Conversion Tracking Is the Part Everyone Gets Wrong

Automation makes bad tracking expensive

There was a period when broken tracking was merely embarrassing. You reported the wrong number, someone eventually noticed, you fixed it.

That is no longer the case, because bidding is now largely automated and automation optimises towards whatever you tell it to count. If the thing being counted is not the thing you want, the system will diligently pursue the wrong outcome and get better at it over time.

An account counting every form submission as a conversion, including the spam and the job applications, will learn to find more of that traffic. It is not malfunctioning. It is doing exactly what it was told, using a definition nobody examined.

The failures worth checking for

Most tracking problems are one of a handful of specific faults, and they are all findable in an afternoon.

Counting the wrong event. A page view on a contact page rather than a submitted form. A button click rather than a completed action. These inflate the numbers and mislead the bidding.

Duplicate counting. The same conversion recorded by more than one tag, or by both the platform and the analytics import. Results look roughly twice as good as they are.

Everything weighted equally. A newsletter signup counted the same as a quote request. The system will chase the cheap one because it is easier to produce.

Nothing recorded for phone calls. For service businesses this is often the majority of enquiries, invisible in the data and therefore invisible to the bidding.

Broken by a website change. Tracking implemented correctly, then silently broken when the form or the confirmation page changed. Nobody tests it afterwards because nobody owns it.

Offline outcomes missing. Leads recorded, but no signal about which became customers. The account optimises for lead volume with no idea which leads were any good.

Decide what a conversion actually is

Before touching any settings, answer the question properly: what event, if it happened more often, would make the business more money?

For most businesses that is a qualified enquiry, not a form submission. The difference matters, because form submissions include the timewasters and the qualified enquiries do not.

Getting closer to that distinction is possible without heroic engineering. Screening at the point of enquiry, through better form logic or a qualifying step before submission, separates the real enquiries from the rest before they enter the count.

Where you cannot fully distinguish them in the tracking, you can at least weight them. A quote request and a brochure download should not carry the same value, and assigning different values changes what the bidding pursues.

Feed the outcome back

The single highest‑value improvement available to most accounts is telling the platform which leads became customers.

The mechanism exists. Lead data is captured, matched against the sales outcome in your own system, and imported back. From then on, the bidding is working towards revenue rather than towards enquiry volume.

It requires the enquiry and the sale to be connected somewhere, which is usually a process problem rather than a technical one. Sales teams that do not record the source of a closed deal make this impossible, and no amount of tracking configuration compensates for it.

Businesses that do this properly usually find the picture rearranges. Campaigns that looked efficient on cost per lead turn out to produce leads that rarely close, and campaigns that looked expensive turn out to bring in the customers worth having.

Test it, then keep testing it

Tracking is not a configuration you set. It is infrastructure that decays, usually silently, usually after a website change.

Submit your own forms and confirm the conversion appears where it should. Do it after every meaningful change to the site. Compare the platform's conversion count against your own records monthly, and treat a divergence as a fault rather than a rounding difference.

Pay attention to the pages the conversions happen on, too. Tracking will faithfully report a low conversion rate without telling you the page is the reason, and the fundamentals of a page built to convert are well covered in most guidance on social media landing page design, which applies equally to paid search traffic.

The same applies to any channel where the conversion happens off your site. Anyone running YouTube lead generation confronts this directly, because the action and the outcome sit in different systems and connecting them is deliberate work rather than a default.

Where this leaves you

Auditing conversion tracking is unglamorous and nobody asks for it. It is also the highest‑return work available in most accounts, because every other improvement is multiplied or nullified by whether the measurement underneath is sound.

If you have not verified yours in the last six months, assume it is wrong somewhere. Check what is being counted, check for duplicates, check whether calls are captured, and check whether anything tells you which leads became customers.

For accounts where the spend is material, this is usually the first thing capable Google Ads specialists such as SIXGUN look at, ahead of keywords, bids or creative. Not because it is more interesting, but because there is no point improving a system that is aiming at the wrong target.