You will be able to run a short data health check that catches broken tracking before it misleads a report.
Jun's monthly report for the furniture shop opened with bad news: website enquiries down 45 percent on the month before (the figures in this lesson are examples). The owner frowned and asked what had gone wrong with the marketing. Jun had already drafted three possible reasons. Then the owner's sister mentioned that the showroom had actually been busier than usual with people who said they had sent a form online. Jun checked. On the 9th, a web developer had updated the contact page, and the thank-you page had moved to a new address. From that day on, GA4 had stopped counting enquiries. Nothing was wrong with the marketing. The tracking had broken.
Every analytics setup breaks sooner or later, usually quietly. This lesson is about catching that before it reaches a report, with a short check you run at the start of every monthly review.
Real customer behaviour tends to move gradually, or with the calendar: busier before festive seasons, quieter in certain months. When a key event drops to zero overnight, halves on a single day or suddenly doubles, the cause is usually a change to the site or the tracking, not a change in customers.
So the first check each month is to look at each key event as a daily line chart over the month, not just the monthly total. A total can hide a cliff in the middle. Jun's enquiries looked merely low as a monthly figure; as a daily line, they fell to zero on the 9th and stayed there.
When you see a jump, ask what changed on that date. Common culprits are a website update or redesign, a new theme or plugin, a change to a form or booking tool, a new cookie banner, someone editing the tag setup, or a second copy of the tag being installed, which doubles counts. Keep a note of site changes as they happen, so this question has an answer.
Next, check that the month's campaigns show up in GA4 the way you tagged them. Open the traffic acquisition report, switch to session source and medium, and then to session campaign, and look for every campaign you ran.
Look for three problems. A campaign that is missing entirely suggests the links went out without tags, and its visits are sitting in direct. A campaign split across two or three spellings means someone ignored the naming rules from lesson 2.2. And a rising Unassigned channel usually means a new medium that GA4's default channel groups do not recognise.
Then glance at direct traffic. If direct jumped in a month when you sent a big untagged email, the two are probably connected. Cross-check against your tagging sheet from lesson 2.4: every link logged there should appear in GA4 under the right source and medium.
The most important check compares GA4's key events with the real results in your own system: enquiries in the inbox or CRM, bookings in the booking tool, orders in the shop platform.
The two will never match exactly, for reasons covered in lessons 4.2 and 5.2: cookie choices, ad blockers, people who enquire by phone or WhatsApp. What you are watching is the ratio between them. If GA4 usually records about 70 percent of the enquiries your inbox receives, and this month it records 30 percent, something has broken. If it suddenly records 140 percent, something is double counting.
This check would have caught Jun's problem in a minute. Real enquiries were normal. GA4's were not. The gap is the signal.
Some changes are deliberate and still break comparisons: a new key event added mid-month, an internal traffic filter switched on, a fix to a double count, a move to a new website. The data after the change is better, but it is not comparable with the data before.
Write these into the report itself, near the top, in plain words. "Enquiry tracking broke between the 9th and the 23rd and was fixed on the 23rd; the GA4 enquiry figure for this month is incomplete, and the inbox count of 74 is the reliable number." Readers then know which comparisons to ignore, and nobody makes a budget decision based on a broken number.
Jun's final report led with the real enquiry count from the inbox, explained the tracking gap in two sentences and fixed the thank-you page address the same afternoon. He also added a line to his tracking log from lesson 3.4, so the next person to touch the contact page would know to check the event.
The checks above take about fifteen minutes once you have done them twice. The way to make sure they happen is to write them as a numbered list and run it, in order, before reading any result. A pilot runs through a checklist before every flight not because they forget how to fly, but because skipping one step is easy when everything seems normal.
Your list should cover events, tags, filters and a comparison with real sales or leads, written for your own setup, with each check phrased as a question that has a yes or no answer.
Write a ten-point monthly data check for your setup, covering events, tags, filters and a comparison with real sales.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).