Counting What Earned Its Keep: A Year-End Lead Source Audit

A lead source audit ranks every channel you paid for this year by cost per signed job, not by lead volume or cost per lead. Done at year end, it tells you which sources earned their place in next year's budget, which ones deserve a fair retest, and which ones should stop drawing money in January.

Why does the end of the year force a lead source audit?

Budgets reset in January, and most of them reset on assumptions made the previous January. That is the problem. The market you bought leads in twelve months ago is not the market you are buying in now, and a plan carried forward without a review is a plan built on a stale reading of demand.

The demand picture itself has moved. Construction spending in September 2026 ran at a seasonally adjusted annual rate of $2,186.1 billion, 2.7% below the September 2025 estimate of $2,245.8 billion, according to the U.S. Census Bureau's monthly construction spending release.

Across the first nine months of the year the total came to $1,649.6 billion, 2.9% below the $1,698.6 billion recorded for the same period in 2025, per that same Census release. A source that looked efficient against last winter's demand may simply have been riding a larger market.

Costs are pressing from the other side at the same time. Seventy-seven percent of firms reported rising costs of goods, services or wages, tariff-related cost increases, or both, in the Federal Reserve Banks' 2026 Small Business Credit Survey.

Among firms that buy inputs from outside the United States, 76% passed at least some of those increases on to their customers while 60% absorbed at least some of them, according to the same survey. When margin is thinner, the gap between a source that converts and one that merely delivers inquiries stops being an academic distinction.

An audit is also the only honest input to a budget. If you are rebuilding next year's numbers from scratch, the sequence matters: audit first, then plan. A plan built for marketing teams buying leads at an agreed cost depends on knowing what each source actually produced, and this is the exercise that produces that answer.

Which number should rank your sources, and which one misleads you?

Most source reviews rank on cost per lead, because cost per lead is the number every platform hands you without being asked. It is also the number least connected to whether you made money. A cheap lead that never answers the phone is not cheap. It is a line item with nothing behind it.

Rank on cost per signed job. Divide what you spent on a source over the period by the number of jobs you actually sold from it. Everything else in the audit exists to explain that figure or to tell you whether it can be improved.

The last pair matters together. A source with a cost per signed job of three hundred dollars that brings small repair work is not obviously better than one at six hundred dollars that brings full replacements. Rank on cost per signed job, then sanity-check the ranking against the average job size each source brings.

What records do you need before you start?

Pull twelve months. A shorter window buries seasonality, and seasonality is real in most trades. Demand swings hard between quarters in roofing, heating and remodelling, and far less in categories such as locksmithing or pest control, so the shape of the year differs by the trade you work in.

You need five things, per source:

• Spend. Everything attributable to that source, including agency or management fees, not just the media cost. A managed channel whose fee is excluded will always look better than it is.
• Lead count. Deduplicated. The same homeowner arriving twice in a week is one lead, and sources differ a lot in how much duplication they push.
• Contact outcome. Whether anyone reached the person, and how long the first attempt took. This one field does more work than any other in the audit.
• Quote or estimate count. The point at which the lead became a real opportunity.
• Signed jobs and their value. Tied back to the originating source, not to the last touch before signature.

That last point causes most of the disputes. If your records credit the final interaction, every source that sits early in the journey will look worthless and every retargeting line will look brilliant. Decide on one attribution rule, apply it to all twelve months, and write it down next to the results so the next person reading the audit knows what they are looking at. Stage metrics are only comparable when they are counted the same way across every source, and an audit that mixes two attribution rules will rank sources by bookkeeping rather than by performance.

How do you build the scorecard?

One row per source, one column per stage, in that order, so the drop-off is visible across the row rather than buried in a ratio. The figures below are round numbers chosen to make the arithmetic readable, not benchmarks to measure yourself against.

Read the table from right to left. The ranking column is the last one, and the moment you sort by it the lesson usually arrives: in the example above, the source with the lowest cost per lead carries the worst cost per signed job, and the source with the highest cost per lead is not the worst performer in the set. Volume and unit price both pointed the wrong way.

Then read each row left to right to find where the loss happens. A row that loses people between lead and contact has a reachability problem. A row that contacts well and quotes badly has a qualification or pricing problem. A row that quotes well and closes badly usually has a competition problem, meaning the same homeowner is holding three other quotes.

How do you tell a bad source from slow follow-up?

This is the question that decides whether your audit is useful or merely confident. Response speed changes outcomes enough to impersonate source quality, and a source blamed for a problem your own calendar created will be cut for no reason.

