Walk into any "customer experience for SMB" article and you'll get the same lecture: track NPS, CSAT, CES, churn, retention, lifetime value, share of wallet, advocacy score, and about six more acronyms. Every one of them, the writer insists, is critical.
For a service business — HVAC, plumbing, cleaning, electrical, restoration — most of those metrics are noise. Tracking them all dilutes attention from the two or three that actually move the business. This is the no-fluff version: what to measure, what to ignore, and how to use the numbers.
The three metrics that matter
1. NPS (Net Promoter Score)
What it is:"On a scale of 0-10, how likely are you to recommend us to a friend?" Promoters score 9-10, passives 7-8, detractors 0-6. NPS = % promoters − % detractors.
Why it matters for service businesses:Service work is overwhelmingly referral-driven. The promoter/detractor split predicts how much referral revenue you'll see next quarter better than any other single number. The detractor count predicts how many 1-star reviews you'll get in the same window if you don't intervene.
How often to look: Daily during peak season, weekly otherwise. Monthly is too coarse — a bad week in July gets buried in a monthly average.
2. Detractor recovery rate
What it is: Of all detractor responses (NPS 0-6), what percentage did you contact + remediate before they posted a public review?
Why it matters:NPS tells you how many unhappy customers exist. Recovery rate tells you whether you're actually doing anything about them. A business with NPS = 20 and a 90% recovery rate is in a better operational state than a business with NPS = 50 and a 10% recovery rate, because the bad reviews from the first business never get published. We covered the recovery workflow in detail in how to recover unhappy customers before they leave a bad review.
How often to look: Weekly. Should be above 75%. If it's below 50%, your alerting is broken and you're losing customers invisibly.
3. Per-tech NPS variance
What it is: The spread between your highest-NPS and lowest-NPS tech, measured over the last 30 days.
Why it matters:The single biggest lever on company-wide NPS for a service business is consistency across techs. A 40-point variance between your best and worst tech means your customer experience depends on a coin flip. A 15-point variance means you've got a culture, not a roster.
Per-tech NPS also surfaces hiring problems early. A new tech whose first month NPS is 20+ points below the team average usually doesn't recover; better to know in week three than month six.
How often to look:Monthly. Don't make per-tech decisions on a single week — sample size matters more here than recency.
What to ignore
CSAT (Customer Satisfaction Score)
"On a scale of 1-5, how satisfied were you?" CSAT is the right metric for transactional B2C — Amazon order confirmation, Uber ride end. For service businesses it's usually redundant with NPS, asks the customer to do double work, and produces a number that's harder to compare across time because the rating scale changes its meaning every few quarters as your customer base shifts.
Just track NPS.
CES (Customer Effort Score)
"How easy was it to get the help you needed?" CES was invented for customer-support contexts. Service businesses don't have a customer-support layer in the same sense — you're the work itself. CES doesn't map cleanly to a tradesperson walking through a door.
Skip.
Industry-benchmark NPS
"Home services industry average NPS is 36." That number combines you with pool cleaners and dryer-vent inspections. Comparing to it tells you nothing about whether you're improving against your own past performance — the only comparison that matters.
Compare to your own last quarter. That's the benchmark.
Anything aggregated quarterly
Service-business cadence is too fast for quarterly readouts. By the time the Q2 summary comes out, you're halfway through Q3 and the seasonality has flipped. Weekly is the operational frequency; monthly is the strategic one. Quarterly is for the deck you show your accountant.
Vanity averages
"Our average review is 4.7 stars." Means almost nothing. The distribution matters more than the average — 20% 5-stars + 80% 4-stars (average 4.2) is healthier than 70% 5-stars + 30% 1-stars (average 3.8) because the second distribution means a third of your customers actively warn people away from you. Look at the histogram, not the mean.
What to also track (but downstream of the three above)
These don't belong in your weekly NPS check, but they matter as monthly review items:
- Detractor verbatims clustered by theme. What are detractors actually complaining about? Communication? Pricing? Punctuality? AI-summarized topic clustering across the last month's detractor responses is what you act on in your operations meeting.
- Per-job-type NPS. Different services deserve different baselines. A no-heat call should hit 70+; a maintenance tune-up should hit 50+. Same number is bad news for different reasons.
- Google / Yelp review velocity.The rate at which new public reviews appear (positive AND negative), per week. Sudden velocity changes — either direction — signal something the survey hasn't caught yet.
- Time-to-first-detractor-contact. Median minutes from detractor response → first human outreach. Under 60 minutes = excellent. Over 240 = your alerting is broken.
How to instrument this
You need three things working together:
- A survey that fires on every job-close (covered in how to send post-job customer surveys).
- A detractor alert that pages a human within minutes of a low-NPS response.
- A weekly dashboard with the three metrics above on one screen — NPS trend, recovery rate, per-tech variance.
You can hand-roll this with a survey tool + Slack + a Google Sheet, and a lot of service businesses do. The once-you-pass-50-jobs-a-week problem is that maintaining the plumbing eats a person's time. Tools like Canopy fold the three pieces into one workflow so you don't.
30-day trial, one click. Trade-specific overviews at /for/hvac, /for/plumbing, /for/cleaning, /for/electrical, and /for/restoration.