How we use data to improve websites over time

In short: this article describes how at PaperPlane we use Google Analytics 4, Search Console and Microsoft Clarity to understand what really happens on a site after it goes live, and how that data guides the choices we make to improve it. The central point: the best decisions don’t come from instinct or from the opinion of whoever has the loudest voice, but from observing how real people use the product.

When we started working on the VolumeBK project, the e-commerce born from the merger of SpazioBK and VolumeVolume, one of the starting points was the existing SpazioBK site. On paper, the search filters worked: they were there, labelled, reachable. Nobody had ever flagged them as a problem. And yet Laura, founding partner and UX/UI expert at PaperPlane, watching the Microsoft Clarity session recordings, was telling a different story. Some users clicked “Filter by” expecting to open a panel, not to select a category directly. Others, once the filter menu was open, tried to close it by clicking outside the area, a behaviour the site didn’t support. Small frictions, repeated, invisible to those who work on that site every day but plain to see for those who know where to look.

Those observations steered the complete rethinking of the filter system in the new VolumeBK: more meaningful language, more explicit interaction areas, clearer visual states. Not because someone had a more authoritative opinion on the right labels, but because the data showed where people got stuck.

The site goes live. And then?

There’s a moment in every web project when there’s a tendency to consider the work finished: the site is online, the content is there, everything works. It’s an understandable feeling. But it’s also the point at which you stop learning.

A site isn’t a printed poster. It’s a digital product that people use, and real use rarely matches the intentions of whoever designed it. Some sections that seemed central get ignored. Some pages thought to be transitional become important landing points from organic traffic. Some calls to action don’t convert, not because they’re badly written, but because they sit in a spot almost nobody reaches.

All of this you discover only by looking at the data. That’s why at PaperPlane, when we work on a project on an ongoing basis, we offer a continuous optimization service that integrates three tools: Google Analytics 4, Google Search Console and Microsoft Clarity. Three tools with distinct roles, that together cover different angles of the same problem.

Three tools, three different questions

The temptation, when talking about “site data”, is to think it’s enough to have Google Analytics installed. Analytics counts the visits, says where they come from, shows which pages get viewed most. It’s useful. But it isn’t enough, because it answers only part of the questions that matter.

Google Analytics 4 answers the question what happens: how many people visit the site, which pages they reach, where they leave, whether they complete the actions we care about, a contact request, a purchase, a newsletter sign-up. It’s the aggregate view of behaviour, indispensable for understanding the structure of traffic and measuring results over time.

Google Search Console answers the question where they come from and what they’re searching for: which queries on Google bring people to the site, which pages they land on, how those pages rank against potential customers’ searches. It’s the tool that connects the site to the reality of the market, and that lets you understand whether the content is intercepting the right people at the right moment.

Microsoft Clarity answers the question how they really behave: where they look, where they click, how far they scroll, where they get stuck. Heatmaps aggregate the behaviour of thousands of sessions into a single image. Session recordings let you watch, in anonymised form, the navigation of individual users, with all the hesitations, the double clicks, the backtracks that aggregate numbers hide.

None of the three tools, on its own, tells the complete story. Analytics tells you a page has a high exit rate; Clarity shows you where users stop on that page and what discourages them. Search Console tells you you’re getting traffic for a certain query; Analytics tells you whether that traffic then does anything useful. Reading them against each other is what produces actionable information.

Why Clarity deserves a discussion of its own

Microsoft Clarity is free, installs in a few minutes on any WordPress site, and offers features that until a few years ago were the preserve of expensive enterprise tools. And yet it’s still underused: many sites have it installed, few use it systematically.

The reason we consider it indispensable at PaperPlane is simple: it’s the only tool that shows behaviour, not just the result. The difference between knowing that 65% of users leave a certain page and seeing a user try twice, scroll up and down looking for something they can’t find, and then leave, is the difference between a number and a story. Stories guide design decisions far better than numbers alone.

On the VolumeBK project, Clarity’s session recordings revealed the filter problem we mentioned at the opening. The heatmaps showed which elements of the product page captured attention and which were ignored. That visibility into real behaviour made the redesign choices far more solid than any internal discussion about “what works best” would have.

One technical detail worth mentioning: Clarity integrates natively with Google Analytics 4. This means you can correlate the qualitative data, a specific session, a pattern of behaviour, with the quantitative data from GA4, filtering the recordings by user segments or by specific events. It’s a powerful combination, and still little exploited.

Data is useless if it stays in a drawer, and opinions cost

Here’s the point worth stating clearly, even at the risk of seeming awkward.

