Conversion Tracking Setup: What to Track Before Running Ads

Before running ads, track the actions that show real business value, not just interest. For most businesses, that means successful form submissions, confirmed bookings, purchases with accurate order value and, where your systems allow it, qualified leads. Track clicks, form starts and add-to-carts as well, but treat them as diagnostic signals rather than results.

A good conversion tracking setup starts with a decision, not a tag. Once Meta or Google Ads is optimizing toward an event, that event shapes who sees your ads and where your budget goes. A button click and a completed enquiry should not mean the same thing in your reports. If they do, the ad platform cannot tell the difference either.

This guide explains what to measure before you spend, how the main tools fit together and how to test the result. If your ads are already live and getting clicks without enquiries, my article on why Facebook ads get clicks but no leads covers the funnel side of that problem.

What is a conversion tracking setup?

A conversion tracking setup is the system that records actions on your website and connects them to your marketing. It usually has four layers, and they are easy to confuse:

• Event: an observed action, such as a page loading, a button being clicked or a form being opened. On its own, an event only says something happened in the browser.
• GA4 key event: an event you have marked in Google Analytics 4 as important to your business. GA4 used to call these “conversions”. Google renamed them so that “conversion” now refers to what Google Ads counts, as set out in Google’s explanation of key events and conversions.
• Advertising conversion action: an action configured in Google Ads or Meta so the platform can count it and, potentially, bid toward it. In Google Ads, primary conversion actions can be used for bidding, while secondary actions are for observation.
• Qualified lead or sale: a business outcome confirmed in your CRM, booking system or order records. Ad platforms only learn about it if you send that information back.

Keep three ideas separate. Observed actions are what the tag saw. Inferred intent is what you assume the action means. Verified outcomes are what your business records confirm.

Decide what success means before installing tags

Service businesses

For lead form tracking, a successful enquiry is a form your system actually received, not a click on Submit. Above that sit booked appointments and qualified leads: enquiries that match your criteria for location, budget or service fit. Decide which of these your ads should optimize toward, and how you will confirm each one.

Ecommerce stores

In ecommerce conversion tracking, the core outcome is a completed purchase, recorded once, with the correct order value, currency and a unique transaction ID. Product views, add-to-carts and checkout starts are useful steps toward that outcome, not the outcome itself.

Phone-led businesses

A click on a telephone link shows that someone tried to call. It does not show that the call connected, how long it lasted or whether it was a sales enquiry. If calls drive your business, look at call tracking that can record connected or qualified calls, where it is available for your market.

WhatsApp-led businesses

WhatsApp click tracking tells you a link opened the app. It does not prove a message was sent, a conversation happened or the enquiry was genuine. Track the click as intent, then record real conversations and qualified enquiries separately, even if that starts as a simple log your team keeps.

What should you track before running ads?

The role of each action depends on your business model, but as a general guide:

Diagnostic signals

These show where people drop off. They are rarely the right optimization goal when a stronger signal is available.

• Landing page visits: context for everything that follows.
• Form starts: engagement with the form. A large gap between starts and successful submissions points to friction in the form.
• Telephone and WhatsApp link clicks: intent to make contact, not completed contact.
• Product views, add to cart and checkout starts: shopping intent at increasing strength.

Outcomes, or signals closer to them

• Successful form submissions: the enquiry reached your system.
• Confirmed bookings: an appointment exists in your calendar or booking tool.
• Purchases: a completed order with value and currency.
• Qualified leads and closed sales: confirmed in your CRM and, where your integration supports it, sent back to the ad platforms.

If you only take enquiries on WhatsApp, a WhatsApp click may be the best signal available at the start. That can be a reasonable choice, as long as everyone reading the reports knows it counts clicks, not conversations.

A simple event-mapping example

An event map is a written plan of what each event means. It is the foundation of GA4 event tracking and of every platform that reuses those events. In the example below, generate_lead and purchase are GA4 recommended events. GA4’s enhanced measurement can also collect form_start. whatsapp_click and phone_click are custom names used for illustration. Meta and Google Ads use their own names, such as Meta’s Lead and Purchase standard events, so names do not have to match everywhere. Definitions do.

