Craig Campbell SEO Training & Consultancy

Tutorials

Google Ads Conversion Tracking and GA4 Guide

By Craig Campbell · 18 Nov 2026

11 min read

Google Ads Conversion Tracking and GA4 Guide

Conversion tracking is where advertising reports either become useful or start misleading you. If the wrong action counts as success, a campaign can look healthy while the business receives little of value. I want tracking that answers a straightforward question: which activity helps generate the outcomes this business actually needs?

In this tutorial I will explain how I plan measurement across Google Ads and Google Analytics 4, choose an implementation and test it properly. I will use a service enquiry as an illustrative example. The same principles apply to ecommerce, although purchase values, refunds and transaction identifiers introduce additional requirements.

Quick answer: how should Ads and GA4 tracking work?

Define a genuine business outcome, implement an event that fires only when it happens, then decide how that outcome becomes a Google Ads conversion action. You can use a suitable Google Ads website conversion implementation or import an eligible GA4 key event through the linked accounts. Do not count both as separate primary successes for the same enquiry.

GA4 uses events to describe activity and key events to identify important activity. Google Ads conversion actions have their own settings and campaign use. Linking the products does not make every event suitable for bidding, and it does not guarantee the two interfaces will show identical totals.

  • Write down the outcome and its exact trigger.
  • Choose a primary source for measuring that outcome in Ads.
  • Implement consent-aware tags and appropriate data handling.
  • Test successful, failed and repeated journeys.
  • Compare reports with an understanding of attribution and timing.

Step 1: write a measurement plan

Begin with the business, not the tag manager. For a service business, distinguish an enquiry submitted, an enquiry accepted as qualified and a customer won. These are different stages. You may initially measure the first stage online and add later stages from a customer relationship management system when the business can record them consistently.

For each event, document its purpose, trigger, data source, owner and intended reporting use. A useful definition might say that a lead event occurs after the server accepts a valid enquiry and the visitor receives confirmation. That is much clearer than asking a developer to track the contact form without specifying what success means.

Separate diagnostic interactions from business outcomes. Telephone-link clicks, file downloads and form starts can help explain a journey, but they should not automatically compete with verified enquiries as primary bidding signals. Otherwise a campaign may appear to improve because it generates more of the easiest interaction rather than more useful demand.

For ecommerce, document how currency, value and transaction identifiers are generated. Decide whether revenue includes tax or delivery and keep the definition consistent. Use actual order information rather than a fixed value copied into every purchase event. Plan for cancellations and refunds where the available integrations support them.

Step 2: choose your measurement architecture

A Google tag implementation, Google Tag Manager and supported platform integrations can all form part of a sensible setup. The right choice depends on your website, consent system and ability to maintain it. Installing several overlapping methods without checking what each sends is a common source of duplicated events.

ApproachGood reason to use itWhat to check
Google Ads website conversionDirect measurement of an Ads outcomeCorrect destination, trigger and conversion settings
GA4 key event imported into AdsUse a defined Analytics outcome in advertisingAccount linking, eligibility and duplicate actions
Offline outcome importMeasure qualification or sales after the website visitMatching data, consent, timestamps and upload rules

I normally nominate one implementation owner. That person needs to know which tags are installed in the website theme, which come from plugins and which are managed through a container. Without that inventory, somebody can add a second installation while trying to repair the first.

Keep production and testing arrangements clear. If a staging site sends events into the live property, test enquiries can contaminate reports. Use suitable test configurations and filters with care, and document their limitations. Do not permanently exclude real customers because an internal-traffic rule was broader than intended.

Step 3: check accounts, access and linking

Confirm you are working in the correct Google Ads account, GA4 property and web data stream. Similar account names are not proof. Record identifiers in your implementation notes so the business can verify ownership later. Use appropriate permissions rather than sharing a single personal login among staff and suppliers.

Link the intended GA4 property and Google Ads account with the required permissions. Review the current options for advertising features and data sharing in line with the organisation's consent and privacy position. Interfaces change, so use the current product labels and help documentation rather than following an old sequence of menu clicks blindly.

Review auto-tagging and ensure redirects preserve the information needed for attribution. Test your actual advert destination, including any tracking templates and intermediate redirects. A working page URL is not enough if a redirect strips attribution parameters before the measurement system has a chance to use them.

