Reports

Support generates more data than almost any other function in a business, and most of it is never looked at. The reports below are worth the time — but only if you are clear about which question each one answers, and which ones are dangerous to read as performance measures.

The reports

All of them can be filtered by date range and exported.

Conversation report

The base picture: incoming volume, resolutions, first response time and resolution time over time.

Read it for shape, not level. The absolute numbers mean little in isolation; the pattern means a lot.

• A weekly rhythm tells you where to staff. Almost every business has one, and almost none staff to it.
• Volume rising faster than customer count means something in the product or process is generating contacts.
• Resolution time rising while response time stays flat means conversations are being picked up promptly and then stalling — usually waiting on someone outside support.

Agent report

Volume and speed per person, and the report most likely to be misread.

It is genuinely useful for spotting imbalance: one person carrying visibly more than everyone else usually means your capacity settings are wrong, not that they are heroic and the others are lazy.

It is a poor measure of performance, because conversation counts do not describe difficulty. The agent who takes the hardest cases will look slowest on every number here, and if you manage on this report you will teach your team to avoid hard conversations. Read it alongside CSAT and alongside what you know about who handles what.

Team and channel reports

The same figures, grouped usefully.

Channel report answers where your load actually comes from — which frequently surprises people who have been staffing to a different assumption. It also shows which channel is slowest, and slowness is usually routing or staffing rather than the channel itself.

Team report compares groups. If one team is consistently slower, look at their volume per person before concluding anything about the team.

Label report

The most commercially interesting report in Tola, and the most neglected.

It tells you what customers contact you about. Not what you think they contact you about — what they actually do. delivery-delay climbing month over month is an operations problem being reported through the support queue. how-do-i dominating is a product or documentation problem.

Two conditions make it useful: your labels must be applied consistently, and there must be few enough of them to mean something. Automate the application with workflows wherever you can — human labelling is inconsistent by nature.

Take this report to whoever owns the product. It is the closest thing support has to a direct line into what the business should fix.

CSAT report

Ratings and written comments from customers after resolution.

The score is less useful than the comments. A drop from 4.6 to 4.4 is noise; five people writing "I had to explain it three times" is a specific, fixable problem.

Two cautions. Response rates are low and skewed — people who reply are disproportionately delighted or furious. And survey fatigue is real; asking after every single conversation lowers both your response rate and your score. Ask selectively.

Read the comments weekly. Read the score monthly.

SLA report

Which conversations met their targets and which breached, by agent, team and channel. See SLA policies.

Read it for clustering. Breaches on one weekday are a staffing gap. Breaches on one channel are routing or capacity. Breaches spread evenly across everyone mean your target is wrong, not your team.

Bot report

How much your bots and AI assistants handled, and how often they handed over to a human.

The number to watch is the handoff rate, and to watch its direction rather than its value. A rising handoff rate means your knowledge base has drifted out of date, or customers are asking about something new — both are signals to act on, and both are invisible if you only look at the total.

Also compare CSAT on AI-handled conversations against human-handled ones. If it is materially worse, narrow what the assistant answers rather than accepting the trade.

How to actually use these

Most teams either ignore reports entirely or check them daily and react to noise. Neither works.

Weekly, fifteen minutes. Unattended conversations, SLA breaches, and CSAT comments. These are the ones where something is actively going wrong and you can still do something about it.

Monthly, an hour, with the team. Volume trends, the label report, response times against last month. Ask one question: what is the biggest repeated thing here, and is it a workflow, a canned response, a help center article, or a product fix? Every recurring support problem is one of those four.

Quarterly, with the wider business. The label report and volume trend. This is where support stops being a cost centre and starts being the part of the company that knows most about what customers actually experience.

Two rules that save a lot of wasted time: compare against yourself, not benchmarks — your trend is the only meaningful comparison. And do not put individual agent numbers on a wall. It reliably produces faster, worse support, because the fastest way to improve every number on this page is to close conversations before they are resolved.