Research in progress
CasaNova is a real e-commerce project that Sageo Estudio uses as an environment for research and experimentation. We do not present it as a success story or as a finished story, because it is not one: it is the place where we observe, test and document how SEO, AEO, GEO and Search Intelligence interact on an e-commerce business that is genuinely operating.
This is the first installment of that line of work. It explains why we research on a project of our own, which questions are open and how findings are recorded. It does not anticipate results, because there are still no results that can be rigorously published.
Applied research is usually done on client projects, and there is a structural problem there: the scope is agreed in advance and the time is limited by a budget. You can measure before and after, but you cannot change course halfway through just because an observation justifies it. Every decision has to be justified to someone else, and many interesting questions stay out of the contract simply because nobody asked for them.
Working on a project of our own removes those restrictions. You can observe the full cycle —decision, implementation and effect— without negotiating every step. If a hypothesis requires changing the category architecture, rewriting a product page or changing how the structured data is declared, it is done and documented. If the result contradicts what we expected, that is documented too, and that contradiction is useful information instead of a problem to be hidden.
There is an additional reason, a less comfortable one. It is easy to repeat what the industry says about how search engines and AI systems behave. When the work is done on third-party data, you never quite know which part of the explanation is observation and which part is someone else's interpretation. An environment of your own forces you to distinguish between what was measured and what is assumed.
The advantage of a project of our own is not having more freedom to do things: it is being able to hold a claim to the end and know what evidence supports it.
Several lines of work are open at CasaNova, and they intersect. None is closed and each one has its own baseline still to be consolidated.
How presence in search engines behaves on a catalogue that changes: which pages are discovered, which are indexed and in which queries the content appears. What matters is the trend and the proportion, not a single value read as certainty.
Which queries arrive with buying intent and how they are phrased. Commercial intent is not always written the way you expect, and that difference defines what type of page should answer each query.
How the category architecture is organized and what relation it has to the way content is crawled, grouped and understood. It is one of the lines where a structural decision affects everything else.
How the brand is represented and how each product is represented: what it is, in which category it competes and with which attributes. If those signals contradict each other between pages or between channels, the system cannot state anything with confidence.
Indexing, semantic markup, cannibalization, pages duplicated by filters and variants. A real catalogue produces problems that a small site does not reproduce, and those problems are the ones worth studying.
What a generative system knows about the brand, how it describes it and whether what it says matches what the brand declares about itself. Here we observe both the mention and the accuracy of what is mentioned.
The relationship between what is published and what a system answers. It is the most difficult line: the generated answer varies by system, language, location and session, so it is read as a share and not as an absolute value.
Documenting is the part that decides whether the work is useful or not. Every finding is logged with its context and its limits, never as a general conclusion. An observation made on a catalogue in one specific category does not become a statement about e-commerce in general.
This log lives in internal working notes. It is published when the method and the observations are up to standard, not before. And when it is published, it will state the period worked, the real scope of the intervention and the source of the data.
Research in development. This line of work is in progress and is not finished. We do not present results or conclusions because there are still no publishable measurements, and no figure will be published until it has a verifiable period, scope and source. What appears here is what is researched, why and with what method — not a performance record.
The goal of this line is not to accumulate anecdotes about an e-commerce business of our own. It is to turn what we learn into methodology: working criteria that can later be applied to any project, with the advantage that they have already been tested in an environment where it was possible to make mistakes and correct them.
Part of that learning has a specific destination. SAGEO Intelligence is the platform for analyzing brand presence in AI systems that we are developing, and it is still under construction. The research on CasaNova feeds that development with real questions: what is worth observing periodically, how a generated answer is logged so that it is comparable with the next one, and which signals about the brand are stable enough to become an indicator. When the product is not finished, the research is what defines what it should measure.
The rest of the learning goes back to the method. Search Intelligence is a cycle —research, measure, interpret, intervene and measure again— that only works when it is run through in full. A project of our own makes it possible to complete it without skipping steps because of contract limits.
The principle that governs all of this fits in one sentence: we research while we build. Research is not a previous phase that ends to make way for execution, nor a later phase that confirms what has already been decided. It advances together with the project, and in many cases it forces it to change direction before the mistake becomes expensive.
That implies accepting a discomfort. A project in progress does not produce a clean story with a beginning and an end. It produces better-formulated questions, observations with a date and declared limits, and criteria that are adjusted when the evidence does not back them up. We prefer to publish it like that: without closing what remains open and without presenting as a result what is still an observation in progress.
The next installments of this series will deal with specific lines, with their method and their limits. If any measurement becomes publishable, it will appear with its period, its scope and its source.
Rate this crystal
Instead of stars we use gems, which is what forms inside a geode. Your rating helps show which crystals are worth polishing further.
Loading ratings…
A comment can correct, expand or challenge what the article says. It is published after review.
Loading comments…
