Live Chat for Support: Find Which Problems You Solve in One Chat and Which Keep Coming Back (Free Resolution Analyser)

Live chat for support works best when you know, topic by topic, which problems your team fixes inside the chat and which ones slip out as follow-ups or return a few days later. Most small teams never look. They answer, close the chat and move on, so the same refund question or login problem eats hours every week. This guide shows how to log two weeks of support chats in a simple three-part format, measure first-chat resolution for each topic, and decide what to do with each one: save a reply, write a help page, build a ticket path, or fix the cause outside chat. A free tool does the counting for you.

3fields per chat to log
2 weeksof chats is enough
5decisions per topic
15 minto log a busy day

⚡ In short

  • Log each support chat as topic, outcome (solved, ticket or came back) and minutes.
  • Measure first-chat resolution per topic, not just one overall number.
  • Quick, solved topics become saved replies. Slow, solved topics become help pages.
  • Topics that need checking get a ticket path with an owner. Topics that keep coming back need a fix outside chat.
Live chat for support in LiveChatBot: the LiveChatBot shared inbox with a conversation open
How it looks in LiveChatBot: the LiveChatBot shared inbox with a conversation open.

The problem

Support chats that end are not always support problems that end

A support chat has a natural ending. The customer says “ok thanks”, the agent closes the window, and the inbox looks cleaner. It feels like a win. But closing a chat is not the same as solving the problem behind it.

Some chats end because the problem really was fixed: the customer got the tracking link, reset their password and logged in, or learned how to download an invoice. Others end because the agent promised to “check and get back”, and the follow-up lives in someone’s memory. And some end politely, only for the same customer to return three days later with the same refund question, a little angrier.

If you only count chats, all three look identical. That’s why so many small support teams feel busy without getting ahead. They’re answering the same unresolved problems again and again, and they have no data to show which ones.

The fix is not a complicated dashboard. It’s a habit of noting, for each chat, what it was about and how it really ended. After two weeks you can see which topics you handle well, which ones always need a follow-up, and which ones are quietly broken somewhere else in the business. That’s the difference between answering chats and running live chat for support as a real service.

What to measure

First-chat resolution, topic by topic

The number that matters most is first-chat resolution: the share of chats where the customer’s problem was fixed in that chat, with no follow-up and no return visit about the same thing.

One overall figure hides too much. A team might solve 60% of chats on the first try, which sounds acceptable, while solving 95% of delivery questions and only 10% of refund requests. Those two topics need completely different fixes. So measure resolution separately for each topic.

For every chat, you need three facts:

  • Topic. A short label in your own words: “refund”, “login problem”, “delivery status”, “setup help”. Keep the list small; ten to fifteen topics is plenty.
  • Outcome. One of three: *solved* (fixed in the chat), *ticket* (needed a follow-up, a check or another person), or *back* (the customer returned about the same issue within a week).
  • Minutes. Roughly how long the chat took, from the first reply to the last. Round to the nearest minute; don’t time it with a stopwatch.

That’s it. Customer names, order numbers and transcripts stay where they are. You’re measuring patterns, not people.

Why “back” matters so much: a chat marked solved that comes back was never really solved. When the agent reopens it, change the original outcome to *back* rather than adding a fresh *solved*. It’s the most honest signal in the whole log.

Two-week routine

How to collect the log without slowing the team

Day 1Agree the topics

Write 10 to 15 topic labels on one shared note. Add a new one only if nothing fits.

Every chatLog one line

After closing, the agent adds topic | outcome | minutes to a shared sheet. Ten seconds per chat.

DailyMark the returns

When a customer comes back about the same issue, change their earlier line to back.

Day 14Analyse

Paste the lines into the free tool below and read the advice for each topic.

Day 15Act on three topics

Pick the three biggest problems and fix them before you log again.

Setting it up

Using live chat for support without losing follow-ups

Logging only helps if the chats themselves are handled in a way that makes follow-ups visible. Here is the minimum setup that keeps nothing hidden.

One shared inbox

Every support chat should arrive in one place that the whole team can see. In LiveChatBot, chats land in the team inbox, and anyone can pick them up from the browser or the Android app. When support lives in personal WhatsApp chats, follow-ups disappear with the phone.

A ticket for anything you can’t finish now

If an agent can’t fix the problem in the chat, the chat becomes a ticket before it closes, with an owner and a status: open, pending or solved. The customer gets a ticket page they can check without logging in. Our guide to turning a chat into a support ticket shows the steps, and the ticketing system page explains the statuses.

Offline messages count too

Support questions don’t stop at closing time. Set opening hours so that out-of-hours visitors see the away form and leave an email or phone number. Those offline messages can become tickets, so they join the same queue as daytime chats.

Saved replies for the answers you repeat

Any answer your team types more than twice a week should be a saved reply, with {name} to personalise it. That keeps the answer consistent no matter who sends it. See how to use saved replies.

With this in place, your log reflects reality: *ticket* means a real ticket exists, and *solved* means nobody owes the customer anything.

Free tool

Analyse your support chats by topic

Paste your log as topic | outcome | minutes, one chat per line. The tool groups by topic, works out the resolution rate, ticket rate, return rate and average time, and tells you which of five fixes each topic needs.

Paste a week or two of support chats, one per line, as topic | outcome | minutes. Outcome is solved (fixed in the chat), ticket (needed follow-up) or back (the customer came back about the same thing).

Runs only in your browser; nothing is sent or saved. Use topics only, never customer names or order numbers.

Reading the results

The five labels and what to do with each

The analyser gives each topic one label. Here is what they mean and the next step for each.

Saved reply candidate

Solved in the chat at least 70% of the time, and quick. This is support working well, but every agent is probably typing their own version. Save the best answer as a saved reply. In the example, *delivery status* (four of five solved in about two minutes) and *change address* fit here.

