---
title: "Customer Service Targets: Metrics for Online Stores"
canonical: "https://www.thrivedesk.com/navigating-customer-service-targets/"
author: ThriveDesk
description: "Define customer service targets for your store with clear clocks, denominators, and review periods. Compare speed, resolution, and satisfaction fairly."
categories: [blog, customer-support]
---

# Customer Service Targets: A Measurement Guide for Online Stores

Define customer service targets for your store with clear clocks, denominators, and review periods. Compare speed, resolution, and satisfaction fairly.

[Zannatun Nime](https://www.thrivedesk.com/author/nime/) 9 min read Updated October 18, 2026

![Support agent wearing a headset and holding an arrow](https://www.thrivedesk.com/_astro/handsome-businessman-suit-headphones-with-microphone-holding-white-arrow-looking-camera-presenting.BWwvjK3z_yI7Ia.webp)

TL;DR

-   A target needs a threshold, a period, and a counting rule. Record the baseline, eligible cases, data source, and owner before comparing results.
-   Separate business hours from elapsed time and a useful first reply from an automatic acknowledgment. Keep email and live chat results distinct.
-   For percentages, show the numerator and denominator. A satisfaction score describes survey respondents, not every customer who contacted the store.
-   Check unresolved work and the slowest cases alongside averages. Closing a request does not establish that the customer received a complete remedy.
-   All numbers in this guide are hypothetical examples. Choose your own thresholds from customer commitments, staffing, and a comparable baseline, then preserve the original target when you revise it.

Customer service targets are numerical thresholds with a review period and a measurement rule. For an online store, that might mean a first-reply time for email during staffed hours, or a satisfaction threshold among valid survey responses. Without a shared rule, two reports can give different answers for the same work.

This guide explains clocks, eligible cases, percentages, and fair comparisons. All numerical scenarios below are hypothetical examples, not industry benchmarks or ThriveDesk results. Start from your own baseline and customer commitments. If you need to choose an outcome, read the [goal examples](https://www.thrivedesk.com/customer-service-goals/). For ownership, resources, and the planning steps, use [how to set customer service goals](https://www.thrivedesk.com/how-to-set-customer-service-goals/).

## What makes customer service targets usable?

A usable target tells two people how to produce the same result from the same cases. “Reply faster” leaves too many choices open. “Median first human reply below four business hours for new support emails next month” gives you a starting definition, but still needs the hours, exclusions, and data source.

Record a baseline using that same definition. Add the period covered, the number of cases, and relevant conditions such as a delivery disruption. Then record the target period, the owner, and the decision date.

Keep a quality check beside the main metric. For a response-time target, review whether the reply answered the question or explained a useful next step. A timer cannot tell you whether an agent sent accurate return instructions.

Agree on these definitions before the period starts. If a reporting rule changes, record when it changed and whether earlier results can be recalculated. Otherwise the team may mistake a new definition for better service.

## Define the cases and denominators first

Decide which requests belong in the report. An email about an existing order, a pre-purchase chat, and a supplier message should not enter the same customer-support measure by accident.

For rates, record both the numerator and denominator. Suppose 12 of 100 eligible closed cases reopen within seven days. The example reopen rate is 12%. If you report only the 12 reopened cases, you cannot compare that count with a period that had twice as many closed cases.

Use one unit consistently. A customer can send several messages in one conversation or raise several issues about one order. Choose whether the measure counts messages, conversations, issues, or orders. Explain how you identify the same issue across channels.

Record exclusions such as spam, test conversations, and unrelated replies. Do not add exclusions after seeing which cases missed the target. For low volume, show the counts and inspect individual cases; a large-looking percentage change may reflect only one customer.

## Choose the right clock for response time

First response time measures the interval between a customer’s initial request and the reply defined by your rule. State whether an automatic acknowledgment counts. For a human-response target, exclude that acknowledgment and measure the first human reply.

A business-hours clock counts time only during your defined support schedule. An elapsed-time clock continues overnight and through closed days. Record the timezone, working days, holiday exceptions, and treatment of schedule changes.

For example, assume support works Monday through Friday, 9 a.m. to 5 p.m., with no holiday. A request arriving Friday at 4 p.m. and answered Monday at 10 a.m. waits two business hours but 66 elapsed hours. Both calculations describe the same case. They answer different questions.

Keep channels separate. Email and live chat have different customer expectations and staffing patterns. If your [live chat](https://www.thrivedesk.com/live-chat/) is unavailable outside staffed hours, explain what customers can expect when they leave a message.

