Chat support metrics that matter: first response time, resolution time, CSAT and SLAs

The chat support metrics that matter most are first response time (how fast a customer gets a first reply), time to resolution (how long until the issue is solved), customer satisfaction (CSAT) (how customers rate the help) and SLA compliance (how often you keep the response and resolution promises you set). Track those four per channel and per team, add volume and backlog for staffing, and you’ll see most problems before customers complain.

The rest is detail, but the detail decides whether the numbers are honest. This guide defines each metric with a formula, shows how messaging channels like WhatsApp and Instagram change the picture, and walks through building SLAs that your team can meet.

Why chat support metrics are different from email and phone

Chat sits between email and phone. Customers expect quicker answers than by email, but conversations pause and resume over hours, unlike a phone call. Messaging apps add two more twists:

• Reply windows. On WhatsApp, a 24-hour customer service window opens when the customer messages or calls you; after it closes you can only send pre-approved templates. Messenger and Instagram also give businesses 24 hours to respond. A slow reply isn’t just a bad experience; it can cost you the ability to reply freely.
• Asynchronous customers. The customer may answer two hours later. If your clock keeps running while you wait for them, your resolution time punishes the team for the customer’s pace.

So define every metric precisely, and decide upfront which waiting time belongs to you and which to the customer.

The core chat support metrics (with formulas)

Metric	What it tells you	How to calculate
First response time (FRT)	How long customers wait for a first answer	Time from first customer message to first reply; report the median
Average response time	How responsive the whole conversation feels	Average of all gaps between a customer message and your next reply
Time to resolution	How long it takes to solve the issue	Time from first message to conversation marked resolved
First contact resolution (FCR)	How often one conversation is enough	Resolved without reopening or a follow-up contact ÷ all resolved × 100
CSAT	How satisfied customers are	Ratings of 4–5 ÷ all ratings × 100
SLA compliance	How often you keep your promises	Conversations within target ÷ all conversations measured × 100
Backlog	How much work is waiting	Open, unassigned or awaiting-reply conversations at a given time
Reopen rate	How often “solved” wasn’t solved	Reopened conversations ÷ resolved conversations × 100

First response time

First response time is the metric customers feel most. Three rules make it trustworthy:

1. Use the median, not only the average. One conversation that waited all weekend can distort an average; the median shows the typical experience. Track the 90th percentile too, to see the long tail.
2. Separate automated and human replies. An instant greeting is useful but it isn’t an answer. Report “first reply” and “first human reply” as two numbers.
3. Measure in business hours if you don’t staff 24/7, and say so on the report. Then track after-hours volume separately, because those customers still waited.

Time to resolution

Define what “resolved” means in your team: the customer confirmed, the agent closed the conversation, or no reply for a set time after a solution. Pick one and keep it.

If your tool supports it, pause the clock while waiting for the customer. Otherwise, at least report time to resolution next to the number of customer replies, so long conversations caused by slow replies are visible.

CSAT, NPS and CES

• CSAT asks about this conversation. Best for support quality.
• NPS asks how likely the customer is to recommend you. It reflects the whole relationship, not one chat.
• Customer effort score (CES) asks how easy it was to get help. Useful when you’re changing processes, for example adding an AI assistant.

For chat, a short CSAT question right after the conversation gets the best response. A simple version:

Thanks for chatting with us today. How would you rate the help you got?
[ Great ]  [ OK ]  [ Not good ]

Then ask one open follow-up only when the answer is negative: “Sorry about that. What should we have done differently?” Our list of customer satisfaction survey questions has more wording options.

On WhatsApp, send the survey while the customer service window is still open, so it goes as a normal message. Later, it would need an approved template.

Volume, backlog and workload

These don’t measure quality, but they explain it. Track:

• Conversations started per hour and day, per channel.
• Conversations per agent per shift.
• Unassigned conversations older than a few minutes.
• Conversations close to a reply-window deadline.

When FRT gets worse, the cause is usually here: a volume spike, a channel nobody watches, or an uneven split between agents.

