Key takeaways
- Have outsiders call from their own mobile numbers; a familiar voice or a blocked caller ID spoils the test.
- Always test the after-hours leak and the missed-call callback, the two paths no office report shows.
- Score outcomes on a 100-point sheet, including whether the promised callback or confirmation actually arrived.
- Use your phone system's own recordings with its announced notice; all-party consent states such as California penalize secret recordings[2].
- Announce the program, fix the system before coaching the person, and share individual results privately.
What a test call shows that reports cannot
Your phone reports say how many calls rang and how many were answered. They do not say what a caller with a leaking water heater hears at 4:55 on a Friday, whether the after-hours forward really reaches the on-call phone, or whether the callback promised to a missed caller ever happens. A mystery shop answers those questions the only reliable way: someone rings your business the way a customer would and writes down what happened, start to finish.
It is a different tool from the ones you may already use. A one-week call audit counts and classifies the real calls that came in. A CSR call scorecard grades how one person handled one recorded call. A mystery shop tests the whole path, including the parts no recording captures: the route the call took, the callback, the confirmation, and whether the details the caller gave made it onto the job.
Customers already run their own version of this test, and so does Google. Local Services Ads rank partly on responsiveness to customer inquiries, and Google says missed calls may count against it[3]. Callers are just as unforgiving about callbacks: only 19% of US adults generally answer calls from unknown numbers, 67% let them ring and check voicemail, and the rest mostly ignore the voicemail too[1]. A test that stops at “was the phone answered?” misses half of what decides whether the job is booked.
- Generally answer 19%
- Don't answer, check voicemail 67%
- Don't answer, ignore voicemail 14%
Source: Pew Research Center, 2020[1]
Show the data
| Item | Value |
|---|---|
| Generally answer | 19% |
| Don't answer, check voicemail | 67% |
| Don't answer, ignore voicemail | 14% |
Who should place the calls
Not you. CSRs know the owner's voice within two words, and a manager calling from the back office tests nothing. Use people the team will not recognize, calling from numbers the team has never seen.
| Caller | Strengths | Watch out for |
|---|---|---|
| Friends or relatives of the owner | Free, trusted, real mobile numbers | They go easy in the write-up; brief them on the sheet first |
| A contractor in another trade or town (a call swap) | Knows what good sounds like; you return the favor | Must not compete with you for the same customers |
| A paid mystery shopping service | Consistent scoring, written reports, many calls | Generic scripts; insist they use your scenarios |
| A manager from another branch | Knows your rules and your service area | May be recognized; best kept for after-hours paths |
- Each caller uses their own mobile number, never a blocked caller ID. Real customers rarely call as “No Caller ID”, and some systems route blocked calls differently.
- Give each caller a realistic address inside your service area that is not an existing customer, and one address just outside it for the out-of-area scenario.
- Ask callers to cancel any booking within the hour, by phone. It frees the slot and tests your cancellation path at the same time.
- Agree a code word the caller can give if a real emergency is about to be dispatched, so a technician never drives to a fake leak.
Eight scenarios worth testing
- Routine drain booking, mid-morning. Tests the basics: greeting, details, two time options and the visit fee explained the way your policy says.
- Price shopper. “How much to replace a water heater?” Tests whether the answer follows your price question policy or turns into a guess.
- After-hours leak, 10 pm. Water coming through a kitchen ceiling. Tests the forward, the triage questions and how long the on-call technician takes to ring back.
- Lunch-hour call, 12:15. Tests whatever covers the phone while the office person is on a break.
- Five to close on a Friday. Tests whether late calls get the same effort, and whether anyone mentions weekend availability.
- Out-of-area caller. Tests whether the team says no clearly and quickly, and offers a referral if that is your practice.
- Hang up and wait. The caller lets it ring through to voicemail, leaves a short message, then records when the callback comes, from which number, and what was said.
- Existing customer with a problem. “Your tech was here Tuesday and it's leaking again.” Tests recall handling and whether the complaint reaches a manager.
Run scenarios 3 and 7 every round. They are the paths where most offices find surprises, and nobody in the office can see them. If you advertise Spanish-language service, add a ninth call in Spanish and treat it the same way.
How many calls, and when
Eight to twelve calls a quarter is enough for a small shop: one of each scenario, plus repeats of the after-hours and callback tests. Spread them across weekdays, evenings and a weekend, because that is where coverage thins out. On an average weekend day or holiday in 2025, 30.3% of employed people worked, against 80.5% on a weekday[4], and customer service roles often need cover during evenings, weekends and holidays[5]. Test the hours your staff are least likely to be at a desk.
Pause the program during a freeze, a storm or any day the board is already overloaded. A test call on the busiest morning of the year tells you nothing you didn't know, and it takes a line a real customer needed. Run an extra round whenever you change phone provider, ring groups, the after-hours forward or your answering setup, since that is when paths quietly break.
Recording the test calls lawfully
You will want to hear the calls, not just read notes. Federal law allows a call to be recorded when the person recording is a party to it or one party consents[6], but several states require every party's consent. California's Penal Code 632 punishes recording a confidential communication without the consent of all parties with a fine of up to $2,500 per violation[2]. In Washington, consent counts as obtained when one party announces the recording to the others in a reasonably effective way, and the announcement must be recorded too[7].
The simplest route is your own phone system. If it already records calls with an announced notice in the greeting, the test calls are captured like any other call, and you can pull them by time and number. If it does not record, have the test caller fill in the sheet during and right after the call instead of recording on a personal phone. The rules for your state are in call recording consent rules for plumbing companies.
