Somebody asked me this month whether he should just start paying plumbers for water-loss calls. He'd done the math on his Google spend, decided it was ugly, and figured a referral fee was the cheaper door. It's the right instinct and the wrong arithmetic, and the reason why takes about fifteen hundred words.
What does a contractor lead actually cost?
About $25 if it came from a referral and about $91 if it came from paid search — $90.92, in the Pipeline On 2026 data, across the home-service trades. That is a 3.6x spread on the same outcome: a phone ringing with somebody who wants work done.
Most owners already believe this without having seen the number. Ask where your best jobs come from and nobody says AdWords. What the number adds is a scale for the gap, and the gap is large enough to reorganize a marketing budget around.
Then you get to restoration and the number falls apart.
Why does the same referral cost $25 in one shop and $1,000 in another?
Because in restoration the referral has a market, and markets set prices. Right now contractors in water damage are paying plumbers as much as $1,000 for a single water-loss referral. The plumber standing in the basement at the burst pipe is holding the highest-intent lead in the trades — the customer is wet, insured, and deciding in the next ninety minutes — and every restorer in the metro knows it. They bid. The price went to four figures.
So the honest version of the headline is this: $25 is a cross-trade average that describes an HVAC shop getting a neighbor's name from a happy customer. It does not describe the restoration market, where 60 to 80 percent of the work originates from an insurance claim and the claim originates from whoever was standing there first. If your business is claims-heavy, the $25 figure is not your number and quoting it at yourself is a way to feel good on a Tuesday.
What the $1,000 tells you is more useful than what the $25 tells you. A four-figure spot price on a phone call is what an auction produces, and an auction has three properties worth naming. The price rises until the margin is gone, because that's what bidding does. The winner buys one call and nothing else — the plumber owes him nothing next Tuesday. And cash is the single currency every competitor in the market can also produce, by lunchtime, in any amount.
Which is the part the arithmetic usually misses. A paid referral and a paid click are the same category of expense. Both are rented.
What does the second job cost?
Zero, if you own the relationship — another $91 if you don't. This is the calculation almost nobody runs, and it's where the whole comparison actually gets decided.
Price a paid-search lead honestly and you're pricing a transaction. You spent $91, the job closed, and the next job from that same homeowner costs you $91 again unless something in your operation captured them on the way through. Most operations don't capture anything. The job closes, the invoice goes out, the customer is a row in QuickBooks, and eighteen months later when their water heater goes they search again and you pay again — possibly for the same person you already served.
The referred job runs the other direction. The homeowner arrived pre-sold, because someone they trusted vouched for you and that vouching did the work your first sales call would otherwise have to do. And the partner who sent them is still standing in basements next month. One referral from a good partner is a subscription somebody forgot to bill you for, and the shop that treats it as a one-time lead acquisition is misreading its own P&L.
Run three years forward. The paid channel resets to zero every month you don't fund it — three years of spend and you own nothing you can point to. Three years of a working partner relationship and you own a channel that a competitor cannot buy, because it isn't for sale and it isn't listed anywhere.
That's the case for referrals, and it is not the case for referral fees. The fee buys the first job on the same terms the click did.
So why doesn't your software know any of this?
Because the tools were built around the invoice. Open any of the five major field-service platforms and look for the place a referral partner lives. Jobber and Housecall Pro treat partner tracking as a spreadsheet problem — meaning, they don't do it. ServiceTitan has commission tracking bolted onto its top tier. Service Fusion and Connecteam have nothing. Your cheapest channel by a factor of three-and-a-half is the one thing your system has no record of.
The pricing model makes it worse in a way that's almost funny. The referral engine works, you win more jobs than your crew can cover, you hire — and the platform charges you for hiring. The enterprise standard in HVAC and plumbing runs roughly $250 to $498 per technician per month, before a $5,000-to-$50,000 implementation and six to twelve months of onboarding. Housecall Pro's top tier covers eight users and then bills $35 for each one after that. The channel that compounds runs headlong into a cost structure that compounds against it.
We built WORKFORCE around that, which is the reason to say plainly what it does and does not do. Partners get their own login: they submit a referral through a structured form, it gets an ID, and they can see its status without calling your office. Every technician is included in the plan — the fourth truck doesn't reprice you. What it does not do is pay anyone. SR Credit, the commission layer, is roadmap and not shipped, so no partner is earning or collecting money through the platform today. Given everything above, I'd argue the payout is the least interesting part of a referral network, but I'd rather you hear the limit here than find it in month two.
What actually changes on Monday?
One question at intake, and it isn't "how did you hear about us?" That question returns "the internet" and "a friend," which is the same as returning nothing. Ask "who should I thank for sending you?" — it presupposes a person, so the caller reaches for a name, and it's a warm question because the reason is true.
Then three things that are less interesting and matter more:
Make it a required field at intake rather than a note at invoicing. By invoicing your office is trying to get paid and the customer has forgotten. Optional fields are blank on your busiest week, which is precisely the week your referrals spike.
Report back. Every partner who sent you something hears what happened to it — the job, the date, the outcome — without having to ask. This is the entire relationship, and it takes about two minutes a month per partner. A partner who hears nothing assumes it went badly and quietly stops sending. They almost never tell you why.
Rank partners quarterly by what flows in both directions. The relationships where you're net-taking are the ones a competitor's cash offer will eventually take from you.
None of this requires software. A spreadsheet carries a referral ledger fine for a while, and it fails the same way in every shop — a slammed week, an untagged job, a partner left blind. Start the spreadsheet this week anyway. You cannot automate a process you have never run manually.
The deeper version of this — the four things you can trade a referral partner that cash can't buy, and why insurance agents in most states legally cannot accept a fee from you at all — is in the Referral Partner Playbook.