Solved, but slow

Usually fixed, but each chat takes more than eight minutes. The problem is understood; the explanation is just long. Write it once as a numbered, step-by-step saved reply, or a short help page with screenshots, and send the link. In the example, *setup help* and *invoice copy* take 9 to 15 minutes each. A help page could cut that to two.

Needs a ticket path

More tickets than solves. These are problems an agent can’t fix alone: refunds that need finance, damaged items that need photos and a courier claim. Don’t try to force them into one chat. Instead, make the ticket path fast and predictable: create the ticket in the chat, name the owner, and tell the customer exactly when they’ll hear back.

Keeps coming back

A quarter or more of these chats return. That is the clearest sign the real problem isn’t in chat at all. It might be a refund policy nobody can find, a confirmation email that doesn’t include tracking, or a login page that confuses people. Better chat replies won’t fix it; fixing the cause will.

Mixed results

Sometimes solved, sometimes not, with no clear pattern. Read three solved chats and three unsolved ones side by side. Usually one detail separates them: the agent had access to something, or the customer shared a screenshot. Make that detail standard.

Example

Same topic, two ways to handle it

Refund question (needs a ticket path)
🚩 Before“Let me check and get back to you.” Chat closes. Nobody owns the follow-up, and the customer returns in three days.
✅ After“I’ve opened ticket #[number] for your refund and assigned it to our accounts person. You’ll get an update by tomorrow 6 pm, and you can check it on this link anytime.”
Invoice copy (solved, but slow)
🚩 BeforeTen minutes of back-and-forth explaining which menu to open.
✅ AfterOne saved reply with four numbered steps and a link to a help page with screenshots. Illustrative.
Login problem (keeps coming back)
🚩 BeforeThe agent resets the password every time.
✅ AfterThe team notices most returns come from the same confusing login screen and asks the developer to add a clear ‘Forgot password’ link.

Speed and resolution

Fast replies matter, but resolved problems matter more

Most advice on support chat focuses on response time: reply within a minute, keep the customer engaged, avoid long silences. That advice is right, and our guide to live chat response time covers it in detail. But a fast first reply followed by an unsolved problem still leaves the customer unhappy.

Think of it as two separate promises. The first is “we’ll answer quickly”. The second is “we’ll fix it, or tell you exactly when it will be fixed”. A team that keeps both promises builds trust. A team that only keeps the first one trains customers to expect a polite reply and no result.

Your log measures the second promise. Track both side by side for a month and you’ll often find the topics with the slowest replies aren’t the ones customers complain about most. The complaints come from topics that keep coming back.

There’s also a staffing angle. Topics labelled *saved reply candidate* or *solved, but slow* are where time savings live. If *setup help* takes 14 minutes and comes up 20 times a week, a good help page can save the team several hours. That’s real capacity for the harder chats, without hiring.

If long pauses during checks are an issue, our guide to support chat silences has wording for “I’m still checking” moments.

Making it stick

A monthly loop for using live chat for support

The first two weeks of logging always produce a few quick wins. The real value comes from repeating it.

Once a month, run the log again for one week. Compare each topic’s numbers with last time. Did the saved replies push *delivery status* from 80% to 95%? Did the new refund ticket path stop customers coming back? Did a new topic appear because you launched a product or changed a policy?

Keep the comparison simple: one row per topic, with last month’s and this month’s resolution rate. Topics that improved get left alone. Topics that didn’t improve get a closer look at three real chats.

A few habits keep the loop healthy:

  • Keep topic labels stable. If “refund” becomes “refund request” next month, you can’t compare. Rename only when you must.
  • Let agents suggest fixes. The people answering chats usually know why a topic keeps coming back. Ask them first.
  • Share one number with the whole team. First-chat resolution for the top five topics, posted where everyone sees it, keeps attention on solving rather than closing.
  • Remove topics that disappear. If you fixed the cause and nobody asks any more, celebrate and take it off the list.

Over a few months, the log turns into a map of where your business makes customers ask for help. Some of the best fixes will have nothing to do with chat at all.

Checklist

Resolution logging, start to finish

Support log progress

Getting started

Start logging this week

Write your topic list today, add a shared sheet with three columns, and ask everyone to log each chat for two weeks. Then paste the lines into the analyser and pick three topics to fix.

LiveChatBot’s free plan gives you the team inbox, the Android app, tickets with owners and a customer ticket page, saved replies with {name}, opening hours with an away form, and canned responses your team can share, for two websites with 100 visitors per website each month. No per-seat charges, so the whole support team can join. See pricing and live chat for customer service, or start at livechatbot.in.

Frequently asked questions

What is a good first-chat resolution rate for live chat for support?

It depends on the topic. Simple information topics like delivery status should be solved in the chat almost every time. Refunds or damaged items often need a follow-up, so a lower rate is normal there, as long as the ticket path is clear.

Do I need special software to log outcomes?

No. A shared sheet with three columns works. The free tool above reads the lines directly. Keep customer details out of the sheet.

How many chats do I need before the numbers mean anything?

Two weeks is usually enough for your main topics. Topics with only one or two chats can wait until you have more data.

What counts as a customer coming back?

The same customer returning about the same problem within about a week. A new, unrelated question is a new chat, not a return.

Does the analyser send my data anywhere?

No. It runs entirely in your browser and nothing is sent or saved. Use topic labels only.

Keep reading

Related guides

📘

Internet Chat Bot Basics for Non-Techies: Script Tags, Browsers, Servers and Cookies Explained

Read the guide →

Solve it once, not five times

Shared inbox, tickets with owners, saved replies and an away form, free on livechatbot.in.