Support dashboard metrics explained, and which ones to ignore

Updated 2026-09-24 · 5 min read · By the Convo360 team

First response time

How long a customer waited for a human to engage. It is the number customers feel most directly, and the easiest to improve dishonestly.

Read it at the ninetieth percentile rather than as an average, because the slowest tenth is where complaints and lost customers live. Read it in working minutes, not clock time, or every weekend evening looks like a failure.

Pair it with resolution time. A team can drive first response down with holding replies while nothing actually finishes faster, and only the pair shows that.

One practical warning: decide once whether an automated acknowledgement counts, write it down, and never change it quietly. A team that changes the definition mid-year loses the ability to compare anything, and somebody will eventually present a graph showing a dramatic improvement that was a definition change.

Resolution time and first-time fix

Resolution time measures how long the issue took to finish. It is the more honest of the two speed metrics and the harder one to influence, because it frequently depends on a courier, a supplier or a refund cycle outside your control.

First-time fix is the share of issues resolved without ever being reopened. It answers a question customers care about more than speed: did I have to come back?

The two together describe quality. Fast resolution with a poor first-time fix means issues are being closed rather than solved, which shows up a week later as a bigger queue and an angrier customer.

Both are worth splitting by tag once you have a few months of data. Resolution time in aggregate is dominated by whatever your slowest category is, which can make a team look sluggish when in fact one type of issue depends entirely on a third party. Refunds are the usual culprit in Indian D2C.

SLA adherence and backlog

SLA adherence is the percentage of tickets that met their deadline. Always read it by priority. A comfortable overall figure routinely hides the fact that every urgent ticket was late while a pile of low-priority ones were answered instantly.

Backlog is simply how many issues are open right now. It is a capacity signal, not a quality one, and it is the number most likely to be misread. A rising backlog during a festive peak is expected. A rising backlog in a flat month means work is arriving faster than it finishes, and the cause is usually one dependency rather than general slowness.

Backlog also needs an age view, not just a count. Forty open tickets where the oldest is two days is a healthy team mid-week. Forty where six have been open for three weeks is a different problem entirely, and the count alone cannot tell them apart. Sort by age occasionally and deal with the tail.

Reopen rate, the one nobody watches

Reopen rate is the share of resolved issues the customer came back on. Almost no small team tracks it, and it is the single most useful number on the dashboard.

The reason is that it catches the damage every other metric encourages. Push first response down and holding replies rise. Push tickets-per-agent up and answers get thinner. Push resolution time down and tickets get closed before they are solved. All three show up as a rising reopen rate, and nothing else shows them at all.

For this to work, a reopen has to mean something. In Convo360 a customer replying within seven days of a resolution reopens the same ticket rather than creating a new one, and the clocks restart from that moment, so the count reflects real recurrence rather than an accident of how tickets were filed.

Read a handful of reopened tickets each week rather than only counting them. The cause is almost always visible within a minute or two: an answer that was correct but not actionable, a resolution recorded before the customer confirmed anything, or a promise made to a customer that depended on somebody else remembering.

Numbers that look useful and are not

Three get quoted often and deserve suspicion.

• Total tickets. Rises with sales and with marketing. It tells you about volume, not about performance, and punishing a team for it discourages them from recording work.
• Average handling time. Lower is not better. The fastest agents are frequently the ones producing reopens, and the metric actively rewards it.
• Satisfaction score with a low response rate. A 4.6 from eight per cent of customers is an anecdote with a decimal point.

Where these belong is as context beside a metric that measures something, not as targets in their own right.

A fourth deserves mention because it is fashionable: deflection rate, the share of conversations handled without a human. It is genuinely useful when paired with reopen rate and satisfaction. On its own it rewards a bot that closes conversations rather than one that answers them, and customers experience that as being fobbed off.

How to run a useful monthly review

Thirty minutes, five questions, one action. What happened to the ninetieth percentile of first response? Did reopen rate move? Which priority missed its deadline most? What did the tags say this month was about? And what single upstream change would remove the most tickets next month?

Convo360 puts these on one dashboard alongside satisfaction, channel mix and a fourteen-day opened-against-resolved trend, so the review is reading rather than assembling. If you use the call features, answered, missed and abandoned calls sit in the same place.

Write down the one action and check it next month. A review that produces analysis and no change is a meeting; a review that removes one recurring cause is a system getting better.

Keep the review short enough that it happens every month. A thirty-minute meeting that runs reliably beats a two-hour analysis that happens twice a year, because most of the value is in noticing a direction early rather than in measuring it precisely.

Frequently asked questions

Related reading

Available across India — see WhatsApp & AI calling automation by state and city.