GA4 for SMEs: the reports you actually need (and how to read them)

The owner of a company that sells coffee machines to bars opens Google Analytics 4 on a Monday morning. He sees “Active users: 3,412”, a map of the world, a chart that goes up and down and a menu with thirty items. He closes it after two minutes with the feeling that the numbers are there but say nothing. On Friday the salesperson asks him whether the campaigns are working and he answers “I think so”.

It’s the scene we see most often on first calls. GA4 isn’t difficult because it has too many features, but because nobody has decided which questions to ask it.

The short answer: an SME needs six GA4 views, not thirty. Traffic acquisition to know which channels bring useful people, key events to count the actions that matter, landing pages to understand where people come in and where they get lost, a funnel exploration to find the step that breaks, the monetisation reports if you sell online, and linked Search Console data for the SEO side. Before reading them you need three configuration checks, and afterwards an hour-long monthly routine that ends with a decision.

This isn’t a manual covering every feature: for the general method of making decisions with numbers, see the article on data-driven marketing. Here we talk about what to open, how to read it and what to leave alone.

What should GA4 do for an SME, and what not?

GA4 should answer a few questions that lead to a budget or website decision: which channels bring in leads or sales, which pages produce them, where people stop. Revenue, customers by name and the real effect of a campaign live in the management system, the CRM and incrementality tests.

GA4 can take 24-48 hours to process data, doesn’t see people who reject cookies and hides some rows to protect privacy: it’s built for comparing, not counting. A useful GA4 has a few clear key events, links to Google Ads and Search Console, and one person who opens it once a month with a list of questions.

Which checks should you do before reading a report?

Three settings: which events are marked as key, whether internal traffic is excluded and how long GA4 keeps exploration data. If one of them is wrong, every reading that follows starts off on the wrong foot.

Are the key events the right ones?

Under Admin > Key events you’ll find the actions GA4 considers important. We often find page_view or scroll marked as key: the “key event rate” goes up and means nothing. For an SME collecting leads, 2-4 events are usually enough: form submission, phone click, WhatsApp click, booking. For an e-commerce store, the purchase and not much else.

There’s also the counting method. GA4 lets you count a key event once per event or once per session: the first is the option Google recommends, the second is labelled “legacy”. For a form submitted twice in the same visit, once per session is closer to reality. The change only applies to future data.

Is internal traffic excluded?

If you and your colleagues browse the site every day, part of the traffic in the reports is you. GA4 lets you define internal traffic by IP address and exclude it with a data filter. I recommend keeping it in “Testing” state for a few days first and checking what it catches: once active, the exclusion is permanent.

How long does GA4 keep data?

By default, standard properties keep user-level data for 2 months, and it’s the first value we change: the maximum free option is 14 months. According to the guide to data retention, the setting concerns explorations and funnels, not aggregated standard reports. With 2 months you can’t compare October with the previous year’s October in an exploration, and the change isn’t retroactive: it’s worth doing today.

Which GA4 reports should an SME look at?

There are six, and each one answers a different business question. This table is the one we leave with clients after setup.

Report or exploration	Where to find it (roughly)	Question it answers	Metric to look at first
Traffic acquisition	Reports > Acquisition	Which channels bring visits that turn into leads or sales?	Key events and key event rate by channel
Key events	Reports > Engagement	How many valuable actions happened, and of what kind?	Count per event, trend compared with the previous month
Landing pages	Reports > Engagement	Which pages do people enter from and which ones produce results?	Sessions and key events per page
Funnel exploration	Explore	At which step do people drop out?	Abandonment rate between one step and the next
Monetisation (e-commerce only)	Reports > Monetisation	Which products sell, and which are viewed without being bought?	Revenue, purchases, add-to-baskets per product
Linked Search Console	Reports > Search Console (needs publishing)	For which searches and on which pages does organic traffic arrive?	Clicks, impressions and position by landing page

Menu item names change from time to time and depend on how the property has been customised: if a report is missing, an editor can restore it from the Library.

