How to know which ad produced which job

Most service businesses judge advertising on a feeling, because nothing tells them where the work came from. Fixing that takes a weekend and changes every decision after it.

August 20, 2026 · 5 min read · by Ido Katz

Ask most service business owners which of their marketing works and you get an answer with the word "probably" in it. Probably the Google ads. Word of mouth is probably the biggest. The Facebook thing probably did nothing.

Those are not opinions about the business. They are the absence of a measurement, filled in with whatever feels true — and what feels true is usually whichever channel produced the last memorable job.

The fix is not expensive and not clever. It is one weekend of setup, and after it every budget conversation stops being an argument.

Why "probably" is expensive

Say you spend $3,000 a month across two channels and you cannot tell them apart. One of them produces almost everything; the other produces almost nothing. You do not know which.

You have two ways to be wrong, and they cost differently.

Cut the wrong one and you lose most of your work overnight, then spend three months rebuilding while concluding that advertising does not work. Keep feeding the dead one and you pay for it every month forever, quietly, and it never announces itself.

The middle option — split the difference, keep both, spend a bit less on each — is the one most people take, and it is the worst of the three. You slow down the channel that works to keep funding the one that does not.

None of that is a marketing problem. It is a bookkeeping problem.

The three things you have to connect

Attribution sounds technical. It is three joins.

One: which ad brought them. A click that arrives with nothing attached is anonymous forever. Tagging the link fixes it — every ad gets its own tag, and the tag travels with the visitor.

Two: how they got in touch. Some fill in a form, some call. A form can carry the tag with it. A phone call cannot, unless you give each channel its own number.

Three: what happened next. Quoted, won, lost, how much. This is the join everyone skips, and it is the one that matters — leads are not the product, jobs are.

Miss any one and the chain breaks. Most businesses have the first, occasionally the second, and almost never the third.

The cheap version, in order

Tag every ad link. Add ?utm_source=google&utm_campaign=emergency&utm_content=ad-3 to the destination. Different tag per ad, not per platform — "Google works" is not a decision you can act on; "the emergency ad works and the maintenance one does not" is.

Put a hidden field on your form that records those tags, plus the page they landed on. Any form builder does this. Now every enquiry arrives knowing where it came from.

Give each channel its own phone number. This is the step people resist and the one that pays the most, because in most service businesses the phone is where the money is. Call tracking numbers cost a few dollars a month each, forward to your real line, and tell you which ad caused which ring. Use a distinct number on the Google listing, on the website, and in ads.

Write the outcome down. One row per enquiry: where it came from, what you quoted, whether you won it, what it was worth. A spreadsheet is enough for the first year, and a spreadsheet you actually fill in beats a CRM you do not.

Ask anyway. "How did you hear about us?" is a weak signal on its own — people misremember, and everyone says Google — but it catches the things tracking cannot, like a neighbour's recommendation.

The one table that answers everything

Once those exist, this fits on a page and settles most arguments:

Numbers like these are made up to show the shape, and the shape is the point. The emergency ad is the business. The map listing is free and doing a third of the work. The Meta ad produced the second-most leads and almost no jobs, which is a filtering problem, not a reason to leave.

You cannot reach any of those conclusions from a monthly total.

What usually turns up first

A channel that produces leads and no work. Almost always the wrong kind of person arriving — a filtering problem, not a volume one.

The map listing doing far more than anyone thought. It is free, it is usually half-finished, and finishing it is the cheapest thing on this list.

A gap between leads and jobs that has nothing to do with advertising. If enquiries arrive and nobody calls them back quickly, every channel looks bad in the same way — and that is a five-minute problem.

Repeat customers hiding inside "new". Worth separating, because a channel that brings people back is worth more per dollar than one that finds strangers.

Two mistakes to avoid

Judging too early. A month of data on a service with a long decision is noise, not a result. Remodelling is the clearest case — a campaign switched off after thirty days for producing nothing was often working, and the jobs were four months out.

Optimising to leads instead of jobs. The cheapest leads are usually the worst ones. If you only measure to the enquiry, you will systematically buy more of what does not close, and the numbers will look like they are improving the whole time.

What to do this week

1. Give every ad its own tag. An afternoon.
2. Add a hidden field to the form that stores those tags. An hour.
3. Put one tracking number on the ads and one on the website. A morning.
4. Start the spreadsheet: enquiry, source, quoted, won, value.
5. Wait a month before drawing conclusions, and then draw them by cost per job.

The first month is uncomfortable, because you will find out that something you were proud of does not work. That discomfort is the entire return on the exercise — and it is what a lead is actually worth to you that tells you what to do about it, and what that means for the budget.