There are only two real answers to how to stop firefighting in your business, and most service-business owners spend years on the wrong one. Answer A is to get better at absorbing fires — faster phone, better dispatch software, someone who screens the chaos. Answer B is to remove the fires — find out why each one exists and shut off the source. Both work. They work on different timelines, ask different things of you, and fail in completely different ways.
This piece puts them head to head, then gives you the 30-day sequence for the second one, because that's the one almost nobody runs properly.
First, name the thing you're actually fighting
Firefighting is not a discipline problem. I've watched owners with genuinely excellent discipline get chewed up by it, and I've been that owner. Discipline is what you're using to survive the loop; it is not what created it. What creates it is that your business has no defined path for a class of event, so the event routes to the only general-purpose processor available — you.
The tell is repetition. A one-off emergency is an emergency. The same emergency in a slightly different costume, eleven times a quarter, is an unbuilt system. Missing purchase order. Tech doesn't know the gate code. Customer calls asking where the crew is. Sub shows up with an expired certificate of insurance. A part that should have been staged isn't. Each one feels unique in the moment because the customer name is different.
It also takes more out of the day than the stopwatch says. Research from Gloria Mark's team at UC Irvine found people needed roughly 23 minutes to get back to an interrupted task. A four-minute call at 10:15 does not eat four minutes; it eats the morning's estimate you were three paragraphs into. That gap between the felt drain and the real drain is why owners underinvest in fixing this — the bill never arrives in one piece.
Head-to-head: absorbing fires vs. removing them
Here's the honest comparison. Both columns describe things real owners do, and neither is stupid.
| Dimension | Path A: Absorb (add capacity) | Path B: Remove (close the source) |
|---|---|---|
| What you actually do | Hire an office coordinator or answering service, buy dispatch/CRM software, add a shared inbox, put a phone tree in front of yourself | Log every interruption for two weeks, sort them by cause, and permanently close the three that eat the most hours |
| Time to relief | Days to weeks. Fast. You feel it immediately. | 4–6 weeks before the first fire stops recurring. Slow start, then it compounds. |
| Resource demand | Ongoing payroll or subscription, plus onboarding time — a permanent line item on the P&L | Roughly 6–10 focused hours across 30 days, front-loaded, then near zero |
| What happens at 2x job volume | Interruption count doubles too. You add another person or another tool. | The closed fires stay closed. Volume grows without the interruption count growing with it. |
| Effect on your calendar | Interruptions get filtered and batched — you're interrupted less often, but the underlying defect count is unchanged | Fewer events exist at all. The calendar gets quieter, not just better-buffered. |
| Primary failure mode | The new person becomes a second bottleneck who escalates 40% of everything back to you, because there's no rule for them to apply | You log for two weeks, feel busy, never do the sort, and nothing changes |
| What it demands from you | Money and a hiring decision | Two weeks of honest self-observation and the willingness to write rules down |
| Durability | Ends when that person quits | Survives turnover if the rule lives in a document, not a head |
Where Path A genuinely wins
I don't want to strawman the absorb path. It wins outright in three situations. When the interruptions are real customer demand — inbound calls from people who want to buy — you don't want to remove those, you want someone to answer them fast. When you're in a seasonal spike and the volume is temporary, adding capacity is the correct temporary answer. And when you're past roughly 8–10 field staff, some of this is simply a role that needs to exist; at that scale, hiring your first operations person isn't avoidance, it's org design.
Where Path A fails is when it's used as an anesthetic. If you hand an unsystematized business to a coordinator, you have not removed the decisions — you've added a routing hop in front of them. Within a month they're texting you at 4:40pm asking whether to comp the trip charge, because nobody ever wrote down who can comp a trip charge.
Where Path B genuinely wins
Path B wins whenever the interruption is a defect rather than demand. Defects are anything that shouldn't have needed a human at all: information that existed but wasn't where the tech was, a decision that has been made identically forty times, a deadline that is on a federal calendar and could have been scheduled a year in advance.
