Google Analytics without analytical jargon

Google Analytics is meant to help make better decisions about the site, campaigns and sales. That’s the theory. In practice, it’s not about staring at charts, but about understanding where users come from, what they do after landing on the site, and where they simply drop off. Well-configured measurement shows which traffic sources deliver contact or a purchase, and which only inflate visits without any effect. Installing the code alone does not solve anything if you have not first determined which user actions are truly important to the business. In this article, I break the topic down without analytical jargon, but with an emphasis on an implementation that can be maintained over time. The key points will be: what to measure, how to set it up and what to watch out for so the data is not just nice, but useful.

What is Google Analytics in practice?

Google Analytics is a system for measuring user behaviour on a website or in an app. It collects information about visits, transitions between subpages, clicks, scrolling, form submissions, purchases and other actions that can be tied to a business goal. This makes it possible to check not only how many people visited the site, but also whether they did something valuable.

The reports themselves are not the most important thing. What matters are the questions the tool is meant to answer. The point is, for example, whether ads are actually generating leads, which subpages push the user towards a decision, and where the journey breaks down. And if you do not know what you are looking for, even correctly collected data remains just noise.

Today the standard is GA4, the version based mainly on events. This shifts the mindset from the model of “visits and goals” to specifics: the user clicked a phone number, started a form, submitted a form, added a product to the basket, completed a purchase. It is these actions that build meaningful measurement. So the question is not “what does Analytics show”, but “what is the user actually doing”.

There is one more catch. Data from Google Analytics is not a perfect reflection of reality. Its quality is affected by user consent, script blocking, browser settings and the way the cookies banner is implemented, and that can shift the results in surprising ways. That is why reports are read as an approximation of trends and problems, not as a 100% register of every click.

What are the key elements of Google Analytics implementation?

Google Analytics implementation rests on five things: the measurement goal, the event plan, correct technical implementation, testing and order in the data. Without this, the tool usually collects a lot, but very little comes from it. The most common mistake is measuring everything at once instead of focusing on a few actions that are truly important to the business.

First, you decide what is to count as a conversion. For a service website, this will usually be a form, a phone call, a click on e-mail or a booking for a consultation. In an online store, there are additional stages along the way, such as adding to the basket, starting payment and completing the purchase, while in more complex sites there are also transitions between domains, a payment system, a booking flow or an external form. And that is where the real work begins, because each of these steps has to be captured unambiguously in the data.

• Measurement goal — determining which user actions really matter.
• User journey audit — checking how the user moves through the site, at which points they may drop out of measurement and which areas are critical.
• Event and conversion plan — a written, concrete list: what to measure, what to call it and which parameters to pass.
• Technical implementation — adding GA4, usually through Google Tag Manager, and configuring events and campaign tagging.
• Testing — checking whether events fire at the right moment and do not duplicate data.
• Data organisation — excluding internal traffic, trimming spam, consistent naming and a sensible conversion list.

The way it is implemented also matters a great deal. When a site runs on several domains or subdomains, has an SPA app or uses external tools, a simple tag installation usually does not close the subject. The problem is that the user does not move within one “box”. You need to make sure they are measured correctly even when they switch between different environments.

Finally, it is a good idea to connect Analytics with other data sources, for example ads, Search Console or CRM. Then you can see not only the traffic, but also its quality and real sales effect. If campaigns are tagged incorrectly and events are named inconsistently, even good technical data loses its decision-making value. And that is not a cliché.

How does Google Analytics work step by step?

Google Analytics works simply. First, you decide which user actions matter to the business, then you implement their measurement, test the data and only then draw conclusions. The starting point is not technical, but business-related. The question is: what should count as an outcome — form submission, a phone call, a click on e-mail, downloading an offer, a booking or a purchase. If you do not define this at the beginning, the tool will collect a lot of data, but little of it will help with decisions.

The next step is to check how the user really moves around the site. What matters are visits to the landing page, transitions to the offer, forms, basket, blog, payments — and all the places where the user may disappear from measurement. On this basis, a measurement plan is created: a list of events, parameters and conversions, plus a decision on what should be measured automatically and what needs to be set up additionally. Without this, it is easy to fall into collecting “everything”, which in practice means nothing useful.

Then comes the technical stage. The GA4 tag is added, most often through Google Tag Manager, and specific events and UTM campaign tagging are set up. In practice, you measure not only page views, but also actions: form submission, clicking on a phone number, downloading a file or making a purchase. Good implementation is not about measuring everything, but about measuring a few actions that genuinely lead to contact or sales.

Implementation is only the start. After that, you need to ruthlessly check whether the data is correct and whether the measurement is not telling fairy tales. You test whether events fire at the right moment, do not duplicate and pass sensible values, for example the form name, order value or contact type. At the same time, you tidy up the data: exclude internal traffic, cut spam, set key conversions and connect the account with ads or other data sources.

And then the real work begins. Analysis and optimisation, without the pomp, but based on concrete numbers. You check which channels deliver valuable traffic, which pages push the user further, where they drop off and which devices or sources have the weakest performance. The question is: what is really blocking the result. On that basis, you improve landing pages, forms, the basket, CTAs and campaigns, instead of twiddling the knobs blindly. The most useful reports answer specific questions: where the conversion came from, where the path broke down and what is worth improving first.

