What this log covers
This entry documents three technical updates to seenutech.com: a rebuilt sitemap, added structured data, and a clearer service-page structure. We publish these because we sell GEO work, and our own site should be the first place we prove what we recommend. None of this is a ranking trick. The point is simpler and more durable: make the site easy for both people and machines to read. AI answer engines like ChatGPT, Perplexity, Gemini, and Google AI Overviews do not cite pages they cannot reliably parse. When entities, services, and questions are ambiguous, a model either skips the page or summarizes it incorrectly, and you have no way to correct an answer you never see. So we treated this as a readability project, not a magic-words project. A clean, well-labeled page does not force a citation, but a messy one reliably prevents it, and removing those obstacles is the only part of the process we actually control. Below we describe exactly what we changed, why it matters for AI visibility, what we measured, and what is still open. We are deliberately not claiming new citations or lead volume from these changes yet, because the re-crawl window is still recent and we would rather show the work than overstate the result. Treat this as the first installment in an ongoing record, not a finished case study.
What we changed
First, the sitemap. Our previous XML sitemap missed several industry and service pages, which meant some of our most buyer-relevant content was only reachable through internal links. We rebuilt it so every service, industry, and blog page appears in one canonical list with accurate last-modified dates. Second, structured data. We added Organization schema so engines can resolve who Seenu Tech is, Service schema on each offering so the service name and description are machine-readable, and FAQPage schema where we answer common buyer questions. Third, page structure. We rewrote service-page headings to lead with the question a buyer would ask, then answer it in the first two sentences before adding detail. For example, our snapshot page now opens with 'What is an AI Visibility Snapshot?' followed by a direct, citable definition rather than a marketing tagline. Each change targets a different failure mode. The sitemap fixes discovery, so pages can be found. The schema fixes entity resolution, so a model knows what each thing is. And the answer-first structure fixes extractability, so the relevant sentence can be lifted cleanly. We sequenced them in that order on purpose, because there is no value in a perfectly worded answer on a page nothing can find, and no value in a discoverable page whose core facts are buried.
Why it matters for AI visibility
AI answer engines build responses by retrieving passages and reconstructing them into an answer. Three things make a page usable for that process. It has to be found, which is the sitemap's job. The entities on it have to be unambiguous, which is what Organization and Service schema provide. And the relevant answer has to be extractable as a clean, self-contained passage, which is what answer-first headings deliver. A worked example: before the change, a query like 'what does Seenu Tech do' had to be inferred from scattered marketing copy. After adding Organization and Service schema plus a direct opening line, the same fact is stated once, clearly, in a place a model can lift verbatim. We are not promising this guarantees a citation. We are saying it removes the obstacles that reliably prevent one. That distinction matters, and we hold to it, because the alternative is the kind of vague promise that has made buyers rightly skeptical of marketing agencies. A model is not persuaded by tone. It is far more likely to reuse a fact that is stated plainly, attributed clearly, and corroborated elsewhere, and our job is to make every one of those conditions easy to meet.
What we measured
We measured the inputs we control, not invented outcomes. We confirmed the sitemap validates with no errors and that every key page is listed. We validated each schema block with a structured-data testing tool and fixed the warnings until it passed clean. We resubmitted the sitemap and confirmed the re-crawl was acknowledged rather than assuming it. We also re-ran a small fixed set of buyer-intent prompts across the major AI engines to capture a current baseline of how, and whether, Seenu Tech surfaces. That baseline is the honest reference point for future entries. What we are not reporting is a jump in citations or leads, because the time since re-crawl is too short to attribute any change to these updates without misleading you. The measurement we trust right now is that the technical foundation is clean and verifiable.
What still needs work
Clean structure is necessary but not sufficient. Three items remain open. First, depth: several service and industry pages still need more concrete, citable detail, like specific examples and defined process steps, because models prefer specifics over adjectives. Second, corroboration: AI engines weigh consistency across independent sources, so our entity details need to match across our profiles and any third-party mentions, and that reconciliation is ongoing. Third, freshness signals: we need a steady publishing cadence on this blog so the site reads as actively maintained rather than static. We are treating these as the next phase, not as failures of this one. Being explicit about open items is part of how a proof log stays credible. If we only ever reported wins, this log would be marketing, not evidence.
Next action
Our next step is to add worked examples and defined process steps to the remaining service and industry pages, then re-run the same buyer-intent prompt set in two weeks to see whether the cleaner foundation changes how engines summarize us. We will report that comparison here, including if it shows little movement. If you want the same technical baseline for your own site, the fastest starting point is an AI Visibility Snapshot, which checks discoverability, schema, and answer-readiness, then tells you which fixes matter first. You can request one through our snapshot service page, or follow this proof log in the Living Lab to see how the work actually unfolds before deciding.