Run Your Own Phone Room, Or Hire One?
Building Your Own Call Center: What It Costs, And What Gets You Fined
The short answer: build it if you have year-round volume in one market and somebody whose actual job is running the floor. Do not build it if your volume is seasonal, or if the person who would supervise it is also the person selling.
Two things decide it, and you can check both yourself -- what one person on the phone really costs you once you add everything up, and which of the calling rules carry a fine big enough to hurt. Both are below, with the source printed next to every figure. We run call centers for a living, which is the only reason we can write the second half.
When building your own is the right call
Build it yourself when
- You have steady, year-round call volume in one market -- not a storm season and then silence.
- Somebody can supervise the floor as their whole job, and it is not you between roof appointments.
- You are in one state, so you are learning one set of state rules rather than a dozen.
- You want the asset. Owning the reps, the recordings and the data is worth something to you beyond the appointments themselves.
- You can survive the learning curve. The first months of an in-house floor are the expensive ones, and they are tuition.
Do not build it when
- You are a three-crew shop. Honestly: do not. A supervisor to oversee two setters is a wage you cannot amortise, and the section below shows why.
- Your demand is storm-driven. A floor you staff for the peak is a floor you pay for in the trough.
- You dial into several states. Every one adds its own registry, its own clock and its own statute.
- Nobody in the building wants the job. A phone floor without a real supervisor becomes a compliance incident with a payroll attached.
- You need appointments this quarter. Hiring, training and script iteration are slow, and the ramp is real.
We sell the alternative, so read that split with the appropriate suspicion -- and then check it against the sources below, which are all federal and none of them ours. If you land on "build it", the rest of this page is the honest version of what you are taking on. If you want somebody to talk it through against your actual numbers, that is what a consultation is for, and it is a different purchase from either building or outsourcing.
What a seat costs, at published wage data
Start with the wage, because it is the only part of the build with a federal source behind it. Three different occupations get called "a call center job" and the Bureau of Labor Statistics prices them differently. Getting the wrong one into your model is the most common mistake in a build-versus-buy spreadsheet.
Read those three together rather than picking one. An outbound appointment setter is not an inbound customer service representative, and the two codes sit far enough apart that pricing an outbound floor at the service-desk wage overstates it while pricing a service desk at the setter wage understates it.
The third number is the one left out of almost every DIY estimate. A floor is not five agent wages, it is five agent wages plus somebody to run them, and that person costs more than any of the agents. Note what the series actually is: BLS publishes no "call center manager" occupation at all, so the figure above is the general office and administrative-support supervisor series and we have deliberately not relabelled it. It is the closest published comparison, not a measurement of your floor manager.
What we have deliberately not done is total this up. Benefits, dialler licensing, telephony, list data, QA tooling, errors-and-omissions cover and recording storage are all real costs and we could find no primary source that publishes a defensible figure for any of them in this context. A cost table with three sourced rows and five guessed ones is worse than three sourced rows, so you get three sourced rows and an honest gap.
The hour you actually spend when a roofer answers the phone
In a small contractor the phone does not get answered by a dedicated person. It gets answered by whoever is nearest, and that is usually the most expensive person available. The pair of numbers below is the whole argument, which is why neither is published without the other.
A roofer out-earns a customer service representative, so an answered call in a small shop costs the more expensive hour AND the roof that hour was meant for. That is the real comparison, and it is the one a wage-only model misses entirely.
The size distribution is what makes it bite. Two thirds of roofing establishments have fewer than five employees, which means there is nobody in the building who is free to answer the phone -- everybody who could is already doing something that pays. Two caveats travel with that Census figure and they attach to different halves of it: it counts ESTABLISHMENTS rather than firms, so a roofer with three locations is counted three times; and it counts employer establishments only, so sole proprietors with no payroll are excluded. The second one makes the count a floor and makes the under-five share conservative rather than inflated, since every excluded non-employer would fall in that same bucket.
Two costs that only appear once you are already dialling
These are the ones missing from every build-your-own spreadsheet we have been shown, because they are not equipment and they are not payroll. They are the cost of being allowed to make the calls at all.
