Time to Resolution: Define, Measure and Reduce It

Time to resolution measures how long it takes from the moment a customer contacts you until their issue is actually solved. Of all support metrics, it comes closest to what the customer experiences: not how fast you said hello, but how long they had a problem. It is also easy to measure wrongly, which is why many teams report a number that looks fine while customers wait for days.

This guide explains how to define time to resolution so it means something, how to calculate it in a small team, how it differs from first response time and handle time, and the changes that genuinely shorten it.

What Time to Resolution Measures

Time to resolution (often shortened to TTR, and sometimes called resolution time) is the elapsed time between the start of a support request and the moment it is resolved. For a single ticket:

Resolution time = time the issue was marked solved − time the customer first contacted you

It is a close cousin of mean time to recovery in IT operations, which measures how long systems stay broken; resolution time applies the same idea to customer problems.

Across many tickets, you report a typical value for a period, such as a week or month. That sounds simple. The difficulty is in four decisions that change the number completely:

When does the clock start?
When does it stop?
Does it run outside business hours?
What happens when a ticket is reopened?

Make each decision once, write it down and keep it fixed. A metric whose definition changes every quarter cannot show a trend.

Defining the Clock

Start: the customer’s first message

Start the clock when the customer first writes, not when an agent first looks at the request. If a ticket sat unassigned for a day, that day was part of the customer’s wait.

Stop: the issue is solved, not the conversation closed

Stop the clock when the ticket is marked Solved, meaning your team has done what was needed and told the customer. Do not use the moment it was archived or Closed, which is often days later for administrative reasons.

Business hours or calendar hours?

Both are legitimate, and they answer different questions.

Calendar hours show the customer’s real experience. A ticket raised on Friday evening and solved on Monday morning took about 60 hours from their point of view.
Business hours show your team’s efficiency. The same ticket took perhaps one or two working hours.

Report calendar time as your headline number, because it is the customer’s reality, and business-hours time as a secondary number for capacity planning.

What about time spent waiting on the customer?

If you asked the customer for a photo and they replied three days later, should those three days count? Many teams exclude time in Pending status, on the grounds that the customer controlled it. That is defensible, but only if Pending is used honestly. If agents park difficult tickets as Pending, excluding Pending time hides exactly the delays you most need to see. When in doubt, report both versions.

Reopened tickets

If a solved ticket is reopened because the fix did not work, the original resolution was not real. Measure to the final solve, not the first one. Otherwise, solving tickets too early improves the metric while making customers wait longer.

Average or Median?

Resolution times are heavily skewed. Most tickets are solved in minutes or hours, and a few take weeks. A handful of long-running cases can pull the average far above what a typical customer experiences.

That is why the median is usually a better headline figure: half of your tickets were resolved faster than this, half slower. Add a high percentile, such as the 90th, to show the long tail. A small team might report something like:

Median resolution time: 3 hours
90th percentile: 2 days

The median tells you about the typical customer. The 90th percentile tells you about the customers most likely to leave a bad review. Improving one without the other is common, and both deserve attention.

Time to Resolution vs Other Support Metrics

Resolution time is often confused with metrics that measure something else.

Metric	Measures	From the customer’s view	Typical trap
First response time	Time until the first human or AI reply	“Did anyone notice me?”	Fast greeting, slow solution
Average handle time	Active agent time spent on a conversation	Invisible to the customer	Rushing agents to look efficient
Resolution time	Elapsed time from first contact to solved	“How long did I have a problem?”	Solving too early, excluding Pending dishonestly
Resolution rate	Share of conversations resolved	“Did they fix it at all?”	Counting unresolved chats as resolved
Repeat contact	Customers coming back about the same issue	“Did it stay fixed?”	Ignoring contacts in other channels

A healthy support operation watches at least first response time and resolution time together. Fast first replies with slow resolution mean you are acknowledging people but not helping them. Our guides to first response time and average handle time cover those two metrics in depth.

How to Measure Time to Resolution in a Small Team

You do not need special software to start. You need consistent statuses and a spreadsheet.

Step 1: Make statuses reliable

The metric depends entirely on when tickets are marked Solved. If some agents mark tickets Solved before telling the customer, or leave solved tickets Open for days, the number is noise. Agree on what Solved means first.

Step 2: Sample, do not boil the ocean

Take 50 to 100 tickets solved last month. For each, note the time of the customer’s first message, the time it was solved, and whether it was reopened. Add one column for the topic, such as delivery, billing or technical.

Step 3: Calculate median and 90th percentile

Any spreadsheet can do this. Calculate them overall and per topic.

Step 4: Read the slowest ten

The numbers tell you where; reading the slowest tickets tells you why. You will usually find the same few causes repeated.

Talkmio’s reports cover conversations per day, the share answered by Mio alone, hand-offs, first-reply time, ratings, busiest hours and team performance. For resolution time, the sampling method above works with any tool, and on Ultimate and higher plans you can also export reports to CSV for your own analysis.

