Conversion tracking with GA4 and the Meta pixel: what to measure, how to set it up and how to check it

Why conversion tracking decides how well your ads perform

Advertising platforms cannot see your business. They only see the signals you send them: a purchase event, a quote request, a submitted form. Google’s and Meta’s algorithms learn from those signals and look for people who resemble the ones who converted. If the signal is wrong, the optimisation heads in the wrong direction: your budget is spent efficiently, but on the wrong goal.

Picture a plumbing company that optimises its ads for “contact page views”. The algorithm will find people who open that page, not people who pick up the phone. The report shows excellent traffic while the phone stays silent. The problem is not the budget or the creative, but the event you asked the platform to chase.

Good tracking has a quieter benefit as well: it gives the whole team a shared version of the truth. When marketing, sales and management look at the same numbers, conversations about what works get shorter. And if you work with an agency, you can check independently what you are being told. For the wider foundations of measurement, see our GA4 guide for small business owners.

What counts as a conversion: decide before you install anything

Before any code goes live, answer one question in writing: which actions on your website matter to the business? A brochure site usually has three: a submitted form, a click on the phone number and a click on the WhatsApp or email button. An online shop has a longer chain: product view, add to basket, start of checkout, purchase.

Separate two kinds of events:

• Primary events (conversions): those that map directly to revenue or a qualified lead. In GA4 you mark them as key events (older interfaces called them “conversions”). There are few of them and they rarely change.
• Supporting events (micro-conversions): add to basket, scrolling 75% of a page, a brochure download, starting a form. You use them for diagnosis and remarketing audiences, not to optimise campaigns.

A practical rule: if you cannot explain in one sentence what an event is worth to the company, do not mark it as a conversion. The classic offender is “click on any button”, which inflates the numbers while saying nothing about sales.

How GA4, the Meta pixel and Google Tag Manager fit together

Three components come up in every conversation, and confusing them causes many of the errors we see:

• GA4 (Google Analytics 4) is your analysis tool. It measures behaviour on the site, traffic sources and events, whether or not they come from ads.
• The Meta pixel is a snippet of code that sends site events to Meta, so you can measure conversions from Facebook and Instagram and build audiences. Meta also offers the Conversions API, which sends the same events straight from your server. It is a useful complement when browsers or extensions block scripts.
• Google Tag Manager (GTM) is a container where you manage all your tags from one place, without editing the website code every time.

The setup we recommend for most businesses is a single GTM container on the site, from which the GA4 configuration tag, the GA4 events and the Meta pixel events all fire. The benefits are a change history, a preview mode and one control point for consent. The drawback is some time spent learning the tool, which usually saves many hours of fixes later.

Which events to measure, by type of business

The table below is a starting point, not a fixed recipe. The recommended Google event names and Meta’s standard events exist so the platforms recognise them automatically, so keep them exactly as they appear in the documentation.

For shops, always include the value, the currency and a unique transaction ID in the purchase event. Without them you cannot calculate the revenue attributed to ads, and you cannot prevent duplicate orders being counted.

How to set up tracking, step by step

Order matters. Many people install pixels first and only then wonder what they want to measure. The process below avoids rework.

1. List the important actions and choose two to five primary events, as explained above.
2. Create the GTM container and install it once, across the whole site.
3. Set up GA4: the property, the web data stream and the configuration tag inside GTM.
4. Create the events with clear triggers: loading the thank-you page, submitting the form, clicking a tel: link. For shops, use a data layer, which the ecommerce plugin usually provides.
5. Mark the primary events as key events in GA4.
6. Add the Meta pixel from GTM and map the same actions to Meta’s standard events. If you use the Conversions API, send an event ID as well, so that duplicates can be removed.
7. Import or recreate the conversions in Google Ads, if you run campaigns there.
8. Test everything, then publish the container and note the date in a change log.

If your campaign structure looks fine but you cannot tell where leads come from, add consistent UTM parameters too. You can build them quickly with our UTM builder, and the naming conventions are explained in the guide to UTM parameters.

How to check that measurement actually works

A tag that “looks installed” is not a tag that measures correctly. Verification happens on three levels.

1. The live test

In GTM, use preview mode: browse the site, fill in the test form and watch which tags fire. In GA4, DebugView shows events as they arrive, with all their parameters. In Meta, the test events section of Events Manager shows what the pixel or the Conversions API is receiving.

2. The sanity test

Place two or three test orders or enquiries and check that each arrives once in GA4, in Meta and in your internal system. An event that appears twice is a classic fault, and so is one that never appears because the thank-you page reloads.

3. Regular reconciliation

Every month, compare the conversions in the platforms with those in your CRM, shop or booking diary. Do not expect identical figures: differences come from consent, blocked scripts and different attribution models. What matters is that the gap stays within a stable, explainable range. If one platform suddenly reports double or half, something has broken.

Consent and GDPR: measuring without breaking the rules

In the European Union, cookies and identifiers used for analytics and advertising generally require the visitor’s prior consent. In practice that means a consent banner that blocks marketing tags until the visitor accepts, not one that merely informs them.

Google offers Consent Mode, through which tags adapt their behaviour to the user’s choice, and for traffic from the European Economic Area consent signals are increasingly important for features such as audiences and conversion measurement in Google Ads. Meta has its own consent settings for the pixel. The details change, so check each platform’s current documentation.

The practical consequence: some visitors will decline, so the figures in the interfaces will be lower than reality. That is not a tracking defect; it is the result of lawful measurement. For banners, forms and policies, see our article on GDPR and cookies on your website.

The above is general information and not legal advice. For the exact setup of your own website, consult a lawyer or a data protection officer (DPO).

Common mistakes that skew optimisation

• Duplicate events. The pixel is installed in the theme, in GTM and through a plugin. The result is two purchases for every order and an inflated ROAS.
• Unprotected thank-you page triggers. Anyone who refreshes the page or returns from their browser history generates a false conversion.
• Too many “primary” events. If you mark ten actions as conversions, optimisation gets diluted and you no longer know what actually brought in money.
• Missing or wrong value. Without the correct value, strategies based on revenue cannot work.
• Payment provider not handled as an exception. When the customer leaves for the payment provider and comes back, the session can be credited to the provider instead of the original ad. GA4 settings let you exclude such domains from referral sources.
• Testing only from your own computer. Check on a phone and in another browser too, and with an ad blocker on if you can, to see what you lose.
• No documentation. Nobody knows which event does what, and after a redesign the tags disappear without anyone noticing.

Key terms

Event: any measured interaction, from a page view to a purchase. Key event (conversion): the event you declare important to the business. Attribution: the rule by which a conversion is credited to a channel or campaign. Deduplication: removing events sent twice, for instance from both the browser and the server. Data layer: an object on the site where structured information, such as basket contents, is stored for GTM to read. UTM: parameters added to links to mark the source, medium and campaign.

How we approach it

We start with a short meeting to agree what matters to the business, then review what is already in place: duplicate containers, old pixels, events without values. Only then do we implement or repair, with a simple document that lists each event, its trigger and its destination.

Tracking is part of our conversion rate optimisation work, because without trustworthy data every test or campaign is guesswork. If you would like an outside view of where you stand, we can talk, with no obligation, through our contact page.

In short

• Measure few events, but tie them to money or genuine leads.
• Use Google Tag Manager as the single control point for GA4 and the Meta pixel.
• For shops, send the value, the currency and a transaction ID.
• Test in DebugView, in Events Manager and with real test orders.
• Respect consent, and accept that the numbers will be lower than real traffic.
• Reconcile platform data with sales data every month.