How do you read the Traffic acquisition report?

The Traffic acquisition report shows where sessions, new and returning, come from, grouped by channel (organic search, paid search, social, email, direct, referral). You read it starting from key events by channel, not sessions: a channel that brings lots of visits and no leads is noise, not success.

Be careful not to confuse it with User acquisition. Google’s guide explains it like this: User acquisition shows where new users come from, that is the first contact; Traffic acquisition looks at all sessions. To decide the month’s budget we almost always need the second.

How we read it:

1. Closed month against the previous month and against the same month the year before, which takes seasonality out of the picture.
2. Key events column filtered to the event that counts. “Form submission” says more than “all key events”.
3. Key event rate by channel. With figures made up for illustration: organic search brings 1,800 sessions and 27 enquiries (1.5%), paid social 2,400 sessions and 7 enquiries (0.3%). Social doesn’t necessarily need switching off, it may be building brand awareness, but it’s a question to take to the meeting.

What to ignore: daily fluctuations and the “Unassigned” row when it’s small (sessions GA4 can’t classify). If “Unassigned” or “Direct” grow a lot, on the other hand, something in the tracking has broken: links in emails without UTMs, redirects that lose parameters, a cookie banner blocking the tag.

Key events or conversions: what’s the difference?

In GA4, key events are the actions that matter for your business; conversions are key events imported into Google Ads and used to measure and optimise campaigns. Until 2024 GA4 called both of them “conversions”, and that’s the source of a lot of confusion in reports and meetings.

According to Google’s guide to key events and conversions, any event collected can become a key event, while a Google Ads conversion derives from a key event and is used for campaigns. Google Ads conversions don’t appear in Analytics’ standard reports: you’ll find them in the Advertising section, in the conversion performance report.

	Key event (GA4)	Conversion (Google Ads)
What it measures	An important action on the website or in the app	An action used to evaluate and optimise ads
Who attributes it	GA4, to all channels	Google Ads, to ad interactions
Attribution model	The one set in GA4 (data-driven by default)	The one set in the Google Ads conversion action
Where to read it	GA4 reports	Google Ads and GA4’s Advertising section

That’s why the two numbers almost never match. The full comparison between sources, CRM included, is in Milica’s article on how to check data from click to sale.

How to read the Key events report

I look at it for two things: the trend compared with the previous month and events that disappear. A key event that drops from 40 to 0 in a week isn’t a market collapse; it’s almost always a modified form or a broken tag.

How do you use the landing pages report?

The Landing page report shows the first page seen in each session, with the sessions, engagement and key events that follow from it. It’s the report most underrated by people who run a company website, because it connects traffic to the content that receives it.

The three questions we ask it:

• Which pages bring the most key events? There are usually only a few: the home page, two or three service or product pages, a few articles. These are the pages to protect when the site is rebuilt.
• Which pages get lots of traffic and no results? An article with 900 entrances a month and zero leads is an opportunity to add a link to the service page.
• Which pages have lost entrances compared with the year before? It’s the first clue to a ranking problem or a page removed without a redirect.

Adding the session channel as a secondary dimension shows whether a page only works with one type of traffic: a landing page that converts well from Google Ads and badly from Meta is speaking to two audiences with the same message.

What to ignore: the “(not set)” row, which collects sessions without a recognised entry page, and average engagement time read on its own. A long time on a contact page may mean the form is hard to fill in, not that people like it.

When do you need a funnel exploration?

You need it when you know people arrive but don’t know where they stop. The funnel exploration lets you define a sequence of steps (for example service page, contact page, form submission) and shows how many people reach each one and how many drop out.

For an SME we recommend building just one and saving it, instead of redoing it every time:

• Lead generation site: entrance on a service page, visit to the contact page, start of form completion (if tracked), submission.
• E-commerce: product view, add to basket, begin checkout, purchase.

The funnel can be open (people can enter at any step) or closed (they start from the first) and can be broken down by device: that’s usually where the problem jumps out, with mobile losing many more people on the form. How to turn that data into a hypothesis and a test is something I cover in the article on how to optimise the funnel.

