In GA4 you see hundreds of events, but you don't know which ones actually matter for your business. The number of sessions grew — but how many people actually left an inquiry or booked an appointment? The answer comes only from proper event configuration. In a small business you don't need to track every click — you need to see how many people performed an action that brings revenue: submitted a form, called, booked an appointment, or bought something. The rest is just noise that distracts.
On this page you'll find a practical guide: which GA4 events are worth tracking in a small business, how to mark key events, how to verify that data is reliable, and how to build a simple report you'll review monthly. You'll also learn why every form submission isn't necessarily a lead and how to distinguish actual conversions from false signals.
Problem: sessions alone aren't enough
When you open a GA4 report, the first thing you see is the number of sessions and users. These numbers grow — and it might look good. But for a small business owner this information is practically useless. A session is just a visit to the site, not an action that brings money.
The real question is: how many people did what the site exists for? The answer requires moving from tracking traffic to tracking events — specific user actions on the site. GA4 automatically collects events through enhanced measurement: scrolls, outbound clicks, file downloads, form interactions, and video engagement. You can read more about exactly which events GA4 collects in the Google Analytics Help article about enhanced measurement.
Events in GA4 fall into three levels. The first level is customer actions that bring revenue: submitting a contact form, booking an appointment online, making a purchase. The second level is steps leading to those actions: user opened a form, clicked the "book now" button, visited the pricing page. The third level is noise: scroll, menu click, return to homepage. Your business should focus on levels one and two, but level three shouldn't be treated as success.
What GA4 collects automatically — and why it's not enough
GA4 in enhanced measurement mode automatically collects several useful events without additional code. Among them are form-related events: form_start (first user interaction with a form in a session) and form_submit (form submission). This data is available immediately after enabling enhanced measurement in GA4 settings. More information is available in the Google Analytics Help article about enhanced measurement.
The problem is that form_submit doesn't automatically mean a lead. The user could have submitted a site search form, subscribed to a newsletter, or made a validation error. Additionally, when users refuse cookie consent, some events may not be collected, and GA4 fills gaps with modeling — in that case the numbers in reports may be higher than actual actions. Each of these situations registers as form_submit, but only one of them is a real sales opportunity. That's why you can't blindly treat form_submit as a lead — you need to know exactly what you're measuring.
Additionally, GA4 collects events related to files (file_download), outbound links (outbound_click), and video (video_start, video_complete). These events might be useful in some industries, but in most small businesses they're not critical.
Why form_submit is not a lead
Form_submit registers every form submission, regardless of type. Typical false lead scenario: user types "Warsaw" in site search, lands on contact page, submits contact form asking about opening hours — system registers form_submit, but that's not a potential customer. Similarly, newsletter subscription registers as form_submit but brings no revenue. Verification means checking in CRM whether you're actually dealing with a lead or just a sent message.
Recommended events: generate_lead and purchase
Among the events recommended by Google (recommended events), there are two that matter directly for small businesses: generate_lead and purchase. The generate_lead event tracks situations when a user submits a form or asks for information — leaving a contact you can later sell to. The purchase event tracks purchase completion if you sell products online. Both definitions are in the Google Analytics Help article about recommended events.
The table below shows what business action corresponds to what GA4 event and how to verify that the event is real:
| Business action | GA4 event | How to verify credibility |
|---|---|---|
| Contact form submission | generate_lead | Check if the form has validation and if the content corresponds to a lead (not newsletter subscription) |
| Online purchase | purchase | Verify in payment gateway or CRM — transaction count must match |
| Online booking | generate_lead | Check in booking system if the appointment was actually reserved |
| Price list download with required email | generate_lead | Add required email field for download — then it's a lead, not a file |
| Phone number click | custom event (e.g. click_phone) | Ask the caller where they got the number — verify the source |
You can track each of these events in two ways: automatically through enhanced measurement (form_submit, but with limitations described above) or manually through GTM (Google Tag Manager), where you define the conditions under which the event should be sent. Manual configuration gives you full control and eliminates false events.
Phone number click and online booking — not yet conversions
A common mistake is treating a phone number click as a phone call. In reality, clicking a number is just expressing intent — the user could have changed their mind, not gotten through, or had a busy line. The same applies to going to the online booking system: the user entered the calendar but didn't select an appointment.
In reports for the owner, name these events honestly: not "calls" or "bookings", but "phone number clicks" and "booking system entries". Only later, combined with data from your CRM or booking system, will you see how many of those clicks turned into real actions. One simple rule: you track the click, but you measure conversion only after confirmation in another system.
The topic of installing a phone tracker on the site goes beyond this article's scope. If you need guidance on implementing call tracking, you'll find it in materials about GTM configuration — here it's enough to know that a phone number click is just the beginning of the funnel, not its end.