Most sites have Google Analytics installed. A much smaller share look at it regularly. A smaller share still use what they see to make concrete decisions. Having the tools active isn’t the same as doing analysis: it’s the same as having a thermometer on the wall and never looking at it.

The opposite problem exists, and it’s just as real: accumulating data without a method for interpreting it leads to paralysis. Too many numbers, too many charts, too many metrics that all seem important. The analysis becomes an end instead of a means, and meanwhile nobody decides anything.

The point of balance is asking the right questions before opening the tools. Not “what does the data tell us?” in the abstract, but “we want to understand why this page doesn’t convert” or “we want to know whether users find what they’re looking for in the navigation menu”. A specific question produces a specific analysis, which produces a specific action.

The other advantage of data, perhaps the most underrated, is that it shifts the conversation onto a different plane. In every project there are opinions: the client has a preference, the designer has another, whoever writes the content has a third. Opinions aren’t wrong, they’re part of the process. But when there’s data showing how real people behave, the question is no longer “I think it works like this” but “what do the people using the site tell us”. It’s a change of register that makes decisions faster, more shared and, in our experience, more often correct.

How we work: analysis first, even on sites we didn’t build

When we propose a continuous optimization service on a new site, integrating Analytics 4, Search Console and Clarity is the starting point. They’re installed, configured correctly, which isn’t trivial, especially for GA4 where setting up events and conversions takes care, and a period of observation is established before drawing conclusions. A week’s data tells you little; two or three months’ starts to be meaningful.

But the moment these tools become even more valuable is when we take on the restyling or redesign of an existing site we didn’t build. In that case, the first thing we ask for, or install if it isn’t there yet, is access to the analytics tools. Designing without baseline data means working on assumptions: perhaps correct ones, but assumptions all the same. Having six months of real data on user behaviour completely changes the quality of the design decisions.

On the SpazioBK site, starting from the analysis of the heatmaps and session recordings let us identify precisely the critical points of the existing experience, the filters, as we said, but also the navigation paths, the pages with a high bounce rate, the areas of the site underused relative to the editorial investment they required. That information made the project brief far more concrete, and the final result far closer to users’ real needs.

A real limitation: data shows the what, not always the why

Being honest is part of how we work, so we’ll say it: data has a boundary.

Clarity can show you that 40% of users abandon the contact form halfway through. It doesn’t tell you whether they do it because the form is too long, because a required field is unexpected, because at that moment they were interrupted by something external, or because they simply changed their mind. To answer that question you need other tools, user testing, interviews, A/B tests, or a reasonable hypothesis to verify with a targeted intervention.

In the same way, Analytics can show you that the organic traffic of a certain page has dropped by 30% in three months. It doesn’t automatically tell you what to do: the cause could be a Google algorithm update, content that has lost relevance, a technical problem that reduced indexing, or a competitor who in the meantime has published something more complete. Interpreting that figure takes experience and context, not just the ability to read a chart.

Data is a lens, not an answer. It reduces uncertainty, but doesn’t eliminate it. And that lens, to be useful, needs someone able to ask the right questions, and to tell a signal from noise.

Where to start if your site doesn’t have these tools yet

First: install the tools and configure them well. Google Analytics 4 and Search Console are free and their value grows with time, every month that passes without data is a month of information lost forever. Clarity is also free and sits alongside GA4 without conflicts. The configuration of GA4, though, deserves care: tracking events and conversions requires conscious choices that depend on the site’s specific goals.

Second: wait before concluding. A week’s data is almost always misleading, too subject to random variation, too dependent on specific events. A minimum observation period of two or three months, better still over a full seasonal cycle for sites with seasonal traffic, is necessary to read meaningful patterns.

Third: start from a question, not from the data. Opening GA4 without a specific question in mind produces confusion, not insight. Choose a concrete goal, “I want to understand why the services page doesn’t generate contact requests” or “I want to know whether mobile users find what they’re looking for”, and use the tools to answer that question.

Fourth: cross-reference the sources. An anomalous figure on a single platform might be a tracking error or a random variation. The same signal that appears on Analytics, is confirmed on Clarity and finds an explanation in Search Console is far more reliable, and far more useful for deciding what to do.

When it makes sense to ask for help

Installing the tools is the easy part. Using them systematically, interpreting the data with the right context, and above all turning what you see into concrete interventions, that’s the part that takes time and method.

At PaperPlane the continuous optimization service is designed for those who want the site to genuinely work over time: not as a static asset delivered once and then left to itself, but as a product that improves on the basis of what real users show they want and do.

If you want to understand where to start, or if you already have the tools installed but don’t know what you’re looking at, we can talk it through in a 30-minute call. We bring your data, not our opinions.