What are the current challenges and limitations of Google Analytics?

Google Analytics has a few hard limitations today. And that is not a cliché. They result mainly from privacy, user consent, script blocking and the complexity of implementation itself. The standard is GA4, which works in an event-based model, not the old approach based mainly on sessions and goals. That gives greater flexibility, but it requires a better measurement plan and iron discipline in naming. Without that, reports quickly become impossible to compare.

The biggest practical problem is simple: the data is not a full reflection of reality. Some users do not give consent to tracking, some use browsers that limit cookies, and some block analytics scripts. On top of that comes the cookie banner implementation, which can significantly change how much data gets into GA4 at all. But note, that does not disqualify the tool, only changes how you read the results. Reports in Google Analytics need to be read as an approximation of trends and proportions, not as a perfectly exact record of every action.

The second limitation affects those sites that do not live in one simple environment. If a user moves between domains, lands in an external form, payment system, booking system or an SPA application, measurement requires additional configuration. Without it, sessions can split, traffic sources can be overwritten, and conversions can land in the wrong place or not land at all. And here a simple progression applies: one transition is a risk, several transitions are a problem, and many transitions are chaos. The more transitions between different systems there are, the greater the importance of post-implementation testing and testing after every change on the site.

There is also a third thing, often swept under the carpet. GA4 on its own rarely is enough for a full assessment of marketing and sales, because it shows what happened on the site, not always what happened later. The tool will record clicks and forms, but it will not always connect them with what happened afterwards in the CRM, in sales or after a phone call. That is why the sensible practice is not GA4 instead of the rest, but GA4 together with Google Ads, Search Console, CRM or a sales system. The facts are these: only by connecting these pieces do you get a picture that can be defended in a budget conversation. Only connecting the data on traffic with lead and sales data allows you to assess which channels really have business value.

What decisions are key when implementing Google Analytics?

First, definitions. The key decisions are about what exactly you consider a valuable user action, how you measure it and who is responsible for data quality. This is not a technical detail, but the foundation of the implementation and later conclusions. If a business wants contacts, measure contact, not “nice traffic”. If sales matter, measure the full path to purchase instead of stopping at page visits alone.

The first important decision concerns measurement priorities. A simple service website usually needs a few key conversions, such as form submission, clicking the phone number or email. And here discipline is enough. A shop, booking platform or more complex site already requires a broader plan: basket stages, payment, form errors, logins, offer selection and cross-domain transitions. Without that, you are only looking at a fragment of the film, not the whole story.

The second decision is the implementation method. Google Tag Manager is most often chosen because it makes it easier to develop measurement, run tests and make changes without having to touch the site code every time. It sounds convenient and it is. It makes particular sense when the site has dynamic elements, several forms, advertising campaigns and the need to add new events quickly. The question is whether you want measurement “hard-coded” or something you can maintain in motion.

The third decision concerns the naming of events, parameters and campaigns. Without consistent names, the data quickly becomes impossible to compare, and reports stop answering simple questions such as: which form works better or which campaigns bring real contacts. And that is not a cliché. In practice, it is worth setting consistent rules for event names, UTMs and the values passed to GA4 straight away, because later “tidying up” usually costs more than the implementation itself.

The fourth decision concerns the technical scope of measurement. You need to establish whether the user moves only within one domain, or also goes through subdomains, an external basket, payment system, booking calendar or embedded form. There is no room for guesswork here. If such transitions are not configured correctly, some conversions may be attributed incorrectly or disappear from reports. And then the hunt for a “drop in sales” begins, even though it may not be a drop at all.

The fifth decision concerns privacy and consent. In the current model, data quality depends on how the cookie banner works, when tags are fired and whether measurement respects user choices. This is practice, not theory. As a result, implementation needs to be planned together with the person responsible for the site, advertising and legal compliance, rather than treating GA4 as a separate add-on that will “somehow be connected up”.

The sixth decision is organisational, but it can make or break the outcome. You need to establish who has access to the accounts, who publishes changes, who tests the implementation and who is responsible for the later review of the data. Without an owner, the topic gets diluted. Even a good implementation loses value if, after changing a form, page template or payment system, nobody checks whether the measurement still works correctly. And then you pay for data you cannot trust.

The last important decision is what to connect Analytics data with. The tool itself will show traffic and behaviour, but only linking it with ads, Search Console, CRM or the sales system gives a fuller picture of channel quality and campaign effectiveness. The key is to slot the pieces together into one puzzle. That makes it easier to distinguish traffic that only generates visits from traffic that genuinely supports sales, in other words delivers what really matters in the business.

What are the most common mistakes and how do you avoid them?

The most common set of sins. Measuring everything without priorities, inconsistent naming and implementation without testing make the pile of data in the tool grow, while answers are still nowhere to be found. The problem is that it is usually not GA4 that fails, but a chaotic measurement plan.