The scale of the effect is documented. An audit of 2,241 United States companies measured how long each took to answer a web-generated test lead: 37% responded within an hour, 16% between one and 24 hours, 24% took more than a day, and 23% never responded at all, according to research published in Harvard Business Review.

Among the companies that did respond within thirty days, the average response time was 42 hours, per that same audit. A lead bought on Monday and called on Wednesday is a different product from the one the source actually sold you.

The researchers also examined 1.25 million sales leads across 29 B2C and 13 B2B companies and found that firms attempting contact within an hour of the enquiry were nearly seven times as likely to qualify the lead as firms that tried just one hour later, and more than sixty times as likely as those that waited a day or longer, in the same body of research.

So before you judge any source, split its leads by time to first contact and compare like with like:

• Take only the leads contacted within an hour and recompute contact rate, quote rate and close rate for that subset.
• Compare that subset across sources. This is the fair comparison, because it removes your own response time from the measurement.
• If a source performs acceptably in the fast subset and badly overall, you have an operations problem, not a sourcing problem. Fix routing and staffing before you cut the budget line.
• If a source performs badly even in the fast subset, the problem is the source. Now you can act on it.

The same split tells you something about delivery. A source that delivers leads at midnight on Saturday and one that delivers them at ten on Tuesday morning will produce very different contact rates in a business that answers the phone during office hours, and that is a scheduling fix rather than a verdict on lead quality.

What should you cut, keep or test next year?

Three decisions, and a rule for each.

Cut a source when it fails in the fast-contact subset, when its cost per signed job sits above the value of the jobs it brings, or when it has had a full twelve months and a fair retest and still cannot clear your threshold. Cutting is also a decision about attention, not only money: a weak source consumes dispatcher time that a strong one would use better.

Keep and fund further a source that clears your cost per signed job threshold with room to spare and shows no sign of saturating. Before you raise its budget, check whether the volume is actually available. Many local sources cap out, and the extra spend goes to broader, worse-matched inquiry rather than more of the same.

Test in two cases. The first is a source that failed on operations rather than quality, which deserves one clean quarter with fast follow-up before any verdict. The second is a source you have never tried, funded deliberately and small, with a decision date written in the calendar before the money is committed. A test with no end date is not a test, it is a subscription.

Set the threshold before you look at the results, not after. Decide what a signed job is worth and what share of that you are willing to spend to win one, and apply that number to every row. Thresholds chosen after the numbers are in front of you tend to be chosen to protect whatever you already wanted to keep.

Where the economics have to work from the first job rather than from an annual average, a model that charges per delivered lead rather than per click shifts the media risk off your side of the table. That trade-off is the basis of performance-based lead generation, and it changes the shape of the audit: the ranking column stops depending on how well someone else managed your campaigns.

How often should you repeat this?

Run the full version once a year, in the last quarter, so the output lands before budgets are set. Run a shortened version quarterly, using just two columns: cost per signed job and contact rate by source. Those two catch nearly every drift worth catching, and they take a morning rather than a week.

Keep each year's scorecard. The second audit is far more useful than the first, because the interesting question stops being which source is best and becomes which sources are getting better or worse, and how quickly.

Frequently asked questions

What counts as a good cost per signed job?

There is no universal figure, because it depends entirely on what a signed job is worth to you and what it costs to deliver. The practical test is the one you can run in a minute: work out the margin on an average job, then decide what share of it you are prepared to spend to win the work. That share, in currency, is your threshold. Any source above it is not a bargain waiting to be understood.

How many months of data does a lead source audit need?

Twelve, if the records exist. Trades with strong seasonal swings will produce misleading rankings on anything shorter, because a source tested only in a peak month gets credit for the season. If you have less than a year, say so on the scorecard and treat the result as provisional rather than pretending the window was long enough.

Should you cut a source that signed nothing last quarter?

Not on one quarter alone, and not before you have looked at response times for that source's leads. A quarter is small enough that a single busy stretch, a staffing gap or a seasonal trough can produce a zero that says nothing about the source. Check the fast-contact subset first, and if that is empty, you have found an operations problem rather than a sourcing one.

Does a cheaper lead ever cost more?

Regularly. Price per lead and price per signed job move independently, because everything between the two is governed by reachability, qualification and competition. Shared leads sold to several contractors at once are the clearest case: the unit price is lower and the close rate is divided among everyone who bought the same inquiry. The only way to know which way it nets out for you is to carry both columns in the audit.

What is the single most common mistake in a lead source audit?

Ranking on cost per lead and stopping there. The second most common is crediting the last interaction before signature, which quietly transfers the credit from the sources that created demand to the ones that happened to be present at the end. Fix the ranking metric and the attribution rule, and most of the rest of the audit takes care of itself.