The registry fee scales with how many area codes you pull, and the first five are free -- which is why a single-market operator barely notices it and a multi-state one does. The rate is reset by the FTC every fiscal year, so treat the figure above as the published rate for the period stated and re-read 16 CFR 310.8 before you budget against it. Note also that 16 CFR 310.8 relieves some callers of the fee entirely; whether that includes you is a question for your lawyer, not for this page.
Five years of per-call detail plus scripts and consent records is a retrieval commitment, not a checkbox -- and that reading is ours rather than the rule's. The rule states the period and the record set. What it costs you to actually produce those records on demand, two years after the agent who made the call left, is the part you should price before you buy a dialler rather than after.
Why the seat is not where the money goes
Everything above prices the infrastructure. This is the part that decides whether the infrastructure produces anything, and it is measured on our own floor rather than asserted. Read the population statement first -- a percentage from a call floor means nothing without the number of calls it was measured on.
That is one campaign, in one vertical, over one window: roofing appointment setting across two months in early 2026. It is not an industry benchmark and it is not a forecast for your floor. It is offered as evidence that we measure the work, and because the finding below is the single most useful thing we can tell somebody deciding whether to build.
Same campaigns. Same lists. Same hours. Same dialler. The person on the phone moved the result by more than four times. Whatever you spend on infrastructure, that spread is the thing that determines your cost per appointment -- and it is bought with recruiting, scoring and coaching, none of which arrives with the software.
Both framings of the set rate are printed on purpose. Per contacted homeowner is the number that describes the phone room; per dial is the number that describes your invoice. If a vendor quotes you one, ask for the other, and ask which one their price is attached to. That question and the rest of the ones worth asking before you sign are written out as a sheet you can take into the call -- the call center vendor question list, with each answer traced to either a federal document or a measurement from this floor.
The rules that turn a phone room into a fine
This is the part every "how to start a call center" article skips, and the reason is structural: most of them are published by companies selling diallers and phone systems, whose interest is in the setup sounding simple. What follows describes published federal rules with a link to each primary source.
This is not legal advice. It describes published rules and links to the source for each. No attorney-client relationship is created by reading it, and nothing here tells you whether a given rule reaches your business -- that turns on facts about your campaign that this page cannot know. Rules move, too. Have a lawyer who works on telemarketing rules review your scripts, your consent records and your scrubbing process against every state you call into, before you dial.
First, the trap: closing face-to-face does not put you outside the rule
Contractors reason that because the sale is signed on a kitchen table, telemarketing rules are somebody else's problem. There is a face-to-face exemption, and it is narrower than that reasoning assumes -- it withholds exactly the four provisions that carry the penalties.
There is a genuine open question underneath this and we are not going to pretend it is settled: 16 CFR 310.2 defines telemarketing as a plan or campaign involving more than one interstate call, so a contractor dialling only inside their own state may sit outside the FTC's jurisdiction while remaining fully inside the FCC's rules and their own state's statute. We found no authority resolving that for a regional contractor floor, so the honest thing is to name the question and send you to counsel rather than to assert an answer in either direction.
What the dialler is allowed to do
Two rules govern the machine rather than the script, and both are the kind a predictive dialler will break by default if nobody configures it otherwise.
Two things worth drawing out, both of them ours rather than the rule's. A cap expressed as a rate means you need a dialler that can REPORT the rate per campaign, which is a procurement requirement most buyers do not know to ask about until after they have signed. And the calling window is measured at the called party's location, not yours -- if you dial across time zones that is a configuration problem before it is a policy one.
Note the asymmetry in the abandonment rule as well, because it is routinely misquoted in the other direction. It governs how long an OUTBOUND caller may leave a person hanging. There is no matching federal standard for how fast an INBOUND call to a contractor has to be answered, and anyone citing the two-second figure as an inbound service level is misreading it.
Who you are allowed to call, and the two clocks that disagree
Scrubbing is where a build-your-own operation most often goes quietly wrong, because nothing breaks visibly when it is done badly -- the calls connect exactly as before.
The two established-business-relationship clocks are the detail worth the most attention, because a shop that scrubs to one is out of step with the other and nothing in the dialler will tell them. Which clock governs a given campaign is a jurisdictional question, so this page states that both exist and differ and stops there.
