Define what counts
Which actions are genuinely worth money to you, agreed in writing before anything is built.
PPC Strategy · Delhi NCR & remote
Without this, every other number is fiction.
Conversion tracking is the least interesting subject in digital marketing and the one that invalidates everything else when it is wrong. Every optimisation decision downstream inherits its errors.
Short answer
Conversion tracking records which advertising produced which enquiries or sales. It requires the right events defined, tags firing once on the correct action, calls tracked as well as forms, and — where deals close offline — the outcome imported back so platforms optimise towards revenue rather than towards form submissions.
Verified, not assumed
Calls tracked too
Server-side where needed
Offline sales imported
Free tracking check
Get a free quote
Four details. A real reply the same day, from me.
Sure your conversions are real?
Get a Free Quote
WhatsApp
01 The premise
The errors I find most often.
The most common is counting the wrong thing. A conversion set to fire on a page view of the contact page, rather than on a submitted form, means the account is optimising towards people who considered enquiring. The number looks healthy and it is measuring the wrong population entirely.
The second is double counting, usually from the same conversion configured in two places — once directly and once imported from analytics. Every figure is then inflated, cost per conversion looks half what it is, and budget decisions are made on a fantasy. It is common enough that I check for it before anything else.
The third is invisibility. Calls not tracked at all, which for local and service businesses can mean the majority of enquiries are simply missing from the account. Or a sale that completes on the phone weeks later never being fed back, so the platform learns to find cheap form fills rather than customers.
None of this is exotic and all of it is quietly ruinous, because it does not look like a problem. It looks like a working account with disappointing business results, which is precisely the situation in which people increase their budget.
02 The setup
What a correct setup includes.
Ask whoever manages your advertising to confirm each of these specifically.
The setup
- The right event. A submitted form or a completed sale, never a page view of a contact page.
- Firing exactly once. No duplicate tags, no double import from analytics. Deduplication checked rather than assumed.
- Calls tracked. From ads and from the website, with a minimum duration so misdials do not count as leads.
- Values attached where they exist. Revenue for retail, an estimated lead value otherwise, so bidding can optimise on worth rather than count.
- Server-side where it matters. Browser tracking is increasingly blocked. Meta’s Conversions API and server-side GA4 recover a meaningful share of lost signal.
- Offline conversions imported. When the sale closes on a call, feed that back. It is what turns cheap leads into qualified ones over time.
- Consent handled correctly. So tracking behaves properly where consent rules apply, rather than silently failing or ignoring them.
- Verified in a live test. A real submission, checked end to end, appearing exactly once in every platform.
03 By business
What to track, by business type.
The conversion event should match how you actually get paid.
| Business | What should count |
|---|---|
| Local service business | Calls of over thirty seconds, WhatsApp clicks, form submissions and direction requests. Calls usually dominate and are usually untracked. |
| B2B or considered sale | Form submissions and qualified opportunities imported from your CRM. Optimising on raw form fills produces junk at scale. |
| E-commerce | Purchases with actual revenue values, plus add-to-cart as a secondary signal. Values must be accurate or ROAS bidding is meaningless. |
| Education | Enquiry forms, brochure requests, and eventual admissions imported back. The lag is long and the import is what makes it measurable. |
| Travel | Enquiries and WhatsApp conversations, with bookings imported later. Most travel conversions complete off the website entirely. |
Free tracking check
Submit your own form and watch what happens.
That is the test. Give me access and I will do it properly — a real submission traced through GA4, Google Ads and Meta to confirm it appears once, in the right place, attributed correctly.
- A real end-to-end test, not a tag scan
- You keep the findings either way
- Duplicate counting checked specifically
- Same-day reply, all 7 days
Free check
Get a free quote
Four details, and a real reply today.
04 Process
How it gets implemented.
Implement through GTM
One container, documented, so tags are not scattered across the site where a redesign will silently delete them.
Add call and chat tracking
Dynamic numbers, WhatsApp click events, minimum durations. Frequently the biggest single gap.
Connect the platforms
GA4, Google Ads and Meta, with deduplication so the same conversion is not counted twice.
Test and document
A real end-to-end submission checked in every platform, then written down so the next developer does not break it.
— PPC Strategy
Related services.
These sit next to conversion tracking and are usually bought with it. Same person doing the work in each case.
— Questions
conversion tracking — your questions.
Straight answers, including the ones that cost me work. If yours is not here, ask it — I reply the same day.
Submit your own form and follow it through. It should appear exactly once in each platform, attributed to the right source, within the expected window. If it appears twice, not at all, or as a page view rather than a submission, the tracking is wrong.
For most service businesses, yes — calls are frequently the majority of enquiries and are frequently untracked, which makes working campaigns look like failures. Set a minimum duration of thirty to sixty seconds so misdials do not count.
Sending the eventual outcome back to the ad platform — which leads became customers and what they were worth. It is what lets bidding optimise towards revenue rather than towards form submissions, and it is the single biggest lever on lead quality over time.
Increasingly yes, particularly for Meta. Browser-based tracking is blocked by a growing share of browsers and extensions, so a meaningful portion of conversions go unrecorded. Meta’s Conversions API and server-side GA4 recover a substantial part of that.
They attribute differently — different models, different lookback windows, and different definitions of a session. Some divergence is expected and normal. Large unexplained gaps usually mean a configuration problem worth investigating.
It is worth verifying rather than assuming. Tracking is frequently installed correctly and then broken by a later website change, and nobody notices because nothing visibly fails. A live end-to-end test takes ten minutes and settles it.
Free consultation · No obligation
Make the numbers true first.
Give me access and I will run a real end-to-end test — a live submission traced through every platform to confirm it is counted once and attributed correctly.
- A real submission tested, not a tag scan
- Calls and WhatsApp included
- Duplicate counting checked specifically
- You keep the findings either way
Start here
Send your enquiry
Name, number, email, what you need.