What Makes Resolution Slow

When teams read their slowest tickets, the causes cluster into a few groups.

Missing information at the start

The first reply asks for the order number, the customer replies hours later, the second reply asks for a photo, and another day passes. Each round trip adds hours or days. A pre-chat form or a well-designed first question collects what you need in one go.

Handoffs between people

Every time a ticket moves from one person to another, it waits in a new queue, and the new owner has to understand it. Good routing to the right agent on the first attempt removes many of these delays.

Waiting for a decision

Refunds above a limit, exceptions to policy, and goodwill gestures often wait for a manager. If the same decision is made the same way every time, give the frontline team the authority to make it.

Waiting on another department or supplier

Warehouse, finance, a courier, a developer. You cannot always speed them up, but you can make the dependency visible with an internal note, chase it on a schedule and keep the customer updated.

Answers that do not exist anywhere

If agents have to research the same question repeatedly, the answer belongs in your knowledge base. This is also what lets an AI assistant resolve the question without any human involvement next time.

How AI Changes Resolution Times

An AI assistant that answers from your own content can resolve routine questions in seconds, at any hour. Mio, Talkmio’s assistant, answers only from your website, FAQ and uploaded documents, and hands the conversation to your team when it cannot answer, when the visitor asks for a person, complains or asks about a specific order.

Two effects follow, and you should measure them separately:

The overall median drops, because many simple questions are resolved almost instantly.
The median for human-handled tickets may rise, because the easy ones no longer reach your team. That is not a failure; your team now works on harder problems.

Report “resolved by AI alone” and “resolved by the team” as two groups. Mixing them hides what is happening on each side.

Keeping Customers Informed While You Resolve

Some issues simply take time: a replacement part has to be shipped, a bank has to process a refund, a developer has to fix a bug. You cannot always shorten those, but you can change how the wait feels.

Say what happens next and when. “Finance will confirm the refund by Thursday, and I will write to you as soon as they do” turns an open-ended wait into a known one.
Update before the customer asks. A short message on the promised day, even if the answer is “still waiting”, prevents a follow-up chat and a repeat contact.
Record the promise. An internal note with the date means whoever is on shift that day keeps the promise.
Give one owner. A ticket assigned to a named person moves faster than one the whole team vaguely watches.

None of this changes the clock, but it is the difference between a customer who waited three days calmly and one who waited three days and left a one-star review.

Setting a Target

There is no universal benchmark worth quoting, because resolution time depends on what you sell and how complex your issues are. A clothing shop and a software company have very different normal ranges. Instead of borrowing someone else’s number:

Measure your current median and 90th percentile for two months.
Pick the topic with the slowest resolution and the most volume.
Fix one cause, such as missing information at the start or a decision bottleneck.
Measure again.

Improvements of this kind compound. Reducing back-and-forth on the most common slow topic often does more than any staffing change. For a broader view of which numbers to track, see customer support metrics that actually matter.

Frequently Asked Questions

What is time to resolution in customer support?

Time to resolution is the elapsed time from a customer’s first message until their issue is marked solved. It reflects how long the customer actually had a problem, which makes it closer to their experience than first response time. Teams usually report it as a median for a week or month, plus a high percentile to show the slowest cases.

Should resolution time use business hours or calendar hours?

Report both if you can. Calendar hours show the customer’s real wait, including nights and weekends, so they make a good headline figure. Business hours show how efficiently your team works during its own hours and are useful for capacity planning. Choose one as the main number and keep the definition fixed.

Why use the median instead of the average for resolution time?

Resolution times are skewed: most tickets are solved quickly and a few take weeks. Those few long cases pull the average far above what a typical customer experiences. The median shows the typical case, while a 90th percentile figure shows the long tail of customers most likely to be unhappy.

Should time spent waiting on the customer count?

Many teams exclude time in Pending status, because the customer controlled it. That is reasonable only if Pending is used honestly. If difficult tickets are parked as Pending to get them out of the queue, excluding that time hides real delays. Reporting both versions is the safest approach.

How is time to resolution different from first response time?

First response time measures how quickly someone first replied, while time to resolution measures how long it took to actually solve the issue. A team can reply within a minute and still take three days to resolve. Watching both together shows whether customers are being helped or just acknowledged.

How can a small team resolve issues faster?

Read the slowest tickets and fix the most common cause. Typical fixes are collecting the right information in the first reply, routing tickets to the right person first time, giving frontline staff authority for routine decisions, and writing recurring answers into the knowledge base so an AI assistant can resolve them instantly.

The Bottom Line

Time to resolution is the metric that matches what customers feel, as long as you define it honestly: start at the first message, stop at the real solve, count reopens, and report a median with a high percentile. Measure a sample of last month’s tickets this week, read the slowest ten, and fix the most frequent cause. An AI assistant can take the routine questions off the clock almost entirely. Try Talkmio free and compare your numbers before and after.

Time to Resolution: How Long Customer Issues Really Take