Tracking conversions on a landing page: setting up Google Analytics 4 without drowning in data

A landing page running with no conversion tracking is like a vendor staffing a stall all day without ever counting what they sell: traffic arrives, visitors leave, and nobody knows how many actually filled in the form, clicked "Book a call", or bought. Without that measurement, every decision — changing a headline, moving a button, testing a new offer — rests on a hunch rather than a number. Google Analytics 4 (GA4), free and now the standard since Universal Analytics was sunset, lets you track those conversions; you still need to know what to track, how to set it up correctly, and how far to trust the numbers it shows. Here's how to set up conversion tracking that's actually useful on a landing page, without turning GA4 into an unmanageable mess.

Why conversion tracking is the invisible prerequisite for every optimization

Testing a new headline, comparing two versions of a CTA, deciding whether a premium offer deserves its own page: every optimization covered in our A/B testing guide assumes you can already count conversions reliably — without that, an A/B test just compares two made-up sets of numbers. A 2011 study by researchers Erik Brynjolfsson, Lorin Hitt and Heekyung Kim, surveying 179 large publicly traded companies, measured what data-driven decision-making actually delivers: firms that adopt it show productivity 5 to 6% higher than what their technology investments alone would predict, with a similar effect on asset returns and market value (study). At landing-page scale, the same principle applies: it isn't traffic volume that separates a page that's improving from one that's stalled, it's the ability to know precisely what happened after the click.

What to track on a landing page (and what you can skip)

The primary conversion event

A landing page almost always has just one business goal: a form submitted, a meeting booked via Calendly, a payment completed on an e-commerce page, a lead magnet downloaded. That single event, and only that one, should become the "key conversion" tracked in GA4 — a page tracking fifteen different events with equal weight ends up watching none of them seriously.

The micro-conversions that explain the why

Around that primary goal, a few micro-conversions help you understand where a page is losing visitors without ever becoming the success metric itself: scrolling down to the form, clicking the CTA before the actual submission, or triggering an exit-intent popup. Combined with the bounce rate and average time on page, they let you tell apart a page that isn't attracting the right traffic from one that attracts the right traffic but loses it along the way.

Setting up the conversion event in Google Analytics 4

The form_submit auto-tracked event trap

GA4 automatically captures certain events, including form_submit as soon as a visitor sends a standard HTML form — convenient, but misleading if the page then shows a validation error: the event fires on send, not on confirmation that the data was actually received. For a contact or booking form, it's better to fire a custom event (generate_lead, for instance) at the exact moment the thank-you page loads or the API confirms receipt, as covered in our article on the thank-you page.

Marking the event as a key conversion

Once a reliable event is in place, it needs to be marked as a conversion in the GA4 interface (Configure → Events → toggle "Mark as conversion") to show up in acquisition reports and, if paid campaigns are connected, flow through to Google Ads or Meta Ads. That number, and only that number, should be the reference for judging a Google Ads campaign, a Meta Ads campaign or a LinkedIn Ads campaign — not raw clicks or traffic volume, which say nothing about the quality of the visitors brought in.

UTMs: knowing where each conversion actually came from

Without UTM parameters (utm_source, utm_medium, utm_campaign) added to every outbound link pointing at the landing page, GA4 lumps a large share of paid traffic under "direct" or "unassigned", making it impossible to compare a TikTok Ads campaign to a LinkedIn one based on measured conversions alone. The rule is simple: no ad link, no email signature, no partner post should ever point to a landing page without a UTM — a Google Ads quality score partly depends on data that only clean tracking can collect over time.

Why your GA4 numbers will never be 100% accurate

Even when properly set up, no tracking tool built on cookies and JavaScript can claim total accuracy. A 2016 study by researchers Dominic Coey and Michael Bailey, then at Facebook, presented at the WWW conference, quantified a bias that's often overlooked: because the same person browses with multiple cookies (desktop, mobile, private browsing), a test measured at the cookie level rather than the person level underestimates the true effect by a factor of roughly three, and needs two to three times the volume to detect the same signal (study). On top of that come visitors who decline measurement cookies through the GDPR consent banner, ad blockers that stop GA4 from loading at all, and device switches between an ad click and the eventual conversion. The number GA4 shows is therefore never "the" real conversion count, but a reliable floor for tracking a trend over time — which is more than enough to arbitrate between two versions of a page, as long as you never mistake it for an exhaustive count.

The most common tracking mistakes on a landing page

• Tracking the CTA click as the conversion instead of the actual form submission — a visitor who clicks then abandons counts as a success when it isn't one.
• Setting no UTMs at all on ad links or in emails, artificially inflating "direct" traffic and making any channel comparison useless.
• Marking several events as key conversions at once, diluting the single number that should drive decisions.
• Forgetting to retest tracking after a page redesign — a renamed or moved button can break a trigger with no error showing up anywhere.
• Comparing GA4 numbers to those of another tool (CRM, payment platform) without ever reconciling the two, when they will never measure exactly the same thing.

The checklist before trusting your numbers

1. One event clearly identified as the primary conversion, fired on actual confirmation rather than a simple click.
2. Systematic UTMs on every outbound link to the landing page, following a consistent naming convention.
3. A manual test of the full journey after every page update, form included.
4. Reading numbers as a trend over several weeks rather than as a single day's absolute value.
5. Cross-checking against at least one other source (CRM, payment platform) at least once a month.

Enhanced Conversions on a landing page: how Google Ads' SHA-256 hashing actually works

A form gets submitted, a real lead happens — but Google Ads only sees half of it: a blocked cookie, Consent Mode denying ad_storage, a browser trimming what it tracks. Enhanced Conversions close part of that gap by sending a hashed fingerprint of the form's email or phone number. How it works, what changed in 2026, and how to install it without breaking it.

