Quick Answer: Most GA4 conversion tracking guides are click-by-click tours of the interface. The clicking is not the hard part. The hard part is deciding what counts as a conversion in a business where half the leads arrive on WhatsApp, and being disciplined enough to leave everything else unmarked.
Nearly every GA4 property we inherit has conversion tracking of a sort. There are events, there are stars next to some of them, and there is a number in the report. Nobody in the marketing team can tell you what that number represents.
That is the real failure. Not a missing tag — a number that nobody trusts, quietly used in a board deck anyway.
This guide is the sequence we use at ZenWeb when we take over a Malaysian SME’s analytics: what GA4 calls a conversion in 2026, which actions deserve to be marked, how to set them up, and how to check the numbers before anyone acts on them.
Source video: Key Events and Conversions in Google Analytics, from the Google Analytics channel, on YouTube
Quick Answer: In GA4, the thing you set up is a key event — an event you mark as important. A conversion now means something narrower: an action used to measure and bid on Google Ads campaigns. Same star icon, two different words, and they are no longer interchangeable.
Google renamed the concept in 2024, and the wording still trips people up in meetings. GA4 conversion tracking starts with knowing which word means what, because the naming is exactly why your GA4 report and your Google Ads report were never going to agree:
For an in-house marketer, the practical translation is short. Build key events for the actions your business cares about. Promote only the ones you want Google Ads to bid on into conversions. Do not assume the two numbers should match, because they are counting different things.
Quick Answer: Pick two or three actions that a salesperson would recognise as a real enquiry. Standard properties allow up to 30 key events, which is precisely why most teams end up with a dozen and no idea which one matters.
You can mark up to 30 events as key events on a standard property. Treat that as a ceiling, not a target. The test for any candidate action is simple: if it fires, would somebody in sales have a name and a way to contact this person?
Engagement events are worth collecting. They simply do not belong in the column your managing director reads as “enquiries this month”. Keep the two apart, and your monthly marketing report stays honest without anyone having to caveat it. This one decision does more for the quality of your conversion tracking than any tag you will ever build.
Not sure which actions on your site are really worth counting?
That decision usually comes down to how the site is meant to earn its keep. See how our SEO team maps traffic to enquiries →
Quick Answer: Across the GA4 properties we audit at handover, the form submission is nearly always marked and often wrong. WhatsApp — the action most Malaysian buyers actually take — is marked less than half the time, and phone taps almost never.
What we find in the Admin panel on day one, and how much of it survives a test:
| Lead action | Has a key event | Fires correctly on test | Most common fault |
|---|---|---|---|
| Contact form submitted | 91% | 54% | Thank-you page view used as the trigger |
| WhatsApp enquiry opened | 44% | 31% | Floating widget click never tracked |
| Phone number tapped | 19% | 17% | Tel link not set up as an event at all |
| Quote or booking completed | 37% | 28% | Booking tool on a separate domain |
| Scroll, download or page view marked as key | 63% | — | Inflates the enquiry number by design |
Source: ZenWeb operational data, GA4 audits at client handover, Malaysian SME accounts, 2024–2026.
Read the last row against the second. Nearly two properties in three are counting a scroll or a download as a conversion, while fewer than half are counting the WhatsApp click that produced most of the actual enquiries. The number in the report is not small — it is simply made of the wrong things.
Quick Answer: GA4 conversion tracking comes down to six steps. Send a clean event for each chosen action, confirm it arrives in DebugView, mark it as a key event in Admin, then wait a day for the standard reports to catch up. One afternoon, no code for most WordPress sites.
form_start and form_submit, outbound clicks, scrolls and file downloads with no code changes at all. Sometimes the event you need is already there.lead_form_submit, whatsapp_click, phone_click. Renaming later means losing history.You need Marketer access or above to mark key events, and Editor access to change the data stream. Chasing down whoever holds the account is often the longest step of the six.
Quick Answer: A thank-you page view is the quickest method to build and the least accurate to live with. A custom event fired on successful submission takes about an hour longer and comes within a few percent of the CRM.
We rebuilt GA4 conversion tracking for the same lead form four ways on client sites, then compared each method’s count against confirmed CRM leads over 30 days.
| Setup method | Accuracy vs CRM | Accuracy | Setup time |
|---|---|---|---|
| Custom event on successful submission | 96% | 90 min | |
| Tag manager click trigger on the submit button | 84% | 45 min | |
| Enhanced measurement form_submit, untouched | 79% | 5 min | |
| Thank-you page view | 61% | 10 min |
Source: ZenWeb client tracking, lead forms rebuilt four ways on Malaysian SME sites, 2024–2026.
The five-minute option is not a bad place to start. Enhanced measurement’s form_submit gets you to a defensible number the same afternoon, and it beats the thank-you page it usually replaces. Just know that it fires when the form is submitted, not when the submission succeeds — so a validation error or a failed spam check still counts.
About to put budget behind these numbers?
Tracking has to be right before the spending starts, not after — the same discipline our guide to launching your first Google Ads campaign → insists on.
Quick Answer: In Malaysia the main lead action is usually a WhatsApp click, and GA4 can only see the click — not whether a message was ever sent. Track it, mark it, and label it honestly as an intent to enquire rather than a confirmed lead.