Key events: how to mark them and why the limit matters
In GA4 you can mark selected events as key events. This way, data about these events appears in dedicated reports and is taken into account when modeling conversions. In a standard property you can mark up to 30 events as key, and in an Analytics 360 property — up to 50 events. Information about limits comes from the Google Analytics Help article about key events.
GA4 automatically marks purchase as a key event. When connected to Google Ads with ad personalization enabled, the following events are also added as key events: add_to_cart, add_to_wishlist, begin_checkout, session_start, view_item, view_item_list, view_search_results — exactly as described in the Google Analytics Help article. These settings are useful in online stores, but in a service company that doesn't sell through a cart, excess key events just distract attention.
For small businesses the recommendation is simple: choose 3-5 events that actually translate to revenue. Usually these are generate_lead, purchase (if you sell online), visit to a thank-you page after form submission, click on a CTA button leading to booking, and maybe price list download (if the price list is a lead). Leave the rest as regular events — you don't need to mark them all as key.
Data verification: test submission and reconciliation with CRM
Even the best configuration is useless if you don't know whether the data is real. Before starting regular report usage, perform a test form submission from your own device and check in DebugView (debug view in GA4) or the real-time report whether the event was registered. Repeat the test from different devices and browsers.
After a week or two, compare data from GA4 with data from your CRM or booking system. A typical situation looks like this: GA4 shows 14 form submissions, but in CRM you have 11 leads. Where's the difference? Differences can come from spam submissions, validation errors, cookie consent refusal (in that case GA4 shows fewer than CRM due to modeling), or integration problems between systems. That's normal — GA4 shows interaction with the form, and CRM shows real leads. Both numbers have informational value, but neither can be treated as the sole truth. This is an example using placeholder numbers — substitute your own.
Regular reconciliation of GA4 data with CRM should become a habit. Monthly check if trends in GA4 match trends in CRM. If leads grow in GA4 but decline in CRM — or vice versa — something is wrong with configuration or with the lead handling process in your business.
How to conduct test verification
Open the form page in incognito mode. Fill out the form with real data and submit. At the same time, open DebugView in GA4 and observe whether generate_lead or form_submit event appears. Also ask a colleague from another location to do the same test — you'll check if data is collected independently from the device. After the test, go to CRM and check if the lead actually appeared. Only this two-way verification gives confidence that the configuration works correctly.
Consent mode and refusal — how it affects data
If your site displays a cookie consent banner, you need to know what happens when a user refuses consent. Consent mode doesn't display the banner itself — it works with the banner you already have on the site. When a user refuses cookie consent, Google Analytics doesn't collect full data, but tries to fill gaps with conversion modeling and behavioral modeling. More details are available in the Google Analytics Help article about consent mode.
This means that numbers in GA4 reports after enabling consent mode aren't equal to the number of users who actually performed an action. Modeling is an advanced feature that helps analyze trends, but it doesn't give precise data about each individual user. If accuracy matters to you, make sure as many users as possible give consent — or accept that reports will contain modeling elements, not just hard data.
The topic of cookie banner and consent mode configuration is extensive and goes beyond this article's scope. If you need to implement GDPR-compliant cookie consent, check dedicated tools or consult a lawyer or DPO (Data Protection Officer).
Who has access to GA4 — and why it matters
A small business owner should have full access to their GA4 account — and on their own Google account, not on the account of the entity that created or manages the site. Access and user management in GA4 requires the Administrator role at the account or property level. Information comes from the Google Analytics Help article about user roles.
Why is this important? It happens that an agency or freelancer creates a GA4 account on their own account and gives the business owner only view access to reports. In this situation, the owner can't independently add users, change configuration, or mark key events. If cooperation with the agency ends, data access can be difficult or impossible. Therefore, when the site is delivered or cooperation ends, make sure the GA4 account is on your Google account and that you have the Administrator role.
If the account is in your name, you can give the agency access or revoke it at any time — data security is on your side.
Minimum report for the owner: 5 key events and CRM reconciliation
You don't need a complicated dashboard with dozens of indicators. You need one simple report you'll review monthly and that answers the question: how many people performed a revenue-generating action and where did they come from?
The table below is a ready template for monthly review:
| Channel | Key event | Count in GA4 | Count in CRM | Difference |
|---|---|---|---|---|
| Organic (Google) | generate_lead | |||
| Direct | generate_lead | |||
| Referral | generate_lead | |||
| Google Ads | generate_lead | |||
| Organic (Google) | purchase |
Fill data once a month, preferably on the same day (e.g., first Monday of the month). Fill the "Count in GA4" and "Count in CRM" columns — the difference will show you how many GA4 events didn't translate to real leads or purchases. With large systematic differences, investigate the reasons: the form might be misconfigured, leads might go to spam, or the booking system might not save all appointments.