The first typical mistake is collecting dozens of events without distinguishing which ones really carry business weight. Clicks, scroll, page views and micro-behaviours can help, but they should not obscure what you are measuring in the first place, namely key conversions. It is better to measure 5 key actions well than to superficially track 50 things that nobody analyses afterwards.

The second mistake is prosaic and painful. Inconsistent naming. If one time you save the form as generate_lead, another time as form_submit, and campaigns use different UTM standards, comparisons start to drift and confusion appears instead of conclusions. The question is who is supposed to keep this under control. It is brought into order with simple documentation: a list of events, parameter definitions, one naming convention and a clear rule for who can change it.

The third mistake is the lack of testing after implementation and after every change on the site. Events can duplicate themselves, fire too early or fail to pass important values, such as the form name, contact type or purchase amount. Great that it “comes through”. But be careful, that is only the beginning. In practice, you need to check not only whether the event “comes through”, but also whether it comes through at the right moment and with the correct data.

The fourth mistake is pretending that the data are perfect. GA4 does not show reality perfectly, because it is influenced by user consent, ad blockers, browser settings and the implementation technique. The data make it clear that this is not a “full record of the world”, but only its useful approximation. That is why you should read reports as a compass, not as a transcript of every action.

The fifth mistake becomes apparent when the user moves between environments. When a site uses an external form, payment gateway, subdomains or a booking tool, it is very easy to lose measurement continuity and get wrong conversion sources. And this is not a detail. You do not fix this later with a report, but with earlier cross-domain configuration and a thorough test of the whole journey.

The sixth mistake is looking only at the number of visits. Traffic alone says very little if you do not know whether users reach the offer, start the form, abandon the basket or return later from another source. Let us look at it differently: volume is only the backdrop, and the meaning lies in quality. In practice, more useful than volume alone are traffic quality indicators and drop-off points, because they show what is worth improving on the site or in the campaign.

The seventh mistake is the lack of regular review after changes. A new landing page, a redesigned form, a different consent system or a new campaign structure can change the way data are collected from one day to the next. Who will catch that if there is no inspection routine. That is why a sensible working model is not a one-off implementation, but a periodic check of measurement, the dashboard and traffic tagging rules.

What should you measure and how should you optimise results in Google Analytics?

In Google Analytics, what matters is what leads to contact, a lead or a sale. The rest can be noise. The number of users alone rarely helps you make a sensible decision if you do not know whether that traffic is doing anything valuable. That is why measurement should cover both the main conversions and the earlier steps that show whether the user is even getting closer to the goal. The question is not “how many came in”, but “who got further”. The most practical approach is to measure not everything, but what allows you to assess traffic quality and quickly identify the places where the user drops off.

• main conversions: form submission, phone call, email click, purchase, booking, sign-up, brochure download,
• micro-conversions: starting the form, going to pricing, clicking a CTA, adding to basket, selecting a date,
• traffic quality: engagement, visits to key subpages, time to complete an action, share of new and returning users,
• paths and drop-offs: basket stages, form errors, exit points, drops between steps,
• differences between segments: devices, traffic sources, campaigns, landing pages, locations and new versus returning visits.

Optimisation starts where you can see crowd and silence. In other words large traffic, and then emptiness. If a given entry page attracts many users but has few clicks through to the offer, form or basket, the problem is that something is not delivering on the promise: the content, the page layout, the CTA or the match between the campaign and the content. Instead of pumping in more visits, it is better to check where the intent and the execution diverge. Do not optimise the site for more visits if you have not first checked why the current traffic is not moving to the next step.

In service websites, you most often start with forms, contact sections and offer pages. That is where the user makes the decision, not in the menu. In online stores and booking systems, you need to look more broadly: product, basket, delivery, payment and return after payment, because each of these stages can kill the conversion. But be careful, the differences between mobile and desktop usually say more than any “averages”. If users are dropping off massively on mobile and completing the process more often on desktop, that is a signal to check form usability, page speed, interface errors and how the buttons actually work.

A lot of bad decisions come from staring only at the acquisition channel. It is tempting, because it is simple. You should assess the traffic source not by the number of visits, but by whether users perform important actions and how far they get in the process. The data make it clear that “a lot” does not always mean “good” when the number of drop-offs rises and session quality falls. A campaign with less traffic may be better than a large one if it delivers more valuable contacts, fewer drop-offs and better user quality.

To make data genuinely optimisable, you need to compare it over time and across segments. Otherwise, you are comparing apples with random numbers. Consistent event names, disciplined UTM campaign tagging and a fixed list of the most important conversions help with this, without adding new “important” actions every week. If the naming is chaotic, the report looks like a set of numbers, not a decision-making tool. In practice, it is best to regularly review a few views: landing page performance, the quality of traffic sources, form or basket abandonment, and differences between devices. And only then optimise, not the other way round.

In the end, what matters is not the report, but the traffic on the site or in the campaign. After the analysis, you need to form a concrete hypothesis and immediately “translate” it into a change: shorten the form, reorder the sections, strengthen the CTA, swap the ad copy, fix campaign tagging, or simplify the purchase path. Because why collect charts. Google Analytics only makes sense when the data points to the next step, not just another slide for a presentation.