and the master unsubscribe turns off everything.