About SiteUpward

Who runs SiteUpward

SiteUpward is provided by SIA "MICRON", a limited liability company (SIA) registered in the Register of Enterprises of the Republic of Latvia under number 42103081578. SIA "MICRON" operates the service and is the company you contract with when you run a check or buy an audit.

The same company details appear in the Terms of Service and in the footer of every page of this website.

What SiteUpward does

SiteUpward audits public websites for AI search. An audit shows whether the crawlers behind six AI engines – ChatGPT search, Perplexity, Google AI Overviews, Gemini, Claude and Microsoft Copilot – are allowed in, whether the text is already in the raw HTML (compared with a browser-rendered sample of up to 5 pages), and whether the pages state clear, specific facts that an AI answer can quote.

A free check covers up to 25 pages. It shows the AI-readiness score, which AI engines can reach the site, every site-wide issue and full fixes for 3 pages. A full audit covers up to 500 pages and a large audit up to 2,000 pages; in both, every finding shows where it occurs, why it matters for AI answers and what to change.

Prices and what each audit includes

Who SiteUpward is for, and who it is not for

SiteUpward is designed mainly for businesses and professionals: website owners, marketers, SEO specialists and agencies. Customers can use it from any country, and prices are shown in US dollars or euros.

SiteUpward is not the right tool for these needs:

• Tracking how often AI assistants mention a brand over time. SiteUpward audits the website itself; AI-visibility trackers monitor the answers.
• Content that is only available after signing in. SiteUpwardBot does not log in or submit forms, so it only sees public pages.
• Legal, accessibility-compliance or security audits. SiteUpward reports cover AI search readiness and technical SEO and are not professional advice in those areas.
• Guaranteed rankings or citations. No audit can promise them (see the limits below).

How an audit works

1. Access tests. SiteUpward reads the site's robots.txt and evaluates its rules for 16 AI and search user agents, including OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot and Bingbot. It also requests the homepage with the user agents of well-known AI and search crawlers to find out whether a CDN or firewall blocks them.
2. Crawling. SiteUpwardBot reads the sitemaps and fetches the HTML pages of the audited website, with at most 2 requests per second and 2 parallel connections per website. It identifies itself with its own user agent, follows robots.txt and does not follow links to other websites, submit forms or log in.
3. JavaScript test. Up to 5 pages are also loaded with JavaScript enabled, the way a visitor's browser loads them, to measure how much of the text is already in the raw HTML. Most AI crawlers do not run JavaScript, so text that appears only after scripts run is invisible to them.
4. Page analysis. Rule-based checks measure titles, headings, status codes, canonicals, structured data and similar signals. An AI evaluation model reads each page and evaluates its content.
5. Score and report. The results become four category scores and an overall score from 0 to 100 with a grade from A to F. Each problem becomes a finding with a priority, the evidence from the page and a suggested fix.

How SiteUpwardBot crawls and how to block it

What SiteUpward checks

SiteUpward runs 81 checks in four areas, and each area has a fixed share of the overall score:

Of the 81 checks, 53 are rule-based measurements, 22 are answered by the AI evaluation model and 6 combine both methods. 56 checks look at individual pages and 25 at the website as a whole.

Three problems cap the overall score, however good the rest of the site is. If robots.txt blocks one of the crawlers that AI search depends on, such as OAI-SearchBot or Googlebot, the score is at most 49. If, on average across the pages rendered in a browser, less than 30% of the text is in the raw HTML, it is at most 39. If the whole site is set to noindex, it is at most 19.

How checks are graded by evidence

Advice on AI search varies widely in how well it is supported. SiteUpward grades every check by the strength of the published evidence behind it – the engines' own documentation, peer-reviewed research and large industry studies. Checks with stronger evidence generally carry more weight in the score, and checks without evidence carry none:

Checks answered by the AI evaluation model show how confident the model was. A yes/no answer with a probability between 30% and 70%, or a graded answer with less than 50% confidence, is marked “Review”: the report asks you to check the page yourself, and the finding counts only half in the score.

These grades describe the checks. In the report, each finding is labelled by severity (Critical, High, Medium, Low or Info).

The “Why it matters” text of each finding explains the reason for the check and, for many checks, names the study or search-engine documentation it is based on – for example the GEO study (KDD 2024) or Google's documentation of the nosnippet directive.

Principles

• No invented numbers. SiteUpward does not publish made-up statistics, customer reviews, testimonials or ratings. Statistics from other organisations on this website name their source.
• No myths sold as fixes. Tactics without evidence, such as adding an llms.txt file, are reported as information only and do not change the score.
• A clearly named crawler. SiteUpwardBot identifies itself with its own user agent, follows robots.txt and limits its request rate on every website.
• Only public content goes to the AI model. The AI evaluation model receives the address, title and text of public pages, never account or billing data.
• No training on audited pages. Pages fetched for an audit are not used to train AI models, and SiteUpward does not keep full copies of them.

Privacy Policy

What an audit cannot tell you

• Whether AI engines will cite a site. No one can guarantee citations, rankings or traffic. SiteUpward does not control ChatGPT, Perplexity, Google, Gemini, Claude or Copilot, and their behaviour changes often. The score estimates AI-readiness; it does not measure how often a site is cited.
• Off-site signals. Mentions on other websites, reviews on independent platforms and discussions in online communities also influence AI answers, but an audit of the website cannot measure them. Instead, the report gives a checklist of the ones that matter most for the type of business.
• Pages it cannot see. Results are incomplete for pages behind a login, pages that robots.txt closes to SiteUpwardBot and content that changes during the crawl.
• Certainty about AI-judged findings. AI-assisted assessments can be incomplete, differ between runs or be wrong. Review recommendations before acting on them.

Terms of Service

Contact

SIA "MICRON" can be reached by e-mail. Please write to the address that fits your question: