Customer service KPIs for a small business: the 10 worth tracking, their formulas, and a one-page dashboard
Most KPI lists are written for call centers with forty agents and a workforce manager. This one is for a team of one to ten people answering WhatsApp, email, chat and the phone between other jobs. You get ten KPIs with formulas, the ones to ignore, a dashboard template with example numbers, and spreadsheet formulas to build it in an hour.
Quick answer
The customer service KPIs worth tracking in a small business are first response time, unanswered rate, after-hours share, resolution time, first contact resolution, CSAT, NPS, enquiry-to-booking rate, repeat-question share and cost per conversation. Measure them monthly from a sample of 30 conversations in a spreadsheet, and act on one at a time.
Which customer service KPIs matter for a small team?
The ones that change a decision. A small team should track speed (do we reply?), quality (did it help?), outcome (did it make money?) and cost (what did it take?). Ten KPIs cover those four questions without needing a help desk.
A KPI is a metric you have chosen to steer by and given a target. Total messages received is a metric; unanswered rate with a target of under 2% is a KPI. If no one would act when a number moves, it does not belong on the dashboard.
The 10 KPIs, with formulas
| # | KPI | Formula | What it tells you | Spreadsheet version |
|---|---|---|---|---|
| 1 | First response time | Median of (first reply time minus first message time), business hours | How fast you pick up | =MEDIAN(F2:F31) |
| 2 | Unanswered rate | Conversations with no reply within 24 hours ÷ new conversations × 100 | How many people you ignore, usually by accident | =COUNTIF(F2:F31,">1440")/COUNTA(A2:A31), logging never-answered chats as 9999 |
| 3 | After-hours share | Conversations started outside opening hours ÷ all conversations × 100 | Whether you need after-hours cover | =COUNTIF(G2:G31,"yes")/COUNTA(G2:G31) |
| 4 | Resolution time | Median of (time issue closed minus first message time) | How long problems really take | =MEDIAN(K2:K31) |
| 5 | First contact resolution | Issues solved in the first conversation, with no return within 7 days ÷ issues solved × 100 | Whether answers are complete | =COUNTIFS(I2:I31,"yes",J2:J31,"no")/COUNTA(I2:I31) |
| 6 | CSAT | Ratings of 4 or 5 on a 1 to 5 scale ÷ all ratings × 100 | Whether the help felt good | =COUNTIF(L2:L31,">=4")/COUNT(L2:L31) |
| 7 | NPS | % scoring 9 or 10 minus % scoring 0 to 6, on "how likely are you to recommend us?" | Loyalty, checked quarterly | Two COUNTIFs, subtracted |
| 8 | Enquiry-to-booking rate | Bookings or sales from conversations ÷ new enquiries × 100 | Whether service turns into revenue | =COUNTIF(M2:M31,"booked")/COUNTA(M2:M31) |
| 9 | Repeat-question share | Conversations about your top 10 questions ÷ all conversations × 100 | How much could be answered once, in writing | =COUNTIF(H2:H31,"top10")/COUNTA(H2:H31) |
| 10 | Cost per conversation | (Staff hours on messages × hourly cost + tool costs) ÷ conversations | What service actually costs | Monthly sum ÷ monthly count |
NPS is the Net Promoter Score defined by Bain: the percentage of promoters minus the percentage of detractors. First response time has its own explainer with a worked example, including why the median beats the average. The cost KPI is worked through in how to reduce customer service costs.
Which KPIs should a small business skip?
Skip anything built for large contact centers or that cannot change a decision at your size.
- Average handle time. In chat, one person juggles several conversations, so time per chat measures multitasking, not efficiency.
- Agent utilization and occupancy. Useful for scheduling 40 agents; meaningless when the person replying also runs the till.
- Total messages received. A busy month and a broken product both raise it. Use repeat-question share instead.
- Monthly NPS with few customers. With 15 responses, two unhappy people swing the score wildly. Run it quarterly and read the comments.
- Customer Effort Score, for now. The idea, popularized in a 2010 Harvard Business Review article by Dixon, Freeman and Toman, is sound: reducing effort matters more than delighting. But for a small team, repeat-question share and first contact resolution capture the same thing without another survey.
What does a simple customer service KPI dashboard look like?
One table, one page, reviewed monthly. Here is a template filled with hypothetical numbers for a small clinic.
| KPI | Target | Last month | This month | Status | Action |
|---|---|---|---|---|---|
| First response time (median, open hours) | Under 10 min | 22 min | 14 min | Improving | Add quick replies for pricing |
| Unanswered rate | Under 2% | 7% | 3% | Close | Check archived chats on Fridays |
| After-hours share | Watch only | 31% | 34% | High | Decide on after-hours cover |
| Resolution time (median) | Same day | 1.5 days | 1 day | Close | None this month |
| First contact resolution | Over 80% | 71% | 76% | Improving | Send aftercare sheet with every booking |
| CSAT | Over 90% | 88% | 91% | On target | None |
| NPS (quarterly) | Up on last quarter | 41 | Next quarter | n/a | None |
| Enquiry-to-booking rate | Over 35% | 29% | 33% | Improving | Offer two slot options in first reply |
| Repeat-question share | Under 40% | 58% | 55% | Off | Put prices and parking on website |
| Cost per conversation | Under $2 | $3.10 | $2.80 | Improving | Follows from repeat-question fix |
Notice the action column: every row has one, or a deliberate "none". A dashboard without actions is a report. Also notice the KPIs move together. Cutting repeat questions lowers cost, speeds first response and lifts first contact resolution in one go, which is why it is usually the best first project.
How to build the dashboard in a spreadsheet in an hour
You do not need help desk software for this. A spreadsheet and ten minutes a week of logging is enough to see trends.
- Create a log tab. Add columns A to M: date, channel, first message time, first reply time, time closed, response minutes, after hours (yes or no), topic (top10 or other), resolved in first conversation (yes or no), came back within 7 days (yes or no), resolution minutes, survey score, and outcome (booked, sold or none). The formulas in the table above use these column letters.
- Log a sample each month. Fill in 30 new conversations spread across the month, or every conversation if you have fewer. Ten minutes a week is enough.
- Add the formulas. Use MEDIAN on the response-minutes column, COUNTIF divided by COUNTA for every rate, and a simple sum for cost per conversation. Examples are in the table above.
- Ask one survey question. At the end of resolved conversations, ask customers to rate the help from 1 to 5, and record the answer in the score column.
- Build the one-page summary. On a second tab, show each KPI with this month, last month, your target and one action, laid out like the dashboard template.
- Review monthly and pick one KPI. Spend 20 minutes on the summary each month, choose the single KPI furthest from target, and agree one change to improve it before next month.
A sample of 30 conversations a month will not give you precise percentages, and that is fine: you are looking for direction and big gaps, not decimals. If a number swings more than you expect, log more conversations the following month before acting on it.
How do targets differ by business type?
The KPI list is the same; the targets and the KPI that matters most are not. These are our suggested starting points, not industry standards; adjust them after two months of your own data.
| Business type | KPI to watch first | Suggested starting target | Why |
|---|---|---|---|
| Clinic, salon, tuition center | Enquiry-to-booking rate | Up month on month | Every enquiry is a potential appointment; speed and slot offers drive it |
| Online shop | Repeat-question share | Under 40% | "Where is my order?" and sizing questions dominate and can be answered once |
| Trades and home services | Unanswered rate | Under 2% | Messages arrive while you are on a job and get forgotten |
| B2B services | Resolution time | Within agreed response terms | Fewer, larger clients judge you on how long issues take |
Where ReplyKit fits in your KPIs
ReplyKit is an AI agent for WhatsApp and website chat. It moves several of these numbers directly: it answers routine questions from your own content within seconds (first response time, repeat-question share), offers and books real open slots (enquiry-to-booking rate), and keeps every conversation in one shared inbox so fewer go unanswered. Leads are scored hot, warm or cold and can be exported as CSV, which gives you the enquiry and outcome columns for your spreadsheet.
It also keeps a list of questions it could not answer, so you can teach the answer once. That list is a ready-made source for your repeat-question work. What it does not do: it has no built-in KPI dashboard, no phone or email channel, and no integrations beyond CSV export, so the dashboard above still lives in your spreadsheet. For metrics specific to the AI itself, such as handover and containment rates, see AI agent metrics that matter.
WhatsApp connects through Quick Connect, a QR-code linked-device pairing that is unofficial and not sanctioned by Meta; use a dedicated number. Measure a month of your KPIs first, then start the 7-day free trial (no credit card, 300 AI replies) and compare. Plans are $29, $49 and $99 a month.
Frequently asked questions
What are customer service KPIs?
Customer service KPIs are the handful of metrics a business chooses to steer its service by, each with a target. Common examples are first response time, resolution time, first contact resolution, customer satisfaction (CSAT) and cost per conversation.
What are the most important customer service KPIs for a small business?
Start with first response time, unanswered rate and enquiry-to-booking rate, because they connect service directly to revenue. Add CSAT and repeat-question share once those three are measured every month.
How do you calculate CSAT?
Ask customers to rate the help they received, usually on a 1 to 5 scale, then divide the number of 4 and 5 ratings by the total number of ratings and multiply by 100. Ask right after the conversation is resolved so the answer reflects the service, not the product.
What is a good CSAT score?
There is no universal target, because scales, questions and industries differ. Compare yourself against last month and last quarter, and read the comments on low scores, which tell you more than the percentage.
What is the difference between a KPI and a metric?
A metric is anything you can measure. A KPI is a metric you have chosen to manage by, with a target and an owner who acts when it moves.
How often should a small business review customer service KPIs?
Monthly for most KPIs, with NPS quarterly. A 20-minute monthly review that picks one KPI to improve is more useful than a weekly review nobody acts on.
Move three KPIs with one change
ReplyKit answers WhatsApp and website chat from your content, books appointments and keeps every lead in one place. 7-day free trial, no credit card.
Sources and further reading
- Bain and Company, Measuring your Net Promoter Score · NPS formula: percentage of promoters (9 or 10) minus detractors (0 to 6)
- Harvard Business Review, Stop Trying to Delight Your Customers (Dixon, Freeman, Toman, 2010) · the case for reducing customer effort rather than exceeding expectations
- HubSpot, Customer loyalty trends · 60% of customers define an immediate response as 10 minutes or less, behind the first response target
- SuperOffice, Customer service benchmark report · only 20% of companies answered questions in full on the first reply, relevant to first contact resolution
ReplyKit is an independent product and is not affiliated with or endorsed by Meta or WhatsApp. Sources were read in October 2026. Example figures in this article are hypothetical. This article is general guidance.