How Do I Know If My Leads Are Being Tracked Properly?

Do this on your phone, on mobile data rather than office wi-fi, because that is how most of your customers arrive and it is the environment where things break.

1. Open your site the way a stranger would — a search for what you sell, not a bookmark.
2. Fill in the enquiry form using a name you will recognise instantly. Something like Test Wednesday Morning works better than Test, which you will later find seventeen of.
3. Submit it, and note the exact time to the minute.
4. Then, separately, tap the phone number and let it ring through to whoever answers.
5. Wait an hour, and go looking for both.

The hour matters. Some systems report within seconds and some batch, so an immediate check produces a false alarm about half the time. Google Analytics has a DebugView that shows events arriving live if you want the instant confirmation, but the hour-later check is the one that tells you what the reports will actually say tomorrow.

Two enquiries — one form, one call — is the minimum useful test, because they travel through completely different plumbing and fail in completely different ways.

One enquiry, four systems, one appearance in each. Anything else is a finding.

That fourth row is the one businesses skip, and it is the one that decides whether any of this pays. Without a place to record the outcome, every conversation about lead quality is two people trading impressions. With it, the question becomes arithmetic. That loop is the whole substance of fixing lead quality — the acquisition side rarely changes much.

Most Australian service businesses take the majority of their real work over the phone, and the phone is the least-instrumented part of nearly every setup we are asked to look at.

A tapped number on a mobile can be recorded as an event. A number that somebody reads off the screen and dials manually cannot. A number that has been changed on one page and not another produces two different histories. And a number displayed as an image, which still happens, is invisible to everything.

The practical consequence is a systematic bias: form enquiries look cheap and calls look free, so the parts of your marketing that generate calls appear to produce nothing. Businesses then defund the channel that was actually working, and the report agrees with them the whole way down.

If the phone is how you get paid and the phone is not counted, every number in your reporting is describing a minority of your business.

You do not need to solve this perfectly. Getting from nothing to a rough, consistent count of calls is worth more than getting from a good count to a perfect one, because the first step changes which channel you believe in.

Under-counting makes performance look worse than it is, and somebody usually notices. Over-counting makes it look better, and nobody investigates good news.

The usual causes are dull and common. The same event fires on the form submission and again on the thank-you page. A tag was added twice, once directly and once through a manager container. A page that people reload — a confirmation screen with a phone number on it — records a fresh conversion every time somebody goes back to check the number. Google Ads also has a documented setting controlling whether every conversion is counted or only one per interaction, and the default is not right for every business.

The damage is not only that the report flatters you. Automated bidding optimises toward the outcome it is shown, so a duplicated event teaches the system that a particular kind of visit is twice as valuable as it is, and the budget follows. You end up paying more for the traffic that double-counts best.

The test enquiry catches all of this in one pass. If your single submission produced two conversions, stop and fix that before you touch anything else. This is usually a container problem rather than an advertising problem: check how the key event is defined in your analytics before you change anything in the ad account, and read how one tag manages to break a whole quarter of reporting before you go looking in the wrong system.

Measurement is a legal obligation as well as a technical one. The Office of the Australian Information Commissioner sets out the Australian Privacy Principles, and APP 5 covers what a person must be told when you collect their personal information — who you are, why you are collecting it, and who else will receive it.

For most small businesses this is not onerous. It means a privacy notice that a person can actually reach from the form, written in language that describes what you genuinely do rather than a template describing something else. Where it gets more involved is when enquiry data is passed to advertising platforms to improve targeting, because that is a disclosure and it needs to be covered by what you have told people.

Two practical notes. First, the obligation belongs to the business, not to whoever built the form, so "the agency set that up" is not an answer. Second, consent tooling frequently breaks tracking in ways nobody notices for weeks — a banner that blocks the measurement script also blocks the count, and the enquiry graph falls off a cliff on the day it was installed. Both things being true at once is normal. Check the OAIC's own material rather than relying on a plugin's description of it.

Work from the failure that costs the most and travels the least distance.

1. Enquiries not arriving. A form that collects and does not deliver is losing you jobs today. Nothing else on this list competes.
2. Double counting. It corrupts every decision downstream, including the automated ones, and it is usually a single change.
3. Uncounted calls. Fixing this changes which channel you believe in, which is worth more than another decimal place anywhere else.
4. Outcomes not recorded. Slower to set up and the highest ceiling, because it is what lets you manage to booked work instead of to form fills.

Then put the test in the calendar. Once a quarter, and after any change to the website, the forms or the phone system, run the same two enquiries and check the same four places. It takes fifteen minutes and it is the only routine that reliably catches a measurement break before it has been governing your budget for a month.

It is also worth knowing what a break looks like from the outside: a sudden step change on a single day, holding at the new level. That shape almost never means the market moved. It is the same failure that sits behind most cases of a cost per lead that has started climbing, and it is the cheapest explanation to rule out. If nobody owns the measurement setup, it will happen again.

How often should I test my tracking?

Once a quarter, and after any change to the site, the forms, the phone system or the consent banner. Those four events cause the overwhelming majority of breaks, and a fifteen-minute test straight afterwards catches the problem before it has spent a month shaping your budget.

My analytics and my ad platform report different numbers. Which is right?

Both, usually. They attribute to different things and use different windows, so they are answering slightly different questions. What matters is that each one is internally consistent over time. Investigate when the gap between them changes, not when a gap exists.

Do I need call tracking?

If most of your work arrives by phone, yes — otherwise your reporting describes the minority of your business that fills in forms. A rough consistent count is far more useful than no count, and the step from nothing to something changes decisions in a way further precision does not.

Is any of this a privacy problem?

It can be. The OAIC's Australian Privacy Principles require you to tell people what you collect and why, and APP 5 covers the notice itself. Passing enquiry data to an advertising platform is a disclosure, so it has to be covered by what you have told people. The obligation sits with the business, not the supplier who configured it.