Almost every guide written for a US audience assumes the conversion is a form. Here, the form often sits untouched while the floating WhatsApp button does the work. That leaves you with a measurement gap the interface will never warn you about.
What GA4 can and cannot see:
wa.me leaves your domain, so enhanced measurement records it as an outbound click, with the destination captured in the link parameters.The workable answer is to fire a whatsapp_click event on the button, mark it as a key event, and reconcile it monthly against the messages your sales team actually received. On the accounts we manage, roughly six or seven in ten clicks turn into a real conversation, so the raw click count is useful once everyone knows what it is. Our walkthrough of GA4 and WhatsApp lead tracking for Google Ads covers the tagging in detail, and if reading the two numbers side by side sounds daunting, our guide to reading GA4 reports without getting overwhelmed shows where each one lives.
Quick Answer: When a client says GA4 is lying, the cause is nearly always mundane. Double-firing tags and thank-you page refreshes explain most of the over-count; ad blockers and consent banners explain most of the under-count.
Causes of the gap between reported key events and CRM-confirmed leads, ranked by how often they turn out to be the culprit:
| Root cause | Share of audits | Share | Direction |
|---|---|---|---|
| Thank-you page refreshed, bookmarked or shared | 47% | Over-count | |
| Same event sent by both the plugin and tag manager | 38% | Over-count | |
| Ad blockers and privacy browsers | 29% | Under-count | |
| Consent banner blocking the tag | 24% | Under-count | |
| Spam submissions counted as leads | 21% | Over-count | |
| Booking or checkout on another domain | 18% | Under-count |
Source: ZenWeb operational data, GA4 audits across Malaysian SME accounts, 2024–2026. Audits often surface more than one cause.
Notice that the two biggest causes both inflate the number, and both are fixable in an afternoon. Before anyone starts a debate about attribution models, check whether the thank-you page is being counted twice.
Quick Answer: DebugView is instant, Realtime takes minutes, standard reports take up to a day. But the count only settles once the duplicates and spam are cleaned out, and that usually takes a month of watching.
The gap between what GA4 reported and what the CRM confirmed, tracked from the day rebuilt GA4 conversion tracking went live.
| Time after go-live | Median gap vs CRM | What gets fixed in this window |
|---|---|---|
| Day 1 (DebugView only) | +41% | Nothing yet — you are only confirming the event fires |
| Day 2 (standard reports appear) | +33% | Obvious double-fires spotted and switched off |
| Week 1 | +19% | Spam submissions filtered, refresh counting removed |
| Week 4 | +8% | WhatsApp clicks reconciled against real conversations |
| Week 8 | +5% | Residual gap only — blockers and consent, accept it |
Source: ZenWeb client tracking, conversion tracking rebuilds on Malaysian SME accounts, 2024–2026.
The lesson is in the last row. A five percent gap is where a well-built setup lands and stays. Chasing it to zero costs weeks and buys nothing, because some visitors will always block the tag. Agree with your stakeholders early that the target is a stable, explainable gap — not a perfect match.
Inherited a GA4 property you do not trust?
A two-hour audit usually finds the duplicate tag and the missing WhatsApp event. Ask us to check your tracking setup →
GA4 conversion tracking is not a technical challenge. It is an editorial one. The interface will happily let you star a scroll event and call the result your enquiry number. It will never warn you that the WhatsApp button, the thing most of your buyers actually press, is invisible.
So do the unglamorous version. Choose two or three actions a salesperson would recognise. Fire a clean event on each, watch it land in DebugView, mark it, then reconcile the count against your inbox for a month until it settles. From there, the same discipline carries into the rest of the channel: knowing what a lead costs, knowing which pages produce them, and knowing which posts to optimise for SEO because they already convert. If you would rather have the tracking, the reporting and the traffic handled together, that is what our SEO team does for Malaysian businesses every month.
A key event is any event you mark as important to your business, and it appears in your standard GA4 reports. A conversion is created from a key event, shared with Google Ads, and used for bidding and campaign reporting. Google renamed conversions to key events in 2024 to align the definition with Google Ads.
Two or three for most Malaysian SMEs. A standard property allows up to 30, but every extra one dilutes the number your management team reads as enquiries. Mark only the actions that produce a contactable lead, and leave scrolls, downloads and page views as ordinary events.
Yes, as a click. Fire an event when the WhatsApp button is pressed and mark it as a key event. GA4 cannot see whether a message was actually sent, so treat the number as intent to enquire and reconcile it against your sales inbox each month.
Usually because a thank-you page view is being used as the trigger and visitors refresh or bookmark it, or because the same event is sent twice by both a plugin and a tag manager. Both inflate the count and both are fixable in an afternoon.
DebugView shows the event immediately and Realtime within minutes. Standard reports can take up to 24 hours. Marking an event as a key event affects reports from that point onward and does not change historic data.
Want a conversion number you can defend in a board meeting?
Book a free 30-minute strategy session. We will review your GA4 setup, tell you which key events are lying to you, and show you what a clean number would look like.
Complete the form and our team will contact you to discuss your goals. Let’s grow your business.

Online