Why does firefighting come back after a good week? Because a good week is usually the result of extra owner effort, not a structural change. You worked ahead, you got lucky with the schedule, and the interruption sources were all still intact underneath. When your energy dips the following week, the same fires reignite. A fire is only actually out when the event that caused it can no longer occur — not when you handled it well.
The verdict
Pick Path A when the interruptions are revenue-bearing (inbound leads, real customer service), when the spike is seasonal, or when you're already over roughly 10 employees and the coordination load is a genuine full-time job.
Pick Path B when you can name the same interruption happening more than twice a month, when the interruptions are internal (your own crew, your own vendors, your own paperwork), or when you've already tried adding a person and the escalations still land on you.
Do both, in this order: run Path B first for 30 days, then hire into a business with fewer fires. Every fire you close before you hire is a fire that person never has to be trained on. Owners who hire first almost always over-hire, because they're staffing to a chaos level that didn't need to exist.
The 30-day sequence for closing fires
Days 1–14: keep an interruption ledger
One document. Every time something pulls you off what you planned to do, you write one line. Do not analyze while you log — analysis during the logging phase is how people quit on day four.
- Time — when it hit
- Source — tech, customer, vendor, sub, bank, software, myself
- What they needed — in six words
- Minutes lost — the call plus the recovery, be honest, round up
- Why it reached me — the one field that matters. "Only I know the answer." "Only I have the login." "Nobody's allowed to decide." "It slipped and now it's urgent."
Two weeks, not one. One week can be an unusual week. Two weeks catches a payroll run, a weekend, and at least one invoicing cycle. Most owners running a field business land somewhere between 30 and 70 entries. Expect to feel slightly ill when you tally the minutes column.
Day 15: sort into four buckets
Sit down with the ledger and put every line into exactly one bucket. This is the whole intervention — the log is worthless until it's sorted.
- Prevention. This interruption should not exist. Information was missing, staged late, or lived in one person's head. The fix is upstream, in how the job gets prepared or handed off.
- Policy. A decision you keep re-making identically. The fix is a written rule plus a spending or discount authority, so someone else can decide it without you.
- Delegation. Real work that simply needs an owner other than you — plus a procedure so it doesn't bounce back. If handoffs keep returning, the problem is usually the missing procedure, not the person; the framework for letting go cleanly is worth reading before you assign anything.
- Keep. Genuinely yours. Pricing a large bid, firing someone, a key account relationship. Be ruthless here — this bucket is where owners hide.
Then sort every bucket by total minutes, not by frequency and not by annoyance. The fire that irritates you most is rarely the fire draining the most hours from your week.
Days 16–30: close the top three. Only three.
Three, because four is how you end up with zero. Each closure gets a named owner, a written rule, and a date it goes live. If you're building the written piece for the first time, a fill-in SOP template beats a blank page every time — most of these rules are four bullets and a screenshot, not a manual.
Here's what closure looks like in practice, using three fires that show up in nearly every service business I've looked at:
- "Where's my tech?" calls. Ledger showed 9 calls, 11 minutes each with recovery. Bucket: prevention. Closure: an automated en-route text with a 30-minute window and the tech's first name, triggered from the schedule. Fires remaining: near zero. This is one of the highest-yield touchpoints in the whole customer communication sequence.
- Expired subcontractor COI discovered on site. Ledger showed 2 events, but 90 and 140 minutes each plus a rescheduled crew. Bucket: prevention. Closure: every COI expiration date goes in one sheet with a 30-day reminder, and no sub gets scheduled if the date is inside 30 days. Certificates run on 12-month policy terms, so this is a fully predictable annual event that you were treating as a surprise.
- Payroll deposits and quarterly filings. Ledger showed 4 panicked scrambles. Bucket: delegation, with a prevention layer. Closure: the IRS employment tax due dates go on the calendar for the next 12 months, and one named person owns them. This one has teeth: late federal deposit penalties climb from 2% at 1–5 days late to 5% at 6–15 days, 10% past 15 days, and 15% once an IRS notice has gone unanswered for 10 days. That's a real number leaving the account, not just stress.