## Measure resolution without hiding repeat work

Decide what resolution means for the issue. An agent may complete the support action before the customer receives a replacement or a refund reaches their account. Record which event stops the timer.

Specify whether customer-waiting or internal-waiting time pauses the clock. An elapsed resolution measure includes those delays; an active-work measure may exclude some of them. Neither should silently replace the other in a comparison.

A first-contact resolution rate also needs a definition. Decide which eligible issues count as resolved after the first exchange and how long you watch for another contact about the same issue. Waiting until that window ends avoids counting a recent case as successful too early.

Read reopened cases alongside the rate. Some reopen because the reply was incomplete; others contain a separate question or a customer saying thank you. Apply the documented rule consistently. The [customer service experience guide](https://www.thrivedesk.com/customer-service-experience/) can help with the qualitative review that the timer cannot supply.

## Calculate satisfaction and explain the sample

For customer satisfaction, define the positive responses on your survey scale. Divide the positive count by all valid responses, then multiply by 100. Keep the question, scale, and timing unchanged when comparing periods.

If 80 of 100 valid responses are positive, the example CSAT is 80%. A later result of 85 positive responses out of 100 is 85%. That is an increase of five percentage points, or a relative increase of 6.25%. Say which change you mean.

Report response volume and, where available, how many customers received the survey. The score describes respondents. It does not establish that 85% of all customers are satisfied, nor that support caused a change in store revenue.

Read comments and [customer feedback](https://www.thrivedesk.com/mastering-customer-feedback/) beside the score. A delivery delay may affect a rating even when the agent handled it well. Keep that context visible instead of treating every negative response as an agent error. Net Promoter Score uses a different question and calculation; it is not interchangeable with CSAT.

## Keep a target register with clear rules

Use a shared record so reviewers can find the definitions. The table below illustrates the shape of that record. Its numbers are hypothetical; they are not recommended thresholds. A real register should also include the owner, data source, dates, and sample size.

| Metric | Example baseline | Illustrative target | Rule to record |
| --- | --- | --- | --- |
| First human reply | Median six business hours | Below four next month | New email cases, published hours, exclude spam and auto-acknowledgments |
| Customer satisfaction | 80 positive responses out of 100 | At least 85% next quarter | Same survey scale, valid responses, report both counts |
| First-contact resolution | 60 eligible issues out of 100 | At least 65% next month | Same issue, agreed exchange definition, seven-day follow-up window |
| Reopened cases | 12 out of 100 eligible closed cases | Below 10% next month | Same issue within seven days of closure |
| Older open requests | 18 open cases over three days old | Fewer than 12 at the next review | Same snapshot time, elapsed age, all pending states shown |
| Reply quality | 16 reviewed replies meet the rubric out of 20 | At least 18 out of 20 next review | Consistent sampling, accuracy and next-action rubric |

A customer-facing service promise needs a clear policy too. Check [service policies](https://www.thrivedesk.com/customer-service-policies-that-drive-loyalty/) before treating an internal trial target as a commitment to customers. Tell the team whether the number is a trial threshold, an internal standard, or a published promise.

## Check slow cases and unfinished work

An average can hide long waits. The mean adds the response durations and divides by the case count. The median is the middle value after sorting them. State which one the report uses and avoid switching between them to present a better result.

For a larger sample, a percentile can describe the slower part of the distribution. Record the calculation method and review the cases above that point. With few cases, show individual durations rather than imply precision from a percentile.

Inspect open requests separately from completed ones. A resolution report limited to closed cases omits customers who are still waiting. Review the age and next action of pending work at a consistent snapshot time.

An agent may be waiting on a carrier or finance team. Show that dependency instead of removing the case from view. Decide who owns the follow-up. Our [shared-inbox guide](https://www.thrivedesk.com/shared-inbox/) explains the conversation workspace; your reporting rule still needs to cover cases that remain open there.

## Compare periods with similar conditions

Compare the same channel, case types, clock, and survey definition. Add volume and staffing context. A promotion that doubles delivery questions is different from an ordinary week with the same response time.

Where possible, separate peak-season results from ordinary demand. Record major incidents, changes in support hours, and the arrival of temporary staff. These details help explain a result without excusing a missed customer commitment.

Avoid ranking agents on raw closure counts when their work differs. Someone handling damaged-order investigations may close fewer cases than a colleague answering routine product questions. Review case mix and answer quality before making a performance decision.