If your journey crosses domains, such as a main website and a separate booking service, investigate supported cross-domain measurement. Unwanted self-referrals and broken sessions can make acquisition reporting confusing. Also check whether the external service lets you implement the event at the actual completion point rather than just tracking the outbound click.

Step 4: implement a reliable success event

For the illustrative enquiry form, I prefer an explicit success signal from the application or its supported integration. A button click is weak evidence: validation can fail, the connection can drop or the server can reject the submission. Ask the developer what happens after the system has genuinely accepted the enquiry.

A dedicated confirmation page can be useful, but consider direct visits, refreshes and repeat navigation. If the event fires on every page view, one enquiry may become several recorded conversions. A successful form event with suitable state or an identifier may be more reliable, depending on the website.

Use stable, documented event names. Avoid creating several near-identical versions because different people capitalised the name differently or used a new plugin. Where an appropriate recommended GA4 event fits the action, follow its specification and parameters. Keep implementation instructions precise enough that the next developer can reproduce them.

Do not put names, email addresses, telephone numbers or other prohibited personal information into ordinary Analytics event parameters or page URLs. Inspect what the form puts into the address bar and data layer. A thank-you URL containing submitted contact details is both a privacy problem and a potential analytics data-policy problem.

For purchases, send a stable unique transaction identifier and verify that the implementation handles reloads and retries correctly. An identifier is not a licence to ignore trigger quality; deduplication behaviour depends on the destination and method. Test the actual setup rather than assuming every platform handles repeated events identically.

Your measurement design must fit the laws and requirements relevant to the business and visitors. For a UK business, that includes considering UK GDPR and PECR. I would involve a competent privacy adviser where needed rather than presenting a tag configuration as legal advice.

A consent management platform should communicate the visitor's choices to the relevant tags. Consent Mode can adjust tag behaviour according to consent signals, but it is not a consent banner and does not obtain permission by itself. Check that the default state and later updates happen in the intended order.

Test accepted, rejected and changed consent choices. Look for tags that fire before the consent system initialises and plugins that bypass the intended controls. If some measurement is unavailable because a visitor declines consent, acknowledge that limitation rather than quietly finding an alternative route to collect the same restricted information.

Enhanced conversions use appropriately handled first-party customer data to improve measurement in supported circumstances. Follow Google's current requirements, supported collection methods and applicable consent rules. Hashing is not the same as anonymisation, and these features do not justify collecting extra customer information without a valid purpose.

Step 6: configure Google Ads conversion actions

Once events are working, create or import the relevant conversion action in Google Ads. Confirm the source, category and intended outcome. Review whether the action is primary or secondary and how it is included in campaign goals. A secondary action can still appear in broader reporting without necessarily being a normal bidding objective.

Check campaign-specific goals as well as account defaults. A correct conversion action can still influence the wrong campaign if the goal configuration is misunderstood. Some goal arrangements behave differently, so inspect the actual campaign settings rather than assuming one account-level label tells the whole story.

Choose counting settings that suit the outcome. For many lead actions, one conversion per ad interaction is more sensible than counting repeated submissions as extra commercial value. Purchases often need every valid transaction. These settings do not replace event-level quality checks, and they do not make duplicate implementations harmless.

Set values deliberately. If you use an estimated lead value, document the assumptions and keep it separate from recorded revenue. Review attribution models and conversion windows against the buying cycle and available options. My Google Ads campaign setup guide explains how these goals connect to campaign decisions.

Step 7: test the complete journey

Use the preview and debugging tools available for your implementation, such as Tag Assistant and GA4 DebugView. Start by checking the event itself: correct name, correct parameters and one trigger at the intended moment. Then check that it is sent to the correct measurement destination.

Run a successful submission, an invalid form submission and a repeat visit. Test mobile and desktop, plus any materially different form versions. If customers can book from several landing pages, test each integration rather than assuming every form shares the same underlying code.

Check consent states, redirects and browser behaviour. Browser extensions and privacy controls can affect collection, so a single blocked test does not prove the entire setup is broken. Equally, a successful test in an unrestricted browser does not prove the consent implementation behaves properly for everyone.

Keep a test record with approximate time, device, event and expected result. Reconcile that record with the receiving systems after their normal processing time. Real-time and debug views are useful implementation checks, but they do not guarantee that standard attribution reports will immediately show the same event in the same way.