A fourth pattern worth checking while you're in there: Thursday scheduling scrambles that push nonexempt field staff past 40 hours. Under the FLSA overtime rules, those hours bill at 1.5x. A chronic Thursday fire shows up in the weekly payroll number, which makes it much easier to justify spending two hours fixing the schedule.
How to tell whether it actually worked
Owners routinely believe a fire is closed when it has only gone quiet for a week. Use three checks instead of a feeling.
- Recurrence count. Pick the exact event you closed and count it for the next 30 days. Nine "where's my tech" calls dropping to one is a closure. Dropping to five is a partial fix, usually because the rule only covers one crew or one job type.
- The vacation test. Take one full day off the phone. Whatever escalates to you anyway is still open, regardless of what the document says. If that thought makes your stomach drop, that's its own finding — and it's the same muscle a continuity plan is built to test.
- Interruptions per job, not per week. This is the number that matters. If you ran 40 jobs last month with 52 interruptions and 48 jobs this month with 50, you're winning even though the raw count barely moved. Absolute counts flatter slow months and punish good ones; the ratio tells the truth.
One caution on measurement: don't re-log for two full weeks every time you want to check. A single Friday tally of "what pulled me off plan this week" is enough for maintenance. The 14-day version is a diagnostic, not a habit.
The fires you should not close
Two cautions. Don't close a fire by making the customer experience worse — a phone tree that stops interrupting you by also stopping buyers from reaching you has moved the problem, not removed it. And don't close a fire by writing a rule so rigid your crew has to break it weekly; a rule everyone violates is worse than no rule, because now you've taught people that written procedures are decorative.
Also accept that some interruptions are the job. If you're a nonemployer firm — and per the SBA Office of Advocacy, roughly 81% of US small businesses are — a certain volume of everything landing on you is structural, not a failure. The goal isn't zero interruptions. It's that the interruption count stops rising in lockstep with revenue.
Run it again next quarter
Three closed fires won't finish the job; they'll just make the next tier visible. Re-run the ledger every quarter for a week, close the new top three, and you're operating on a cadence rather than a crisis. Owners who bolt this onto their weekly operating rhythm — one 30-minute slot to review what interrupted them — stop needing the full 14-day version entirely.
If you're staring at a ledger with 60 lines and no idea which thread to pull first, the sequencing question is its own problem, and it's worth reading where to start systematizing when everything feels urgent before you start building. The eventual destination is a business where the schedule, the books, the website inquiries, and the field work each have a defined path that doesn't route through your attention — which is the whole point of building an operating system for a service business rather than just working harder inside one.
At Turnkey Services we spend most of our time on exactly this: the unglamorous back-office plumbing that determines whether a Tuesday is a workday or a rescue mission. The ledger is free. Start it Monday.
Frequently asked questions
How long does it take to stop firefighting in your business?
Expect two weeks of logging, one afternoon of sorting, and two weeks of building. The first fires stop recurring around week four to six. Each quarterly repeat gets faster because the ledger is shorter.
Should I hire someone or fix the systems first?
Fix first, then hire — unless the interruptions are inbound sales calls or you're past roughly 8–10 field staff. Every fire you close before hiring is one your new person never has to be trained on, and it keeps you from over-hiring to a chaos level that didn't need to exist.
Why does the firefighting come back after a good week?
A good week usually comes from extra owner effort, not a structural change. The interruption sources were still intact underneath. A fire is only out when the event that caused it can no longer happen.
Which interruptions should I close first?
Sort by total minutes lost, not frequency or annoyance. Two 90-minute events beat nine 11-minute ones. Then pick exactly three — attempting four or more reliably ends with none of them finished.
What if every interruption feels like it genuinely needs me?
That's normal on day one and usually false by day fifteen. Add a 'why it reached me' column to your ledger. Most entries resolve to 'only I have the login' or 'nobody's allowed to decide' — both are fixable without you.