• form_start
  • Business action: someone begins filling in the enquiry form.
  • Trigger: the first meaningful interaction with a form field, fired once per form per page view so repeat typing does not create duplicates.
  • Proves: the visitor engaged with the form.
  • Does not prove: that they intended to submit, or that anything was submitted.
• generate_lead
  • Business action: an enquiry is received.
  • Trigger: the form’s confirmed success response or success message, not the Submit click.
  • Proves: the form was submitted without errors.
  • Does not prove: that the lead is genuine, qualified or new.
• whatsapp_click
  • Business action: someone taps a WhatsApp link or button.
  • Trigger: an outbound click to a WhatsApp URL, such as a wa.me link.
  • Proves: the visitor tried to open WhatsApp.
  • Does not prove: that a message was sent or a conversation took place.
• phone_click
  • Business action: someone taps a phone number.
  • Trigger: a click on a tel: link.
  • Proves: the visitor tried to start a call.
  • Does not prove: that the call connected or was a sales enquiry.
• purchase
  • Business action: an order is completed.
  • Trigger: the confirmed order, sent with a transaction ID, order value and currency.
  • Proves: an order was placed.
  • Does not prove: that the order was not later cancelled or refunded.

How the tools fit together

• Google Tag Manager is a container for deploying and managing tags. It does not report anything itself. A Google Tag Manager setup is optional: GA4 can also be installed with the Google tag directly or through a platform integration.
• GA4 records events and key events and shows how all channels perform, including organic search and email.
• Google Ads conversion tracking counts conversion actions for Google Ads reporting and bidding. If Google Ads management is part of your plan, you have two common routes: create a conversion from a GA4 key event, or use the Google Ads conversion tag directly. Conversions created from Google Analytics start as secondary and can only be made primary inside Google Ads. Either route can work. The risk is setting both as primary for the same outcome, which can count one enquiry or sale twice for bidding. Choose one primary source per outcome and keep any overlap secondary.
• Meta Pixel sends events from the visitor’s browser to Meta.
• Meta Conversions API sends events from your server, platform or CRM to Meta. For Meta Ads conversion tracking that uses both, Meta’s deduplication guidance matches a browser and server event when the Pixel’s eventID equals the server’s event_id and the event names match, for example Lead and Lead. Checking this is one of the first steps in my Meta Ads campaign management, because unmatched events can double-count leads.
• CRM or order records remain the reference for what actually happened.

Not every business needs every tool. A small service business might start with GA4, the Google Ads tag and the Meta Pixel, then add the Conversions API or CRM imports as spend and complexity grow.

Consent and data handling

• Apply consent controls that suit your audience and the requirements that apply to you. Take legal advice if you are unsure what those are.
• Google’s consent mode adjusts how Google tags behave based on a visitor’s choices. It does not provide a consent banner, and it is not a substitute for obtaining consent where consent is required.
• Do not send names, email addresses, phone numbers or other personal information into GA4 event parameters, page URLs or page titles. Google’s guidance on avoiding personally identifiable information covers common leaks, including URLs.
• Some platform features, such as Google’s enhanced conversions or Meta’s customer information parameters, are designed to receive hashed customer data under their own terms. Treat these as deliberate, separate configurations, not as part of ordinary analytics collection.
• Server-side tracking changes where events are sent from. It does not remove consent obligations.

Common mistakes before launching ads

• Counting clicks on Submit, Call or WhatsApp as completed leads.
• Duplicate tags, such as a hard-coded tag and a GTM tag sending the same event.
• Events firing on failed submissions, such as validation errors.
• Thank-you page refreshes or revisits recording extra leads.
• Purchases sent without value or currency.
• Missing transaction IDs, which make duplicate purchases harder to catch.
• Internal tests counted as genuine leads.
• Missing cross-domain measurement when checkout or booking runs on another domain, which can split one visit into two and credit the result to the wrong source.
• Optimizing toward a weak proxy, such as landing page views, without accepting the tradeoff: more of the proxy, not necessarily more customers.
• No CRM feedback on lead quality, so campaigns keep finding cheap leads that never buy.