The mechanism is distinct from Consent Mode, which it's often confused with. Consent Mode governs the right to send a signal; Enhanced Conversions improve the quality of that signal once the right has been granted. In practice, alongside the standard conversion tag (fired on the thank-you page, never on button click — the same logic used for the Meta pixel's Lead event), the landing page sends Google Ads one or more values captured in the form: email, phone number, or postal address. That data never leaves in plain text: it's hashed with SHA-256, a one-way function — there's no way to recover the original email from the hash. Google then compares that hash against the ones it already holds for its own signed-in user accounts, and when it matches, it links the conversion back to the originating ad click, even without a usable cookie.

For web or for leads: the distinction that decides who hashes what

• Enhanced conversions for web — built for a standard conversion tag placed on the thank-you page. The form sends the data in plain text to the Google tag (via gtag or a GTM tag), and it's Google's own script, running in the visitor's browser, that performs the SHA-256 hashing before the data is sent. Never hash the data yourself on this path: double-hashing silently breaks the match.
• Enhanced conversions for leads — built for offline conversions: a contact form that turns into a signed quote a week later, a sales call that converts into a customer three days after. Here, it's on you (or your CRM) to hash the data in SHA-256 before sending it to Google, through offline conversion imports or the Google Ads API.

Google Consent Mode v2 on a landing page: what changed for your Google Ads campaigns

Since March 2024, a landing page running Google Ads campaigns in Europe without Consent Mode v2 stops building remarketing audiences from its European visitors, and loses part of its measured conversions. What changed, what to actually install, and what research shows about this trade-off between privacy and ad measurement.

Basic mode or advanced mode: the difference that matters

• Basic mode — Google tags (GA4, Google Ads conversion) don't load at all until the visitor has given consent. Simple to set up, but no data exists at all for visitors who decline or ignore the banner: no measurement, no basis for statistical estimation.
• Advanced mode — tags load regardless, but send an anonymized "cookieless ping" even without consent. Google then aggregates these signals through conversion modeling: a statistical model that estimates, from the observed behavior of consenting visitors, the likely conversion volume among non-consenting visitors. This modeling only activates above a certain volume (Google cites a threshold around 700 ad clicks over 7 days per country and domain) — below that, advanced mode adds nothing over basic mode.

Conversely, a study by Goldberg, Johnson and Shriver, published in 2024 in the American Economic Journal: Economic Policy, measured a 12% drop in page views and recorded revenue across 1,084 sites in Europe after GDPR enforcement began, with an effect twice as large on small e-commerce sites as on large ones (see the study). In other words: compliance carries a real measurement cost, and it falls hardest on small operations that have neither the traffic volume needed for reliable modeling nor the technical team for a careful implementation — one more reason to start from clean source code rather than piling scripts on over successive campaigns.

Call tracking on a landing page: measuring the calls your campaigns generate

A visitor clicks your phone number in the hero, calls, becomes a customer — and that conversion exists in none of your tools. Call tracking fills that blind spot, but two very different visions compete on how to implement it.

A tel: link opens the device's phone app and instantly leaves the browser context: no conversion event, no UTM parameter, no cookie survives that jump. Without specific instrumentation, a click on a number hardcoded into the HTML doesn't even send an event to GA4 — as far as the browser is concerned, it's just another link. And even after adding an onClick event that captures the click (which we cover further down), that click measures the intent to call, not the call actually placed, let alone its duration, whether it turned into a sale, or its precise advertising source. That's exactly the gap call tracking is built to close.

Two ways to measure a call, two philosophies

Attribution models: who really deserves credit for your landing page's conversions?

One landing page, five traffic sources, a single form submission: who gets the credit? The answer depends entirely on the attribution model you apply — and that choice, usually made by default, decides where your ad budget goes next month. A tour of the six models, the reasons your tools will never agree, and the stance to take when you have no data team.

This is well documented academically. In "Mapping the customer journey: Lessons learned from graph-based online attribution modeling" (International Journal of Research in Marketing, 2016), Eva Anderl, Ingo Becker, Florian von Wangenheim and Jan Hendrik Schumann model customer paths as Markov walks, across four large customer-level data sets each covering at least seven distinct online channels. Their conclusion: the sequence of touches carries information, and the usual heuristic rules — last click chief among them — value channels quite differently from a model that accounts for real journeys. Hongshuang (Alice) Li and P. K. Kannan, in "Attributing Conversions in a Multichannel Online Marketing Environment" (Journal of Marketing Research, 2014), estimate the carryover and spillover effects of one channel on another and show that ignoring them distorts each channel's measured contribution.

The six attribution models, plainly

The most-cited experiment on this point is Thomas Blake, Chris Nosko and Steven Tadelis's "Consumer Heterogeneity and Paid Search Effectiveness: A Large-Scale Field Experiment" (Econometrica, 2015). The authors actually switched off eBay's paid search ads in selected markets in order to measure their causal rather than correlational effect. The result: the genuinely causal return on paid search was only a fraction of what non-experimental methods estimated, with the gap widest on brand queries, where users would have arrived through the organic result anyway. The most flattering line in your last-click report is often the one with the least real incremental value.

Tracking never replaces a good landing page — it only tells you which of two versions is the good one. Our 10 LanderKit templates (€89 each, €229 for the full pack) ship with a page structure that makes it easy to wire up clean conversion events, from the SaaS waitlist to the single-product e-commerce template.