Metrics for AI assistants and chatbots

If an AI assistant answers first, measure it on its own terms:

Metric	Definition	Watch out for
AI resolution rate	Conversations closed by the assistant with no human involved ÷ all conversations it handled	Customers who gave up count as “resolved” unless you check CSAT or reopen rate
Handover rate	Conversations passed to a person ÷ all conversations the assistant handled	Too low can mean the assistant blocks escalation
Time to handover	Time from customer request for a person to agent pickup	Long waits after the bot says “connecting you” hurt CSAT most
AI CSAT	CSAT for AI-only conversations	Compare with human CSAT on the same topics

Our guide to testing an AI agent before launch explains how to measure resolution rate honestly, and AI-to-human handoff covers the handover itself.

How to set chat support SLAs

An SLA is a measurable promise: a target, a scope and a percentage. For example, “We send a first human reply to 90% of priority conversations within 10 business minutes.” The targets below are an example structure, not industry benchmarks; set yours from your own data.

Priority	Example trigger	First human reply target	Resolution target	Escalation if at risk
Urgent	Payment failed, service down, VIP customer	5 business minutes	Same business day	Team lead and senior agent
High	Order problem, booking change	15 business minutes	1 business day	Team lead
Normal	Product questions, general requests	1 business hour	2 business days	Reassign to available agent
Low	Feedback, partnership requests	4 business hours	5 business days	Weekly review

Six steps to an SLA your team can keep

1. Measure for two to four weeks first. Know your current median and 90th percentile per channel.
2. Define priorities by trigger, such as keywords, customer tier, channel or the topic chosen in a menu.
3. Set targets slightly better than today, not the ideal. Raise them once you hit them consistently.
4. Add an “at risk” stage. Alert before the breach, for example at 75% of the target time, so someone can act.
5. Respect channel windows. On WhatsApp, Messenger and Instagram, every target must leave time to reply within 24 hours of the customer’s last message.
6. Publish the promise. Tell customers in your greeting or away message when to expect a reply.

Hi, thanks for your message. Our team replies Monday to Friday, 9:00–18:00, usually within 15 minutes. Outside those hours, we’ll answer first thing the next working morning.

How to review support metrics every week

Keep the review short and the same every week:

1. SLA compliance per priority. Which breaches happened, and why?
2. FRT median and 90th percentile per channel. Which channel slipped?
3. CSAT with the number of responses, and every negative comment read in full.
4. Backlog at the start of each day.
5. AI resolution and handover rates, with a sample of AI conversations read by a person.
6. One change to try next week.

How MonoChat tracks chat support metrics and SLAs

MonoChat measures support across every channel in the unified team inbox, so WhatsApp, Instagram, Messenger, TikTok, Telegram and web chat use one definition of each metric.

Dashboards and conversation analytics

The analytics page lists conversation intelligence with response time and SLA tracking, CSAT and sentiment insights and agent productivity metrics, in KPI dashboards per team, channel, campaign and agent, with drill-down reporting and scheduled exports. You can also ask your data questions in plain language. The multi-agent call center adds real-time dashboards, agent KPIs and SLA monitoring.

SLA rules and escalation

With escalation and SLA workflows, MonoChat tracks first response and resolution SLAs with business hour rules and waiting time alerts. Rules detect overdue replies, negative sentiment, VIP status or urgent intent, and route the case to the right team, manager or senior agent with a conversation summary and the customer’s history.

Automated CSAT surveys

CSAT survey automation sends a survey through the same channel after a support interaction ends, collects ratings and comments, and routes low scores and negative feedback to the right team for recovery, while positive feedback feeds reporting.

To start, sign up for the free Basic plan (1 agent, 1 channel) at app.monochat.ai. When you add agents and channels, compare plans on the pricing page. If you’re still bringing your channels together, read our guide to the omnichannel inbox first.

Sources

• Send messages and the customer service window (Meta for Developers, WhatsApp)
• Messenger Platform and Instagram Messaging API policy (Meta for Developers)
• Instagram Messaging API (Meta for Developers)