Choose the review period to fit the work. A weekly check can find aging cases, while a later review evaluates a trial after its observation windows finish. If your main problem is capacity, the [objectives guide](https://www.thrivedesk.com/customer-service-objectives/) can help frame the staffing decision behind the number.

## Review misses and automation honestly

At the decision date, record the original target and the actual result. Read a sample of misses to distinguish a broken workflow, missing information, insufficient coverage, and an unrealistic threshold. Each needs a different response.

Keep the original record when you revise a target. Explain the changed assumption and the date the new rule takes effect. If you recalculate earlier periods, label the revised series so readers can see what changed.

Treat automation as a separate part of the measurement. An automated acknowledgment can confirm receipt without answering the question. A bot answer that later reaches an agent is not necessarily a completed resolution. Define the follow-up window and how handoffs count before claiming success.

Use [automation](https://www.thrivedesk.com/features/automation/) for a specific workflow and review its exceptions. If the main measure improves while complaints, reopened cases, or reply-quality failures increase, investigate before expanding the change. Record that conflict in the review and check the cases before deciding what to change.

## Review the conversations behind the numbers with ThriveDesk

ThriveDesk’s [analytics features](https://www.thrivedesk.com/features/analytics/) give support teams reports to use in service reviews. Check the definitions and filters available in your account before comparing them with your target register. This guide does not imply that every custom denominator, percentile, or follow-up window is a built-in report.

Use the shared inbox to inspect the conversations behind a result. Assignment and internal notes can help reviewers understand who owns the next action. For a WooCommerce store, [customer and order context](https://www.thrivedesk.com/woocommerce/) can also help explain what information the agent had when they replied.

Keep your target register in the team’s planning record. Where a required measure is unavailable in the product, use a documented case review or a suitable export rather than substitute a different metric without saying so. Check that you can obtain the data before committing the team to a target.

Basic is $29 per month on monthly billing for 2 teammates and 1 inbox. Confirm the reporting features and limits on the [pricing page](https://www.thrivedesk.com/pricing/) for the plan you need. The linked cost calculator can help with the teammate comparison.

Choose a report because it answers the question in your register. Then read the affected cases to check whether the numerical change also reflects useful help for customers.

## Frequently asked questions

### What should customer service targets include?

Record the metric, baseline, threshold, review period, eligible cases, and counting rule. Add the clock, data source, owner, and a quality check where relevant. For an online store, separate delivery questions from complex refund investigations when their work differs. A number without those definitions is difficult to compare or act on.

### What is a good customer service response-time target?

Choose a threshold from your published customer commitments, channel, staffing, and current baseline. This guide does not prescribe a universal benchmark. Email during staffed hours and live chat need different expectations. Define whether the timer uses business or elapsed hours and whether an automatic acknowledgment counts before comparing results.

### Should you use average or median response time?

They answer different questions. The mean reflects the total response time divided by the number of cases; a few long waits can move it substantially. The median describes the middle observation. Review the slower cases as well, because neither measure alone shows whether some customers wait far beyond your service commitment.

### How do you calculate a customer satisfaction target?

Define which survey responses count as positive, then divide positive responses by all valid responses and multiply by one hundred. Keep the survey scale and timing consistent. Report both counts with the percentage. The result describes respondents; it should not be presented as the satisfaction level of every customer or every resolved case.

### Can automated replies count toward a service target?

Only if the metric explicitly measures automated acknowledgments. For a human-response target, keep them separate. A bot answer also needs an agreed success definition and follow-up window before you count it as a resolution. Review requests that return to an agent so automation does not hide work that remains unresolved.

### When should you revise a target?

Review it at the agreed decision date, or earlier if a major change makes the original assumptions invalid. Record changes in staffing, demand, policy, or reporting definitions. Preserve the original threshold and result alongside the revised plan. Changing a target after seeing a miss should not erase that miss from the record.

Helpful free tools

 [![Helpdesk cost calculator preview](https://www.thrivedesk.com/_astro/og-helpdesk-cost-cal.Df-sCCCd_Z2q8PXW.webp) Helpdesk Cost Calculator Stop guessing. Compare real monthly costs for ThriveDesk, Freshdesk, Gorgias, Intercom, and more based on your team size and feature needs. Try it free](https://www.thrivedesk.com/tools/helpdesk-cost-calculator/)

Share this article

Written by

[Zannatun Nime](https://www.thrivedesk.com/author/nime/) ThriveDesk

Covers customer service best practices, support workflows, and team operations.