Step 8: understand reporting differences

Google Ads and GA4 can differ because they answer different reporting questions. Attribution models, eligible channels, date assignment, time zones, identity, consent and conversion windows all matter. Investigate those definitions before assuming that the smaller number is broken or the larger number is correct.

Ads reporting can attribute a conversion to an earlier advertising interaction, while another report may organise information around when the event occurred. Compare equivalent date bases where the reporting options allow it. Also allow for processing delays and conversions that occur after the initial reporting period.

Keep the website's actual lead or order records as an independent operational reference. They will not necessarily match attributable advertising conversions, but a sudden event spike without additional real enquiries is a warning. Reconciliation should expose measurement faults, not force every system to display an artificially identical number.

My Google Analytics setup guide covers the broader reporting foundation. I use Analytics to understand acquisition and behaviour, while keeping commercial decisions grounded in verified business outcomes and the known limitations of each measurement source.

Step 9: add qualified leads and sales feedback

Once basic enquiry tracking is reliable, connect later outcomes where practical. A CRM can record whether a lead was valid, qualified, won or lost. Use clear stage definitions so one salesperson's qualified lead means roughly the same thing as another's. Better stage discipline matters more than a complicated dashboard.

Choose a supported offline conversion or enhanced-conversions-for-leads workflow and follow its current requirements. Matching may use eligible click information, appropriately collected customer data or supported integrations. Check upload timing, consent, duplicate handling and conversion-action configuration before relying on the resulting reports.

Do not send fabricated values simply to make value-based bidding available. If values are estimates, make the estimation method explicit and review it against sales. The aim is to provide a more useful business signal, not to turn uncertain figures into precise-looking numbers. My PPC planning guide puts that measurement into context.

Keep a short measurement change log alongside the campaign notes. Record who changed a trigger, consent configuration or conversion goal, when the change went live and which test confirmed it worked. This makes later reporting discrepancies much easier to investigate.

Conversion tracking checklist

  1. Define each business outcome and its success trigger.
  2. Inventory tags, plugins, integrations and account ownership.
  3. Confirm account linking and attribution parameter handling.
  4. Check consent behaviour before and after a visitor's choice.
  5. Test success, failure, refresh and repeat submission journeys.
  6. Review primary actions, campaign goals, values and counting settings.
  7. Compare test records with Ads, GA4 and real enquiries.
  8. Assign an owner to recheck tracking after website changes.

Common tracking mistakes

The most damaging mistakes are counting a button click as a lead, importing the same outcome twice and making every engagement event primary. Another is changing event definitions without recording when they changed. A rising conversion rate means little if last month measured completed enquiries and this month measures form starts.

I also see tracking treated as a one-off installation. Forms, consent tools, booking providers and website themes change. Add measurement checks to the release process and investigate sudden discontinuities. Do not let automated bidding run indefinitely against a broken signal just because the campaign itself still serves adverts.

Frequently asked questions

Do I need both GA4 and Google Ads tracking?

You may use both products for different purposes, but you do not need duplicate primary conversion actions for the same outcome. Decide which source supplies the Ads conversion, then keep other measurements clearly labelled for analysis or comparison.

Why does GA4 show a lead that Ads does not?

The lead may not be attributable to an eligible advertising interaction. Consent, attribution windows, processing and configuration can also matter. Check the actual event and reporting definitions before reinstalling tags. A real website lead is not automatically an Ads conversion.

Can I track conversions without a thank-you page?

Yes. A reliable application success event or supported form integration can trigger measurement. What matters is that the event represents an accepted submission. A click listener alone cannot reliably distinguish a completed enquiry from a validation error.

Should every conversion have a monetary value?

No. Use a defensible value when it improves decision-making. Actual purchase revenue and estimated lead value are different concepts. If you cannot support an estimate, focus on accurate outcome counts and sales feedback before creating artificial precision.

What should I do after a tracking fault?

Fix and retest the implementation, document the affected dates and assess whether campaign decisions relied on corrupted data. Review the platform's available remediation tools and their eligibility. If you need help assessing the setup, contact me with the measurement problem and business context.

Let's work together.

Whether you need an SEO partner, a consultant, a speaker for your event or training for your team — get in touch and let's chat.

Get in touch
Get in touch