Key takeaways
- Twelve columns are enough: when, who, how they found you, why they called, who answered, what happened, and every callback attempt.
- Let the phone system fill the columns it knows; type only the outcome columns by hand.
- Never put card numbers, security codes, lockbox codes or gate codes in a log; PCI rules bar storing card security codes after authorization[2].
- One week of rows, sorted by hour and outcome, shows where calls leak better than any monthly total.
What the log is for
A call log earns its keep on Monday morning, when the owner asks how many people tried to book last week and nobody can say. The phone system can count rings and minutes. It can't tell you which calls were new jobs, which ones the office turned away, or which callbacks never connected. A log is the one place where every inbound call gets a reason and an outcome, written while the call is fresh.
The callback columns matter more than most offices think. In Pew's survey, only 19% of US adults said they generally answer calls from unknown numbers; 67% said they don't answer but check voicemail, and 14% said they ignore voicemail too[1]. A note that says “called back, no answer” hides the difference between a lead that heard your message and one that never will. The log has to record how each attempt was made.
- Wages and salaries $25.22 69%
- Other benefits (insurance, retirement, other) $6.16 17%
- Paid leave $2.77 8%
- Legally required benefits $2.58 7%
Source: U.S. Bureau of Labor Statistics, 2026[3]
Show the data
| Component | Cost per hour |
|---|---|
| Wages and salaries | $25.22 |
| Other benefits (insurance, retirement, other) | $6.16 |
| Paid leave | $2.77 |
| Legally required benefits | $2.58 |
Most shops that need a log are small. In 2023, 64,871 of 111,207 plumbing, heating and air-conditioning establishments had fewer than five employees[4], which usually means no dedicated CSR and an owner answering between jobs. The template below is built for that office: one row per call, mostly drop-downs, under a minute to complete.
The columns that matter
| Column | What goes in it | Format |
|---|---|---|
| Date and time | When the call started | Automatic |
| Caller number | The number that called | Automatic |
| Name | Caller's name as given | Text |
| New or existing | Whether they're already a customer | New / Existing |
| Source | How they found you, in their words | Drop-down: Google, ad, referral, truck, repeat, other |
| Reason | Why they called | Drop-down: emergency, repair, drain, water heater, estimate, open job, billing, not a job |
| ZIP code | Where the job is | Five digits |
| Answered by | Who took it | Drop-down: office, owner, tech, answering service, voicemail, missed |
| Outcome | What happened | Drop-down: booked, quoted, message taken, declined, lost, no job |
| Lost or declined reason | Why it didn't book | Drop-down: price, timing, out of area, out of scope, no callback |
| Callback attempts | Time and method of each try | Text: “2:10 voicemail left; 4:30 text sent” |
| Job number | Link to the booked job, if any | From your scheduling software |
Resist adding columns. Every extra field is a few more seconds per call, and a log that takes too long gets filled in at the end of the day from memory, which defeats the point. If a column hasn't changed a decision in a month, delete it. Reason and outcome codes should stay short enough to apply in seconds; the fuller version of each customer's details belongs in the call intake, not the log.
Example row: Tue 4:52 pm · (555) 014-2290 · Dana R. · New · Google · Water heater · 44120 · Missed · Booked · (blank) · 5:06 pm voicemail left; 5:31 pm reached and booked · Job 18842. One glance shows a missed call that still turned into a job, and how.
“Missed” and “voicemail” are different outcomes. A missed call that left no message still has a number, and those are the rows a daily callback pass works through; how fast to call back a plumbing lead sets the target. Google lists responsiveness as a Local Services Ads ranking factor and says missed calls may count against you[5], so the missed rows are worth reading even if you don't advertise there now.
Coding the awkward rows
A few kinds of call cause most of the arguments about what to write. Settle them once, write the rules in a note at the top of the sheet, and the log stays comparable from week to week.
- Same caller, three tries before anyone answers: three rows, because three calls rang. Mark the first two “missed” and the third with its real outcome, so the missed count stays honest and the booking is counted once.
- An existing customer asking about an open job: reason “open job”, outcome “no job”. These calls cost time but aren't lost sales, and lumping them with new inquiries makes the booking rate look worse than it is.
- A transferred call: one row, with “answered by” set to whoever finally handled it.
- A caller who gets a price and says they'll think about it: outcome “quoted”, not “lost”. It becomes lost only when follow-up ends without a booking.
- A wrong number or a sales pitch: “not a job”, no further columns.
Repeat callers are worth a second look on their own. If the same number appears three times in a week with no booking, someone is chasing you, and the log is how you notice before they give up. Recognizing repeat customers when they call covers what the person answering should see.
What never goes in the log
A shared call log gets copied, exported, emailed to the owner's phone and left open on a front-desk screen. Treat it as a document a stranger might read. Card numbers and security codes never go in it: PCI rules say card verification codes must not be stored after authorization, and encrypting them doesn't change that[2]. Lockbox, gate and alarm codes stay in the job record for the visit only. Health details a caller volunteers (“my husband just had surgery, we need the water back”) stay out entirely; “priority: medical” is enough.
The FTC's guide for businesses boils it down to a principle the log should follow: keep only what you need for your business, and don't keep customer credit card information without a business need for it[6]. If something goes wrong, every state, plus D.C., Guam, Puerto Rico and the Virgin Islands, has a law requiring businesses to notify people when their personal information is breached[7]. A spreadsheet full of names, phone numbers and gate codes is exactly the kind of file those laws were written about. Securing customer data covers the wider setup.
Sheet, phone system or software
There is no single right tool; there is a right split. Let machines record what machines know (time, number, duration, missed or answered) and have people record what only a person knows (reason, outcome, lost reason).
| Option | Good for | Weak spot |
|---|---|---|
| Shared spreadsheet | A one- or two-person office starting today at no cost | Every column typed by hand; easy to skip on busy days |
| Phone system call report | Accurate times, numbers and missed calls | Knows nothing about reasons, outcomes or jobs |
| Call tracking numbers | Filling the source column automatically | Still needs a person to code the outcome |
| Field service software | Booked calls already tied to a job | Calls that didn't book often leave no record |
| Answering tool with call summaries | Calls it answers arrive with a recording, transcript and summary | Calls your own team takes still need logging |
Setting up the sheet takes about an hour. Put the log on one tab and the drop-down lists on a second, protected tab so nobody edits a code mid-week. Freeze the header row, format the date column as date and time, and use data validation so each coded column only accepts values from its list. Add a third tab for the weekly summary, built from counts of the coded columns, so the review takes minutes instead of an afternoon. Share edit access only with the people who answer phones.
If you run call tracking numbers, the source column fills itself. If an SI Employee answers your overflow, each call it takes already arrives in your account with a recording, transcript and summary, and signed webhooks can send each call to a sheet through tools such as Zapier, Make or n8n (integrations); the hand-typed log then only has to cover the calls your own team answered.
Hand typing has a real cost, which is why the split matters. Private employers paid an average $36.73 an hour in total compensation for office and administrative support workers in June 2026, $25.22 of it wages[3]. The calculator below puts a number on typing every row by hand; change the inputs to match your office.
| Input | Example |
|---|---|
| Inbound calls a day | 35 |
| Minutes to type each row | 1.5 |
| Rows typed fully by hand (%) | 100% |
| Hourly cost of office time | $36.73 |
| Calls a month | 760 |
| Staff hours on these calls | 19 |
| Hours handed off | 19 |
| Monthly value of that staff time | $697 |
How it is worked out: Calls a day × 21.7 working days × minutes per call ÷ 60 × share handled = hours; hours × hourly cost.
Assumptions
- Hourly cost of $36.73, the average total compensation per hour worked for private-industry office and administrative support workers in June 2026.[3]
- Thirty-five calls a day and one and a half minutes per row are set for this example; time a few rows in your own office and replace them.
- Lower the share typed by hand to see the effect of exporting time, number and missed-call columns from the phone system.
Example only: modelled, not measured. With every row typed by hand, the example comes to about 19 office hours a month.
Reading one week of rows
Totals hide leaks; rows show them. After one full week, sort the log three ways: by hour of the day against “answered by”, by outcome against reason, and by callback attempts against final outcome. Each sort answers one question: when do calls go unanswered, which kinds of call don't book, and which callbacks never land.
Inbound calls logged
Real job inquiries (not vendors, spam or billing)
Answered live by a person
Booked
Example figures for a fictional shop, not industry benchmarks.
Example: the fictional shop above finds that 21 of its 33 unanswered job inquiries came between noon and 1 pm or after 5 pm, that eight lost calls were price questions from first-time callers, and that eleven callbacks went out as calls with no voicemail and no text. Three fixes follow: lunch cover, a better answer to price-shopper calls and a callback rule that always leaves a voicemail. None of those would have shown up in “212 calls this week.”
The log is the raw material, not the analysis. For a structured first review, the one-week call audit gives outcome buckets and a one-page report, and the call-to-revenue funnel follows the booked rows through to paid invoices. Two or three numbers from the log belong on your weekly scorecard from then on.
Keeping it going after week one
Name one owner
One person owns the template, the drop-down lists and the weekly review. Everyone else just fills in rows.
Check blanks daily
Five minutes before close: any row without an outcome gets one, and every missed call has a callback attempt logged.
Review weekly
Twenty minutes, same three sorts every week, one fix chosen and an owner for it.
Prune monthly
Delete columns nobody used, merge drop-down values nobody can tell apart, and clear out codes that crept in.
Clear old rows
Decide how long rows for non-customers are kept, then delete them on schedule, as the FTC's guidance on keeping only what you need suggests[6].
Techs who take calls in the field need a way in too, or their calls vanish. A two-tap form on the phone, or a text to the office, is enough. A one-van owner-operator has the hardest version of this problem: in the solo plumber example, about three in ten calls are missed while he's on the tools, and in the modelled setup the emailed summary of every forwarded call becomes his log. Answering calls as a solo plumber compares the other options.
Every few months, test the log against reality with a mystery shop of your own line: if a test call that rang out never shows up as a missed row, the log has a gap.
If you record calls and link recordings from the log, check the consent rules for your state first; call recording consent covers one-party and all-party states.
Frequently asked questions
Should we use Excel or Google Sheets for a call log?
Either works. What matters is that everyone writes to the same file, drop-downs keep codes consistent and access is limited to the people who need it. A shared online sheet is easier when techs or a remote CSR add rows.
Do spam and vendor calls go in the log?
Yes, as “not a job” with a one-word reason. It takes seconds and shows how much of the phone time goes to calls that will never book. Leave them out and your answer rate and booking rate look worse than they are.
How long should we keep call log rows?
As long as they're useful for decisions, then delete them on a schedule. The FTC's guidance is to keep only what you need for your business[6]. Rows linked to booked jobs live on in your job records anyway.
Can we write card details in the log to save time?
No. Card security codes must not be stored after authorization under PCI rules[2], and card numbers belong only in your payment system. Write “card taken” in the outcome column and nothing else.
Our phone system already has call reports. Why keep a log at all?
The report records that a call happened. The log records why it happened and what came of it. Most offices end up exporting the report and adding the reason and outcome columns by hand, which is the split this template is built for.
References
Pew Research Center (2020). Most Americans don't answer cellphone calls from unknown numbers (opens in a new tab) ↑
PCI Security Standards Council (2026). FAQ: Can card verification codes/values be stored for card-on-file or recurring transactions? (opens in a new tab) ↑
U.S. Bureau of Labor Statistics (2026). Employer Costs for Employee Compensation, Table 4: Private industry workers by occupational and industry group (June 2026) (opens in a new tab) ↑
U.S. Census Bureau (2023). County Business Patterns 2023, U.S. national file (cbp23us) (opens in a new tab) ↑
Google (2026). About ad rankings - Local Services Help (opens in a new tab) ↑
Federal Trade Commission (2026). Protecting Personal Information: A Guide for Business (opens in a new tab) ↑
National Conference of State Legislatures (2022). Security Breach Notification Laws (opens in a new tab) ↑