Website creation

Specific indicators and analytical instruments for measuring AI-generated traffic.

Measuring the effectiveness of AI optimisation strategies requires a fundamentally different approach from traditional analytics. Traffic generated through interactions with AI systems does not always appear in standard analytical tools in any obvious or legible form. Users who receive answers directly from AI assistants — answers built upon your content — may never visit your website at all, yet they are benefiting from your information and contributing to the construction of your brand's authority. This reality demands the development of an entirely new set of metrics and measurement instruments.

Traditional metrics — page views, bounce rate, time on page — remain relevant, but they must be supplemented by indicators specific to AI optimisation. These include the frequency of citations in AI responses (measured through mention-monitoring across various AI platforms), the quality of featured snippets (assessed by the position and context in which citations appear), and growth in referral traffic from unknown or unidentified sources that may well be generated by AI systems.

Specialist tools for monitoring performance in AI-driven search are beginning to emerge in the market, offering insight into how your content is perceived and utilised by different artificial intelligence systems. These include tools for tracking citations in AI responses, sentiment analysers for feedback generated by virtual assistants, and platforms that monitor brand presence within the AI ecosystem.

Building a comprehensive measurement system requires combining data from multiple sources: traditional analytics, social media monitoring, direct user feedback, and analysis of conversations with internal chatbots. For organisations serious about AI optimisation, investing in adequate measurement tools and processes is crucial — both for understanding the real impact of their efforts and for steering future strategy towards the areas with the greatest potential for improvement.

There is a moment when a law becomes something more than a text published in the Official Gazette. It becomes a practical problem. It becomes a question posed to a headteacher who doesn't know where to begin, a teacher staring at an open Google Drive folder wondering whether what she's doing is even correct, an IT technician at a provincial secondary school who suddenly realises that setting up a server requires rather more than knowing the administrator password.

That moment arrived in May 2026.

What happened in May 2026 — and why it matters

Through Order No. 3,858/2026 of the Ministry of Education and Research, published in the Official Gazette of Romania, Part I, No. 392 of 8 May 2026, the Framework Methodology for the Development and Management of the Professional Portfolio of Teaching Staff was formally approved. This is not mere bureaucratic circularity. It represents a genuine shift in how the state understands the documentation, evaluation, and archiving of a teacher's career.

The provisions are clear in their consequences. Beginning with the 2026–2027 school year, every teacher will be required to maintain a comprehensive professional portfolio containing documents relating to professional activity, assessment, continuing development, and classroom work — including in digital form. That portfolio will carry weight at decisive stages of the career: the definitivat examination, the award of teaching grades, annual performance reviews, school inspections, staff mobility, and the granting of increments and prizes.

This is not, then, an optional tool — a folder a teacher might build when the mood takes her, or when the headteacher persists. It is compulsory. Schools have no more than sixty days from the date of publication of the order to produce their own internal procedure — the deadline falling on 6 July 2026 — a procedure to be brought to the attention of all teaching staff and published on the school's website.

The Ministry has announced its intention to create an integrated national platform. The Secretary of State enumerated modules for standardised assessment, financial management, records and electronic registers, together with a module for teacher portfolios and pupil portfolios, the platform's stated ambition being to reduce school bureaucracy. At the time the order was published, the project was still under evaluation at the Ministry of European Projects and Investment.

In other words: the promised national platform does not yet exist in operational form. Schools must manage now, with what they have.

This tension — between the Ministry's centralising ambition and the decentralised reality of several thousand educational institutions spread across Romania — is precisely the space in which open-source platforms become relevant.

There is, in the relationship between user and technology, a fundamental tension that deserves to be explored not merely in technical terms but as a broader cultural phenomenon — a tension between what we imagine might be possible and what is, in fact, achievable within the material and temporal constraints of our present technological moment.

This gap is far more than a simple failure of communication, or a matter of insufficient technical literacy. It reveals something deeper about the way we understand and assimilate technological progress in contemporary culture.

Developers of components, plugins, and modules for Joomla — let us take this particular ecosystem as our vantage point, a microcosm of a much wider dynamic — find themselves in a paradoxical situation. Their users arrive with questions and requests that appear to have been shaped, in large part, by the vague, seductive, and often unsubstantiated promises of artificial general intelligence. This is a discrepancy not so much between technical competence and incompetence as between two fundamentally different ways of thinking about what digital technology actually is.

The Functional Utopia: The Case of the Omniscient Chatbot

Consider the first of these recurring expectations: the desire for a chatbot capable of answering "any question about the website." On the face of it, the request seems modest, even reasonable — after all, do we not speak daily with voice assistants, do we not receive instant answers to complex questions? But this apparent simplicity conceals layers of technological complexity worth unpacking carefully.

The user assumes, in a manner that is almost magical, that an AI chat module can substitute for human presence — that it can navigate the labyrinth of content, services, and functionality without prior training, without painstaking configuration. It is an expectation that reflects, in a certain sense, the central promise of contemporary AI: the elimination of friction between intention and execution, the disappearance of the intermediate effort.

But the technical reality — and here we arrive at the heart of the problem — is far more nuanced, far more capricious.

A functional AI chatbot does not emerge spontaneously from the installation of a plugin. It requires what developers call "vectorisation," "embeddings," "semantic mapping" — terms that, beneath the technical jargon, signify something fundamental: the transformation of human information, with all its ambiguities and nuances, into a form that algorithms can process.

 My father never asked questions. He simply fixed things. I can still see him, bent over an engine with his head all but buried in the innards of a Dacia 1300, both arms blackened to the elbow, listening. He never asked the customer anything. He never said "describe the sound to me" or "when exactly does the problem occur?" He listened to the engine directly, laid his palm flat against the block as though taking someone's pulse, and within a few minutes he knew. He diagnosed without questionnaires, without forms, without protocol. His knowledge was tactile — accumulated across years of mistakes and corrections — and it operated through resonance rather than interrogation.

I inherited from him a fascination with mechanisms. What I did not inherit was that gift for direct listening. I write. I formulate. I build questions and rebuild them whenever the answer that comes back is wrong — which means the question was not properly put to begin with.

And it was precisely that thought which came to me one evening and refused to leave: artificial intelligence works in the opposite direction to my father. It needs you to formulate correctly. If the question is vague, the answer is vague. If the question is formal, the answer is formal. If the question resembles the language of an official who has copied from another form that was itself copied from a third, the answer will read like a planning notice from 2004.

And yet — people do not ask good questions. Not because they are incapable, but because they do not know how to translate anxiety into clear prose. When someone types "can I get the increased child benefit?", they are not conducting a legal enquiry. They are making a survival gesture. They are looking for confirmation. They want to hear: yes, you can.

The FAQ — the Frequently Asked Questions section of a website — ought to be the place where that translation happens. In reality, it is, more often than not, the place where official language finds its most comfortable and least scrutinised refuge.