First Response Time: The One Support Metric Customers Feel Instantly

Most support metrics are invisible to customers. Ticket backlog, agent occupancy, average handle time. Useful internally, felt by nobody outside the team.

First response time is different. It's the one number your customer experiences directly, with a clock running in their head. They sent a message. They are now waiting. Everything they think about your brand in the next few minutes is shaped by how long that wait lasts.

That makes first response time worth getting right before you optimise anything else. Here's how to define it properly, calculate it without fooling yourself, and improve it.

What first response time actually measures

First response time is the gap between a customer contacting you and receiving their first substantive reply.

"Substantive" is doing a lot of work in that sentence. An auto-responder saying "We've received your message and will get back to you within 24 hours" is not a first response. It contains no information the customer didn't already have. If your help desk counts it, your FRT is fiction.

A real first response does at least one of these things:

• Answers the question
• Asks a specific clarifying question that moves things forward
• Confirms a concrete action has been taken ("I've cancelled order #4821")
• Gives a genuine, specific timeframe from a named person

FRT is not resolution time. A customer can get a reply in 30 seconds and still wait four days for their refund. Both matter, they measure different failures, and confusing them is a common reporting error in small teams.

How to calculate first response time

The formula is straightforward:

First response time = (Sum of all first response durations) ÷ (Total number of tickets responded to)

Worked through step by step:

1. Pick your window. A rolling 30 days is standard. Anything shorter gets noisy; anything longer hides trends.
2. Pull every ticket that received a first reply in that window. Exclude tickets still awaiting a first response. They belong in a separate "unanswered" count, or they will silently flatter your number.
3. For each ticket, record the timestamp of the customer's first message and the timestamp of your first substantive reply. Subtract.
4. Sum those durations and divide by the ticket count. That's your average FRT.
5. Now calculate the median as well, sort every duration and take the middle value.

The fifth step is the one teams skip, and it's the one that tells the truth.

Say you handle 100 tickets. Ninety get answered in four minutes. Ten arrive on Friday evening and sit until Monday. Your average FRT balloons past four hours while your median stays at four minutes. Neither is wrong, but only one explains why ten customers are furious.

Report both. Use the median for "what a typical customer experiences" and the 90th percentile for "how bad it gets at the tail".

What quietly distorts your first response time

Before you benchmark yourself against anyone, audit these. Most reported FRT numbers are wrong in at least one of these ways.

First response time benchmarks by channel

Expectations vary enormously by channel, because the channel itself signals urgency. Nobody expects an instant email reply. Everybody expects an instant chat reply, that's why they chose chat.

These are the service-level targets ecommerce teams commonly work towards, not measured industry averages. Treat them as a planning framework, then set your own targets from your actual data.

Two notes. Chat is unforgiving: a widget that takes eight minutes to answer is worse than no widget, because it made a promise it didn't keep. And pre-purchase messages deserve tighter targets than post-purchase ones, someone asking "will this fit a 90cm waist?" has a card in their hand. For chat specifically, see our live chat response time benchmark.

How to improve first response time

In rough order of effort-to-impact:

1. Segment before you optimise. Split FRT by channel, by hour of day, and by ticket type. The problem is almost never uniform, it's usually one channel, one shift, or one category of question.
2. Find your repetitive questions. In most stores, order status dominates. Tackling "where is my order?" alone can transform the queue. Start with reducing WISMO tickets.
3. Fix the pages that generate tickets. If half your enquiries are about delivery times, the answer belongs on the product page and in the checkout, not in your inbox.
4. Write proper saved replies. Not templates you paste blindly, but strong 80% drafts an agent edits in fifteen seconds.
5. Triage on arrival. Route by intent so pre-sale questions and angry customers jump the queue instead of waiting behind a routine address change.
6. Set an internal SLA per channel and make it visible. Metrics nobody watches don't move.
7. Cover the hours you don't staff. This is where most FRT damage happens: nights, weekends, and the gap between time zones. See after-hours customer support.
8. Automate the answers that never change. Order tracking, returns policy, shipping windows, sizing. These are pure latency, not judgement.

For the tactical version of this list, read how to answer customer questions faster.

Illustrative example: Consider a mid-sized Shopify apparel store handling 900 tickets a month. Roughly half are order status and delivery questions. If those were answered on arrival rather than joining the queue, the remaining tickets would inherit a much shorter wait, not because agents got faster, but because they stopped queueing behind questions that never needed a human.

Where AI fits, and where it doesn't

Automation is the only way to hold a sub-minute first response time around the clock without staffing a night shift. An AI support agent replies the moment a message arrives, at 3am on a bank holiday, in whatever language the customer wrote in.

AskZoye is built for exactly this on Shopify: it answers in about three seconds, covers 50+ languages, handles order tracking, returns and shipping questions directly, and goes live in under 60 seconds with no code. Because it's connected to your store data, its first response is usually also the last one needed.

But be honest about the limit. AI should not be the first response to an emotionally charged complaint, a genuine service failure, or anything needing a policy exception. A three-second reply to "my order arrived smashed and it was a gift" is not a win if it's the wrong reply. Design it so AI takes the repetitive volume instantly and anything sensitive reaches a human fast, with context already gathered.

The goal isn't "no humans". It's "no waiting for things that shouldn't require waiting".

The bottom line

First response time is the metric your customers feel before they feel anything else. It's also one of the easiest to accidentally lie to yourself about, auto-replies, business-hours clocks and excluded tickets can make a struggling queue look healthy on a dashboard.

Measure it honestly: substantive replies only, segmented by channel, median alongside average. Then attack it where the volume is, which for almost every Shopify store means the repetitive order and shipping questions that pile up overnight. Pair this with the wider set of customer support KPIs and you'll have a clear picture of what your queue is really costing you.