Do it yourself in 2 hours: practical step list
If you want to configure event tracking in GA4 yourself, follow the list below. The whole process should take about 2 hours, depending on the number of forms and systems you need to connect.
Make a list of actions leading to revenue in your business: contact form, appointment booking, online purchase, price list download, phone number click. For each action, assign the corresponding GA4 event from the table in section 3 of this article.
Go to GA4, go to the Configuration section, and check which events are already being collected automatically. For each action on your list, decide whether form_submit from enhanced measurement is enough, or you need manual configuration through GTM.
Go to the Key events section and uncheck all events that don't match your list. Add a maximum of 5 events that actually translate to revenue. Leave the rest as regular events.
Perform a test form submission from your own device and check in DebugView whether the event was registered correctly. Ask someone from your team or family to perform a test submission from another device.
After a week, compare GA4 data with data from your CRM. Write down differences in the table and think about what you can improve in configuration or in the lead handling process.
Repeat reconciliation monthly — after a few months you'll have enough data to evaluate which channels bring real leads and which only generate traffic without business value.
Tracking scheme: from click to report
- 01customer action
- →02GA4 event
- →03key event
- →04monthly report
- →05CRM reconciliation
The above scheme shows the complete path: user clicks submit button, GA4 registers form_submit, you mark generate_lead as key, monthly you review the report and compare with CRM. This way you see the real impact of your site on the business.
What it looks like when the system takes over
When events on the site are properly configured and key events are marked, you can build an automatic report that combines GA4 data with CRM data. The system shows you the full funnel: source channel (where the user came from) → key event (what they did on the site) → lead in CRM (whether they were registered as a potential client) → visit or purchase (whether the transaction happened).
This report allows monthly analysis of discrepancies between GA4 and CRM and quick responses: if Google Ads brings many clicks but few leads in CRM — the problem might be on the site (form, landing page) or in the ad itself. If organic channel brings little traffic but every third one leaves a lead — maybe it's worth investing in SEO.
See how this mechanism works in practice:
- Analytics and BI — measurement setup and a monthly report: where clients come from and what one inquiry costs
- Dashboards — leads, sales, marketing and finance on one screen
- AI Reports — a weekly plain-language summary of leads, campaigns and sales
- CRM and automations — requests from every channel in one funnel
- Google Ads — campaigns judged by inquiries and orders, not clicks
Read more about how businesses analyze customer behavior on websites and why it's worth tracking specific actions rather than just traffic. Also learn why customers don't leave inquiries, how to automate reporting, how to automate query handling, and how to measure restaurant KPIs.
Frequently asked questions
Do I need to track all events that GA4 collects automatically?
No. GA4 collects several events automatically through enhanced measurement, but most of them are noise that doesn't matter for a small business. Focus on 3-5 key events that actually translate to revenue: generate_lead, purchase, CTA button click leading to booking. The rest can be collected, but you don't need to mark them as key or analyze them daily.
Can I treat form_submit from enhanced measurement as a lead?
Not without verification. Form_submit means the user submitted a form, but it doesn't mean it was a lead. The user could have subscribed to a newsletter, submitted a form with incorrect data, or the form could have gone to spam. When users refuse cookie consent, GA4 may show fewer events than CRM due to modeling. For form_submit to be a credible lead, the form should have validation, and you should reconcile GA4 data with CRM at least monthly.
How many key events can I mark in GA4?
In a standard property you can mark up to 30 key events. In an Analytics 360 property the limit is 50. For most small businesses 3-5 key events are enough — there's no need to mark more. Excess key events complicate reports and make drawing conclusions harder.
What to do when GA4 data differs from CRM data?
That's normal. GA4 measures interaction on the site (user submitted a form), and CRM measures real leads (someone was actually registered as a client). Differences come from spam submissions, validation errors, cookie consent refusal (in that case GA4 shows fewer due to modeling), or integration problems between systems. Regularly reconcile data and look for causes of systematic discrepancies — they most often point to a problem that needs fixing.
Can I track phone number clicks as calls?
A phone number click is just expressing intent, not a confirmed call. The user might not have gotten through, had a busy line, or changed their mind. To measure real calls you need a dedicated call tracking tool (e.g., integration with a phone system or call tracking). Treat a phone number click as "phone number click", not as "call".
Does consent mode affect data accuracy?
Yes. When a user refuses cookie consent, GA4 doesn't collect full data but tries to fill gaps with conversion modeling and behavioral modeling. This means numbers in reports are estimates, not precise. If accuracy matters to you, make sure users give consent — or accept that reports will contain modeling elements, especially with a small sample.