How to test your conversion tracking setup

1. Create an event map listing each event, its trigger and its meaning.
2. Confirm trigger conditions with whoever built the form, booking tool or checkout.
3. Test in GTM Preview or Tag Assistant, where applicable, to see which tags fire and when.
4. Inspect events and parameters in GA4 DebugView.
5. Use the Test Events tool in Meta Events Manager to check browser and server events.
6. Check conversion status and diagnostics in Google Ads.
7. Submit failed and successful forms, and confirm only successful ones count.
8. Refresh confirmation pages and repeat actions to look for duplicates.
9. Place test orders and check value, currency and transaction ID.
10. Compare test events with real form entries, bookings, CRM records or orders.
11. Test with consent accepted and declined. DebugView will not show events when consent mode is active and analytics consent has not been given, so plan for both states.
12. Repeat key tests on a phone, including links opened inside the Facebook and Instagram apps.

Seeing a tag fire only shows that the tag ran. It does not show that the trigger was right, the data was correct or the platform accepted it. Standard reports also take time to process, so allow for that delay before comparing totals.

Why GA4, Google Ads and Meta report different numbers

• Attribution: each platform credits conversions using its own rules and touchpoints.
• Conversion windows: platforms count conversions over different periods after a click or view.
• Time zones: reporting days may start at different times in each account.
• Consent and blocking: some activity is never recorded, and some platforms model what is missing.
• Event definitions: one tool may count every submission while another counts one per person.

Expect differences. The aim is to understand them and agree which figure you use for which decision, not to make every platform match.

A practical pre-launch checklist

• Success is defined in writing for each campaign.
• The event map is complete and agreed.
• The primary conversion is an outcome, not a click.
• Each outcome has only one primary conversion source in Google Ads.
• Pixel and Conversions API events are deduplicated, if you use both.
• Purchases include value, currency and transaction ID.
• No personal data appears in URLs or GA4 parameters.
• Consent behaviour has been tested.
• Failed submissions and page refreshes do not count as leads.
• Test leads are excluded or clearly labelled.
• Someone compares reported leads with CRM or order records on a regular schedule.

Frequently asked questions

What should I track before running ads?

Track the outcome that matters most to your business, such as successful enquiries, bookings or purchases, plus the steps leading to it. Add diagnostic events like form starts, call clicks and add-to-carts so you can see where people drop off. Where possible, connect qualified leads or sales from your CRM.

Is a WhatsApp click a conversion?

It can be configured as one, but it only shows that someone tried to open WhatsApp. It does not prove a message was sent or the enquiry was genuine. Treat it as an intent signal, and record actual conversations and qualified enquiries separately.

Should a form start be a primary conversion?

Usually not. A form start can fire on an accidental tap and says nothing about whether the form was completed. It works better as a diagnostic signal for spotting form friction. Use a confirmed submission, or a qualified lead where available, as the primary conversion.

Do I need GTM to use GA4?

No. GA4 can be installed with the Google tag directly or through many website platforms’ built-in integrations. Google Tag Manager becomes useful when you manage several tags, need custom triggers such as confirmed form submissions, or want to change tracking without editing site code each time.

Why do GA4 and Meta report different lead numbers?

They use different attribution rules, conversion windows and event definitions. Meta can credit conversions after an ad view, while GA4 attributes based on site visits. Consent choices and blocked scripts affect each tool differently. Compare both with your actual form or CRM records rather than expecting them to match.

How do I know my conversion tracking works?

Test each event in GTM Preview, GA4 DebugView and Meta’s Test Events tool, then compare the results with real records. Submit failed and successful forms, refresh confirmation pages and place test orders. If reported conversions closely reflect what your business actually received, the setup is doing its job.

Talk through your conversion tracking setup

If you are preparing to run ads and want a second opinion on your conversion tracking setup, I can help. My tracking and analytics services cover GA4, Google Tag Manager, Google Ads conversions and the Meta Conversions API. Get in touch through my contact page and tell me which tools you use, how enquiries or orders reach you, and which numbers you do not trust. I’ll reply with what I would check first.