If the funnel stops at two months ago, the reason is almost always that data retention has been left at 2 months.

What should you look at in the monetisation reports if you run an e-commerce store?

If you sell online, the Monetisation section tells you which products are viewed, added to the basket and bought, with the revenue generated. The report to start from is E-commerce purchases, which shows the detail by item.

The useful reading isn’t the bestseller ranking, which you already know from your management system. It’s the ratio between views, add-to-baskets and purchases for each product:

Situation	What it might mean	What to check
Lots of views, few add-to-baskets	Price, photos or product page aren’t convincing	Product page, availability, shipping costs shown early
Lots of add-to-baskets, few purchases	A problem at checkout or with the final costs	Shipping costs, payment methods, checkout errors on mobile
Few views, good purchase rate	A product that people who find it like	More internal visibility, dedicated campaigns

GA4 revenue will never match the figure in your management system, because of returns, cancelled orders and cookie consent. If the gap stays stable, GA4 is fine for comparing products and channels; if it changes from one month to the next, there’s a tracking problem. More detail in the guide to e-commerce conversion rate.

What’s the point of linking Search Console to GA4?

It lets you see in GA4 the Google searches that bring people to the site and the pages that receive them, alongside behavioural data. The link adds two reports, Google organic search queries and Google organic search traffic, which however aren’t visible by default: the collection has to be published from the Library.

Some limits to know before reading them:

• Search Console data reaches GA4 after about 48 hours;
• the history is Search Console’s, 16 months;
• a web stream can be linked to only one Search Console property, and vice versa;
• Search Console metrics can only be combined with a few Analytics dimensions (landing page, device, country).

The report we use most is the one on pages, with impressions, clicks and average position next to key events. A page with lots of impressions, an average position between 8 and 15 and a few leads is the first one to improve. For in-depth SEO analysis Search Console is still more convenient; GA4 connects search to the result.

Why are some GA4 numbers hidden or estimated?

Because GA4 applies thresholds to protect privacy and, if you use Consent Mode, it can estimate the behaviour of people who reject cookies. They’re two different phenomena, and you need to recognise them before drawing conclusions.

Data thresholds

When a report contains few users and sensitive dimensions, GA4 hides some rows. According to the guide to data thresholds, this happens mainly with demographic data and search queries; the thresholds are set by the system and can’t be changed. You can tell from the data quality indicator at the top of the report, which flags the threshold applied.

For an SME with little traffic it happens often. The practical solution is to widen the date range and remove demographic dimensions, which on 300 users a month won’t lead to any decision anyway.

Modelled data with Consent Mode

If the site uses Consent Mode, GA4 can estimate the behaviour of people who reject analytics cookies. It doesn’t happen for everyone: according to the guide to behavioural modelling, the property must collect at least 1,000 events a day with consent denied for at least 7 days and have at least 1,000 daily users with consent granted for at least 7 of the last 28 days. Even then, eligibility isn’t guaranteed. Estimated data only appears with the “Blended” reporting identity, and the quality indicator tells you whether you’re seeing it.

In other words: many company websites don’t reach those thresholds. GA4 then counts only people who accepted cookies, and the absolute numbers are lower than reality. That’s fine, as long as you know it: comparisons between channels and between months remain useful, and leads are counted in the CRM. The Google Ads side of the same issue is explained in Gaetano’s guide to Enhanced Conversions and Consent Mode.

How do GA4 and Google Ads link up, and what changes in the reports?

Linking GA4 and Google Ads does three things: it brings campaign costs and clicks into GA4, lets you import key events into Google Ads as conversions and makes audiences available for remarketing. According to the linking guide, Google Ads data appears in GA4 reports within 48 hours, and for remarketing Google Signals or user-provided data collection must be switched on.