The scoring sheet
Score outcomes, not style. The test caller cannot judge tone the way a trained manager can, but they can say exactly whether something happened.
| Check | Points | A pass looks like |
|---|---|---|
| Answered within four rings | 15 | A person or your answering setup picks up; no voicemail, no full mailbox |
| Business name in the greeting | 5 | The caller knows at once they reached the right company |
| Urgency checked early | 15 | Leak, sewage, gas or no water asked about before booking details |
| Name, address, ZIP and callback number | 15 | All four taken and read back |
| Problem in the caller's words | 10 | The job note matches what the test caller said |
| A next step with a time | 15 | Two windows offered, or a callback time from the on-call tech |
| Fee policy stated as written | 10 | Visit or diagnostic fee explained the way your policy says |
| Promise kept | 15 | The confirmation or callback arrives when and how it was promised |
The last row is scored a day later, and it is the one most offices fail. Ask the caller to note when the confirmation arrived, from which number, and whether the callback came at all. Then open the job or message the call created and compare it with the caller's notes: a wrong ZIP, a missing gate code, a ceiling leak written down as “drain issue”. Those small errors are what cost a technician a wasted hour on the day.
Example: a first round of ten calls
Test calls placed
Answered live
Urgency checked before booking details
Job notes matched what the caller said
Promises kept on time
Example figures for a fictional six-truck shop, not survey data.
Example: a fictional six-truck shop runs its first round in March. Eight of ten calls are answered live; the two that are not both land at lunch, and one of them reaches a voicemail box that is full. Six callers are asked about urgency before the office starts on names and addresses. Five job notes match what the caller said. Only three of the promised callbacks and confirmations arrive on time: the after-hours leak caller waits 52 minutes for the on-call technician, whose phone battery died, and two booking confirmations never send because a software update switched them off.
None of that showed in the monthly phone report, which recorded a healthy answer rate. The fixes in this example are dull and effective: a lunch overflow, a weekly mailbox check, a charger in every truck, and a test confirmation after every software update.
Sharing results without an ambush
A mystery shop goes wrong when staff first hear about it in a meeting where their worst call is played to the room. Set it up so the team expects it and gains from it.
Announce the program first
Tell the team test calls will happen every quarter, who scores them and what the sheet checks. Do not say when.
Fix the system before coaching the person
Most failures are routing, settings or missing information. Fix those first, and say that is what you found.
Share privately, then as a team
Each person hears their own calls one to one. The team meeting sees totals and patterns, never a ranking.
Lead with what went right
Play a good call for every problem call, and name the people who handled them well.
Change one or two things
Pick the two lowest checks, fix them, and retest those paths within a month.
Keep the first round out of discipline
The first round is a baseline. Using it to punish guarantees the next round gets gamed.
Hold your outside help to the same sheet. An answering service, an on-call rota or an SI Employee on your overflow should face the same scenarios, including the after-hours leak and the hang-up. With an SI Employee, every forwarded call already has a recording, transcript and summary in your account, so matching the test caller's notes against what was captured takes minutes (see what it does). The one-van plumber example models how an owner-operator would check what callers hear while he is under a sink.
Turning scores into fixes
- Answer checks failing at particular hours: fix routing before anything else, starting with the overflow plan.
- Urgency checks failing: put the triage questions on a card beside every phone and on the on-call tech's lock screen.
- Details wrong on the job: add a read-back step to the script and score it next round.
- Promises failing: trace the broken handoff, whether that is on-call phones, confirmation settings or nobody owning callbacks.
- The same failure two rounds running: the fix did not stick; ask the people answering why.
Keep every round's scores on one sheet. A year of rounds says more than any single one, and it shows whether a change stuck or faded after a month. Read it alongside your call answer rate and the same weakness usually shows up in both.
Frequently asked questions
Is it legal to mystery shop my own plumbing business?
Calling your own business is fine. Recording needs care: federal law allows recording with one party's consent[6], but states such as California require all parties' consent[2]. Use your phone system's recordings with the notice your greeting already gives, or have test callers take notes instead.
Should I tell my staff about mystery shopping?
Tell them the program exists, what the sheet checks and who scores it, but not the dates. Staff who know about it accept the results; staff who are surprised by it argue with them.
How many test calls should a small shop make?
Eight to twelve a quarter covers the main scenarios, plus a round after any change to your phone system, ring groups, after-hours forward or answering setup. Skip storm days and freezes.
Can I mystery shop my answering service the same way?
Yes, and you should. Use the same scenarios and compare the message they send you with the test caller's notes. The answering service checklist lists what to require before and after you sign.
What if a test call takes a real appointment slot?
Have the caller cancel by phone within the hour. The slot is freed and your cancellation handling gets tested in the same round.
References
Pew Research Center (2020). Most Americans don't answer cellphone calls from unknown numbers (opens in a new tab) ↑
public.law (California Penal Code) (2026). California Penal Code section 632 (opens in a new tab) ↑
Google (2026). About ad rankings - Local Services Help (opens in a new tab) ↑
U.S. Bureau of Labor Statistics (2025). American Time Use Survey, Table 4: Employed persons working and time spent working on days worked, 2025 (opens in a new tab) ↑
U.S. Bureau of Labor Statistics (2025). Occupational Outlook Handbook: Customer Service Representatives (opens in a new tab) ↑
Legal Information Institute, Cornell Law School (U.S. Code) (2026). 18 U.S. Code § 2511 - Interception and disclosure of wire, oral, or electronic communications prohibited (opens in a new tab) ↑
Washington State Legislature (2026). RCW 9.73.030 Intercepting, recording, or divulging private communication (opens in a new tab) ↑