It is also worth knowing what the people you are about to dial think of the exercise. The FTC number above measures that from the consumer side. Our own floor measures the same thing from the other end, on the corpus stated earlier:
Those six codes do not add up to the whole campaign and are deliberately not drawn as a chart: they cover under half of it, and the rest carry dispositions the source documents never enumerate. The load-bearing one is the do-not-call share. Every one of those is a request you are then obliged to record and honour, and an internal list you have to keep for years. Two independent measurements -- the FTC's from the consumer side and ours from the dialling side -- both say the same thing: a meaningful slice of the people you call do not want to be called, and managing that is an operational discipline rather than a one-time setup task.
Consent: what is settled, what is vacated, and what is still moving
This is where the roofing-marketing internet is least reliable, so both entries below carry their status explicitly rather than being summarised into a rule of thumb.
Half the roofing-marketing web will tell you that new FCC rules made shared leads illegal. That is false, and it has been false since the rule was struck down before it ever operated. Shared leads remain lawful; the homeowner who filled in one form is still called by several contractors, which is an argument about lead economics rather than about legality.
The revocation rule is the one to actually watch, and it is genuinely unsettled. The core duty is live now. The broader component is waived, has already been pushed back more than once, and is under active reconsideration -- so treat the date above as the current state of a moving file rather than as a deadline you can plan around, and check it again before you rely on it.
Everything above is federal. Your state is a separate problem.
We have deliberately kept this page to federal rules, because those are the ones that apply to everybody reading it and they can be stated once. State law is where a build-your-own operation actually gets caught, and it does not generalise: several states run their own do-not-call registry with its own refresh clock separate from the federal one, several set their own calling windows and their own damages, and several have their own statutes on what a contractor may say about an insurance deductible on a recorded call.
Storm-restoration states add a second layer that most people do not connect to the phone at all: contract-cancellation windows, deductible-rebate prohibitions, and rules about acting as a public adjuster. Those statutes are written about contracts, but the exposure is created by what a setter says on the call that books the inspection. The script is upstream of the contract.
We are not going to enumerate every state and get one wrong. The honest guidance is that the number of states you dial into is one of the strongest predictors of how hard your build will be, and that a lawyer reviewing your scripts against your actual call footprint is cheaper than finding out the other way.
If you would rather not run one
This is the part where we sell something, and it is at the bottom on purpose -- the decision above is worth making properly whichever way it goes. We build and run outbound floors: dialler deployment, scripts and rebuttals, recruiting and training bilingual reps, CRM and calendar integration, call scoring and ongoing coaching. You own the operation; we run it.
That per-appointment price belongs to the roofing storm-damage programme specifically, and we are not going to quote it as a generic rate for an industry it was not measured on. The build-out itself has no published price and is quoted per engagement -- what moves it is seats, hours and days covered, whether Spanish is required, how many lead sources and dialler integrations have to be wired, whether your CRM accepts a booked appointment cleanly, and how much script work runs before go-live. Full detail is on our pricing page. To put your own seat count and hours into the same arithmetic before you talk to anyone, the agent headcount calculator sizes the floor from your dial volume, and the dialer capacity calculator tells you what that floor needs in lines and licences.
There are three different purchases here and they are worth telling apart. If you want somebody to answer calls you already receive, that is inbound call center services. If you want somebody to place calls you are not making, that is outbound call center services, which is also where the rest of our measured campaign data lives. If you already run a floor and want a diagnosis rather than an operator, that is what call center consultants sell.
Still set on building it yourself? Good -- genuinely. The step-by-step version, including the startup costs question this page deliberately refuses to invent a total for, is in how to start a call center business. And if you want to see what the work looks like when it is aimed at storm-damage homeowners, that is our roofing call center.
Frequently Asked Questions
The questions owner-operators actually ask before they decide.
Should I build my own call center or outsource it?
What does a call center seat actually cost?
Does the FTC Telemarketing Sales Rule apply to a roofing contractor?
What is the abandoned-call limit?
When may I legally call, and whose clock counts?
Are shared leads illegal now under the new consent rules?
How long do I have to keep call records?
Is the hard part the dialer or the people?
What do you charge, and what does the build-out cost?
Is this page legal advice?
Talk it through against your actual numbers
Bring your call volume, your markets and your headcount. We will tell you which of the two answers above fits, including when it is "build it yourself and do not hire us".