Three things to know:

• One source for each conversion used in bidding. If the same form is tracked with the Google Ads tag and with the key event imported from GA4, one of the two actions must be left as secondary, or Smart Bidding counts the same lead twice.
• There are three attribution models. GA4 now offers the data-driven model and two last-click variants; first click, linear and the other rule-based models have gone since November 2023. If someone in a meeting asks about “first contact”, the answer has to be looked for in the attribution paths or the CRM.
• GA4 isn’t the place to manage campaigns. It’s for seeing what people who arrived from ads do on the site. To follow spend and results from Google Ads, Meta, LinkedIn and TikTok in a single view, we use RunAlytics with our clients.

Which monthly reading routine do you recommend?

One hour a month, on a month that closed at least a week ago, with the same questions every time and a written decision at the end. It’s the routine we suggest to clients who handle marketing in-house, and the same one we use to prepare the report we walk them through.

Minutes	Report	Question	Example decision
0-5	Data quality indicator and key events	Is the data complete? Has any event disappeared?	Flag to the developer a form that no longer records
5-20	Traffic acquisition	Which channel brought the most key events, and at what rate?	Move budget from a low-rate channel to a high-rate one
20-30	Landing pages	Which pages produce leads and which have lost them?	Add links to a service page from a much-read article
30-40	Linked Search Console	Which pages are close to Google’s first page?	Schedule an update of two pages
40-50	Funnel exploration (or Monetisation)	Where do people drop out? Has anything changed?	Simplify the form on mobile
50-60	Summary	What did we decide, who’s doing it, how will we measure it next month?	Three lines in a shared document

Two GA4 tools help, and few people use them. Annotations are dated notes on charts for marking a campaign launch, a site rebuild or a price change: after six months nobody remembers why traffic went up in March, but the annotation does. Scheduled email reports let an administrator receive a report every month with the same filters and comparisons.

The last column matters more than the others: a report that doesn’t produce a decision, even “we’re not changing anything”, is a reading you could have skipped.

What can you ignore in GA4?

Beyond what we’ve said report by report, in our experience an SME can leave alone:

• The Realtime report, except when you’re checking that a new event works.
• Demographics and interests with fewer than a few thousand users a month: thresholds and small samples make them unreliable.
• Bounce rate read on its own: in GA4 it’s the opposite of the engaged sessions percentage, and says little without key events next to it.
• Changes under 10% on small volumes: with 30 leads a month, 3 more or fewer is noise.

Mistakes to avoid

Marking worthless events as key. Scrolls and page views inflate the key event rate and hide the real leads.

Leaving data retention at 2 months. Explorations can’t look further back, and the change doesn’t recover the past.

Reading the current month. Data is completed in the following days; comparisons make sense on closed months.

Importing the same lead into Google Ads from two sources. The Google Ads tag and the imported key event must be used so that only one drives bidding.

Looking at reports without a question. Without a decision to make, GA4 becomes a gallery of charts.

Want a GA4 that answers your questions?

We start from the decisions you make every month, sort out events, consent and links, and build the few views you need. Then, if you like, we read the report together once a month on a call. It’s the work we do on analytics and data.

You can book a call with Federico or write to us via the contact form.

Frequently asked questions

Why does GA4 show fewer leads than the CRM?

Because it only counts people who accepted analytics cookies, unless the property exceeds the thresholds for Consent Mode modelling, and because some leads come by phone, email or word of mouth without passing through the tracked website. The CRM remains the source for counting leads; GA4 is for understanding which channels and pages they come from.

What’s the difference between a key event and a conversion in GA4?

A key event is an important action measured in GA4, such as a form submission. A conversion is a key event imported into Google Ads and used to measure and optimise campaigns. Until 2024 GA4 used the word “conversion” for both.

How often should you look at GA4?

One structured reading a month, on the closed month, is enough, plus a weekly glance at key events to spot in time if a form stops recording. Reading it every day leads you to react to meaningless fluctuations.

What does “thresholding applied” mean in a report?

GA4 has hidden some rows with too few users, to prevent anyone from being identifiable. It happens mainly with demographic data and search queries: widening the date range or removing those dimensions usually solves it.