Your dashboard is not a report
The most common reporting mistake is forwarding the tool's own summary and assuming it speaks for
itself. It does not. A dashboard is built for someone who works in it daily and already knows what a
normal Tuesday looks like; an executive reading it once a month has none of that context and will
either ignore it or draw a conclusion you did not intend.
Leadership reports exist to support decisions: whether to hire, whether to extend hours, whether a
product problem is costing more than fixing it would. Everything in the report should serve one of
those questions. If a number cannot change a decision, it is decoration, and decoration is what
teaches people to stop reading your reports.
The practical test before sending is to ask what you want the reader to do differently. If the
honest answer is nothing, send a shorter update, or none at all that month. A report that arrives
only when it matters gets read.
Pick three numbers and hold them still
Changing which metrics you report is the fastest way to lose credibility, because it reads —
often correctly — as choosing whichever numbers look best this month. Pick three, define them
precisely in writing, and keep them stable for at least a year.
A defensible trio for most teams is volume, speed, and outcome: how many conversations you handled,
how quickly people got a first real reply, and how many were resolved without escalation. Those three
describe demand, service level, and effectiveness, which is enough to support most decisions a
leadership team makes about chat.
Write the definitions down and attach them to the report. Whether an automated greeting counts as
a first response, whether a conversation with no visitor reply counts as handled, whether an abandoned
chat counts as missed — all of these are judgement calls, and the value of the report depends far
more on holding them constant than on which way you decided.
Report a trend rather than a snapshot. One month's figures are mostly noise; three months of the
same three numbers is a direction, and a direction is what people can act on.
Add the thing the numbers cannot carry
Metrics tell a reader what happened and almost never why. The most valuable half-page in any chat
report is the short qualitative section: the three themes that came up repeatedly, and one or two
short quotes from real conversations.
Quotes work because they are concrete in a way that percentages never are. A line where a customer
explains exactly which part of the checkout confused them will move a decision that a deflection rate
would not, and it is the strongest argument you have for fixes that live outside your team.
Keep the themes to three, and tie each to something specific: how many conversations, what it costs,
what would fix it. A theme with a number attached is a proposal. The same theme without one is an
anecdote, and it will be treated as such.
Take care with quotes. Strip anything identifying, and never include personal details from a
conversation just because they make the point vivid. The reason to quote is to explain a pattern, not
to expose a customer.
Reporting a bad month
Every chat team eventually has a month where the numbers are worse and there is no flattering way
to present them. How that report is written determines how much room you get to fix it.
Lead with the number rather than burying it. A reader who finds the bad figure themselves, three
paragraphs in, stops trusting the framing of everything else. Say plainly what got worse, by how much,
and then explain the cause if you know it and say you are still investigating if you do not.
Separate the causes you controlled from the ones you did not. A response time that slipped because
two people were ill is a different conversation from one that slipped because volume grew thirty per
cent and staffing did not. Both are worth reporting; only one of them is a performance problem, and
conflating them either invites unfair criticism or wastes a genuine capacity argument.
Then say what changes and by when. A bad month with a named action and a date is a managed
situation. A bad month with reassurance attached is the thing that erodes confidence, because the
same numbers usually come back the following month.
What to measure
Track whether your report is actually read and acted on. If no decision in six months traces back
to it, the format is wrong or the audience is wrong, and it is worth asking directly which parts are
useful.
Keep your metric definitions under version control of some kind, even if that is a dated note at
the bottom of the report. When a number moves sharply, the first question worth asking is whether the
measurement changed rather than the performance.
Watch the gap between what your report says and what other teams believe about chat. A persistent
gap usually means the reporting is technically accurate and not landing, which is a communication
problem rather than a measurement one.