What a move in each plumbing contractor KPI tells you

What to take away

A move in a measure is a question to answer, not a verdict. Find the cause before you change anything.
Read the trend across at least three months and the distribution, not the monthly level. One unusual job moves a small firm's average badly.
Ask first whether the business changed or the recording changed.
Hold the review on the same day each month, for one hour, with the same definitions.
If a measure has never changed a decision, remove it.

Read the trend, the spread and the recording first

Three habits keep a monthly number from misleading you: look at the trend across at least three months, look at the distribution rather than the average, and ask whether the business changed or the recording changed.

Three habits for reading KPIs

Look at trend across at least three months
Look at distribution, not the average
Ask whether business or recording changed

A single month in a small firm is dominated by whatever unusual thing happened in it. A lone bad month is a prompt to look at the jobs behind it, and the trend over three months is what you act on.

Two firms with the same average days to collect are in different situations if one has a handful of very slow payers and the other has uniform lateness. The average cannot tell them apart, and the two firms need different fixes.

A sudden improvement usually means someone stopped recording something. Check that first, every time. A number that moved because a habit or a definition changed has told you nothing about the work.

The six measures and what a move means

The six measures are first visit resolution, callback rate, unbilled completed work, days to collect, booked contribution by channel, and billable hours per truck per day. The table gives each formula and the first place to look when it moves.

Measure	Formula	When it moves, look here first
First visit resolution	Jobs resolved on the first visit, divided by jobs attended	A fall: stocking, then diagnosis, then whether the service menu has quietly widened
Callback rate	Return visits for the same issue within a stated period, divided by jobs completed	A rise in one job type: scope or materials. A rise for one person: training or supervision
Unbilled completed work	Value of work finished and not yet invoiced, plus its average age in days	Growth: an administrative bottleneck
Days to collect	Average days from invoice issue to payment received, by customer type	Lengthening: terms, invoicing accuracy, or a customer in difficulty
Booked contribution by channel	Revenue minus direct costs for jobs from each source, divided by jobs from that source	High call volume: its contribution per job, not its call count
Billable hours per truck per day	Hours billed to customers, divided by trucks on the road, divided by working days	A fall: travel, supply house trips, or unpaid diagnosis

First visit resolution moves with diagnosis quality, truck stocking, scheduling accuracy and skill, so a fall tells you to look in four places rather than none. A callback costs twice: the unpaid visit and the paying job that could have filled it. Track callbacks by technician and by job type, because those answer different questions.

When callbacks rise for one person, the answer is in sign-off and supervision, and the structure for it is in the hiring and training guide.

Unbilled work is the fastest cash improvement available to most firms, because finished and uninvoiced work is money you have already spent to produce. Days to collect needs care, since a profitable month can still be a cash crisis, and residential and commercial customers behave differently enough that a combined figure hides both.

Channels are usually judged on call volume, which is how a source producing small distant jobs looks successful for a year. Billable hours is the capacity measure and the pricing input at once, so a fall in it explains a fall in profit that the price list cannot. Measure where the day goes before adjusting prices.

Settle the exclusions before you read the move

Each measure needs five fields written down before the first month of data, and the exclusions field prevents most later arguments. Without it, a move may only reflect a changed rule.

Five fields before first data

Nametwo people mean same thing
Formulanumerator and denominator explicit
Data sourcewhich record it comes from
Exclusionswhat is left out and why
Ownerwho produces and answers questions

Decide now whether warranty visits count as jobs, whether a canceled call counts as an inquiry, and whether internal work counts at all. Then settle the exclusions for each measure:

First visit resolution:jobs that were always planned as multi-visit, and jobs where a part had to be ordered because the item was genuinely unusual.
Callback rate:the period that counts as a callback, and whether a return caused by a customer's own action is included.
Days to collect:how disputed invoices and retainage are treated.

First visit resolution vs callback rate

First visit resolution

Formula: Resolved first visit / attended
Source: Job records
Why: Diagnosis, stocking, scheduling, skill
Exclusions: Planned multi-visit, ordered parts

Callback rate

Formula: Return visits / completed
Source: Job records by property
Why: Costs unpaid visit plus lost job
Exclusions: Callback period, customer-caused returns

Both service measures come from job records, with the formulas in the table above, and both depend on the exclusions listed here.

The monthly hour

Hold the review on the same day each month, for one hour, with six numbers and the same definitions. A review that produces five actions produces none, and a review that produces a discussion produces nothing at all.

The monthly hour routine

Same day each month, one hour
Note which measures moved and by how much
Pick one change, name who by when
Check last month's action happened

Pull the six numbers from job records and the bank, with the definitions unchanged.
Compare each against the last three months, and note the spread as well as the average.
Pick the one move that needs a cause found, and write down who finds it.

All six numbers come from two sources: job records and the bank. A spreadsheet updated monthly is sufficient for a solo or two-truck operation. The records underneath must be captured properly in the field, which is a discipline problem rather than a software one.

The IRS lets you choose any recordkeeping system suited to your business, as long as it clearly shows your income and expenses. Operational records deserve the same treatment, and before discarding old ones, check whether you need to keep the records longer for other purposes.

What not to track

Revenue alone says nothing about whether the work paid. Number of jobs rewards taking small unprofitable work. Anything you cannot produce from records without a special exercise will not survive a busy month, and anything you have never once acted on should go.

Two additions deserve a mention only because they are often demanded. Utilization percentages calculated from estimates rather than records measure your optimism. Customer satisfaction scores gathered inconsistently measure who happened to reply.

The systems that hold all of this, and the wider function map, are covered in the software and KPI guide.

Common questions

How soon after a bad month should I act?

Treat one bad month as a prompt to look at the records behind it. Act on the trend across at least three months, once the recording has been ruled out as the cause.

A number improved suddenly. Should I celebrate?

Check the recording first. A sudden improvement usually means someone stopped recording something, and the number then says nothing about the work.

What is a good first visit resolution rate?

No published figure is comparable, because definitions and service mixes differ. Establish your own baseline over three months and work on the trend.

When should I change a definition?

Rarely, and never mid-year. When you must, keep both versions running for a period so the history remains readable.