Most contractors can tell you roughly where their leads come from. Very few can tell you which source produces the most profit, and those are frequently not the same answer.
The short answer: judge sources by cost per booked job and by the margin of the work they produce, not by volume. A source generating forty leads a month at 14% margin is worse than one generating eight at 32%.
The two numbers that matter
Cost per booked job, not cost per lead:
Cost per booked job = total spend on that source ÷ jobs actually won
A source at $40 a lead that closes at 10% costs $400 per job. A source at $150 a lead that closes at 50% costs $300 per job — and is cheaper despite looking almost four times more expensive.
Average margin of work from that source. This is the one nobody tracks and it changes every conclusion.
Together:
| Source | Spend | Leads | Booked | Cost/job | Avg margin |
|---|---|---|---|---|---|
| Referral | $0 | 14 | 9 | $0 | 31% |
| Repeat customer | $0 | 8 | 7 | $0 | 29% |
| Google search | $2,400 | 22 | 6 | $400 | 26% |
| Lead marketplace | $3,200 | 41 | 8 | $400 | 15% |
| Door knocking | $1,800 | 19 | 5 | $360 | 24% |
Read it by volume and the marketplace wins — 41 leads, 8 jobs.
Read it properly and it is the worst source on the list. Same cost per job as Google search, eleven points less margin. Those eight jobs occupied crew time that could have gone to 26% work.
That is the real cost of a poor lead source: not the fee, but the capacity it consumes.
Why marketplaces produce thin work
It is structural, not a failing of any particular platform.
When a lead is sold to several contractors at once, the customer's frame is comparison. They are expecting multiple quotes and the easiest thing to compare is price. You arrive already positioned as one of several, before you have said anything.
Referrals invert this. The customer arrives pre-sold, often without shopping at all. You are not one of four — you are the one their neighbour recommended. Price sensitivity drops, close rate rises, and the work runs smoother because trust is already established.
The margin difference between those two situations is typically ten to fifteen points on identical work.
That does not mean never use marketplaces. It means know what they are: volume at low margin, useful for filling a thin schedule, dangerous as a primary channel — because a business built on them cannot raise prices.
Tracking it without new software
You need one question and one column.
The question, asked on every single call: "How did you hear about us?"
Not optional, not sometimes. Every call, logged before anything else happens. Whoever answers the phone needs to know that this is part of the job.
The column, in whatever you already track jobs in: lead source. When the job closes, you already have margin from job costing. Now you can sort by source.
After ninety days you will have something most contractors never have — an evidence-based answer to where the good work comes from.
Expect a surprise. Almost everyone finds that one source they were proud of is producing thin work, and one they treated as incidental is producing their best jobs.
The sources that usually win
Patterns hold across most trades businesses.
Referrals. Highest close rate, highest margin, zero cost. The only problem is that they are not controllable — which is a reason to systematise them, not to ignore them. Asking at the right moment matters more than most owners think. The right moment is at completion, when the customer is happiest, and the ask needs to be specific: not "tell your friends" but "if you know anyone with a roof like yours, I'd appreciate the introduction."
Repeat customers. Cheapest work you will ever get, and most trades businesses have no system for staying in touch. A list of past customers and one useful message a year is close to free money.
Organic search. Slower to build, compounds indefinitely. Someone searching for a specific problem in your city has high intent and is not necessarily comparing four quotes.
Trade and referral partnerships. Other contractors in adjacent trades, property managers, insurance adjusters, real estate agents. High margin, high trust, and durable once established.
Your own past quotes. The most overlooked source in the trades. People who asked for a price and did not proceed are the warmest list you own, and most contractors never contact them again. A follow-up at six months costs nothing.
The honest limits of this
Two cautions.
Ninety days of data is a signal, not proof. Eight jobs from one source is a small sample. Do not kill a channel on one bad quarter, and do not pour money into one on one good one.
Attribution is genuinely messy. Someone sees your truck, hears you mentioned, then searches your name. They will say "Google," and Google will get the credit for work that was really generated by the truck and the conversation. Brand-building activity gets systematically under-credited by any tracking system, which is worth remembering before cutting anything that does not show up directly.
What to do with the answer
Spend more where cost per booked job is low and margin is high. Obvious, and almost nobody does it because they never calculated it.
Do not just cut the worst source — replace it. Capacity you free up is only valuable if something better fills it. Cutting a channel without a replacement means a thinner schedule, and a thin schedule leads to discounting, which is where margin really goes.
Build the sources you control. Referrals and repeat customers have the best economics and the worst predictability. Systems that make them more frequent — asking at completion, staying in touch, following up old quotes — are the highest-return marketing work available to a trades business, and they are almost free.
Watch the mix over time. A business whose lead mix is shifting toward paid, price-comparison sources is a business whose margin will follow it down. That trend is visible in this table months before it shows up in the P&L.
Start Monday
Add one column. Ask one question on every call.
In ninety days you will know something about your business that most of your competitors will never know about theirs — and you will be able to stop guessing about where the marketing budget should go.
The variable that beats the source
Before changing where leads come from, check what happens to the ones you already get — because response time moves close rate more than source does.
The pattern is consistent across every trade that has measured it: the contractor who responds first wins a disproportionate share of the work, largely independent of price. A lead answered within minutes closes far better than the same lead answered the next day, and the same lead answered three days later is usually gone.
The reason is not mysterious. Someone with a leaking roof is contacting several contractors in one sitting. The first to reply is talking to a person with an urgent problem and no alternatives yet. The third to reply is talking to someone who has already had a good conversation with somebody else.
This matters for the table above, because a slow response makes every source look worse than it is. A source producing forty leads that you answer in a day is not a weak source — it is a strong source being wasted, and swapping it for a different source will not fix anything.
Three things to check before touching the marketing budget:
How long does an inbound call or form take to get a real reply? Measure it for two weeks. Most owners guess under an hour and find the average is over six.
What happens to calls during working hours? If the phone rings while you are on a roof and goes to voicemail, you are paying for leads that end up with whoever answered theirs.
What happens to a quote after it is sent? Most contractors send it and wait. A single follow-up at three days is the cheapest close-rate improvement available to a trades business, and almost nobody does it.
Fix response time first. It costs nothing, it improves every source simultaneously, and it will change which sources look good in your table — sometimes enough to reverse the conclusion.
The 10 Numbers I Check Every Monday
One page. Ten minutes, every Monday. Every number has a line next to it that tells you when to act.
Get the sheet — freeCommon questions
How long before I can judge a lead source?
About ninety days, or thirty booked jobs, whichever comes first. Less than that and you are reading noise — one good referral or one bad month will swing the average. Do not cancel a channel on a single poor quarter, and do not double the budget on a single good one.
Why is my close rate so different between sources?
Mostly because of how the customer arrives. A referral arrives pre-sold and often without comparing quotes. A marketplace lead arrives expecting to compare several, with price as the easiest comparison. Same work, completely different conversation, and usually ten to fifteen points of margin between them.
How do I get more referrals?
Ask at completion, when the customer is happiest, and be specific. 'Tell your friends' produces nothing. 'If you know anyone with a roof like yours, I would appreciate an introduction' produces referrals. Then stay in touch with past customers — one useful message a year is close to free work.