Frequently Asked Questions

UITruth measures how your brand and competitors appear in real, web-searching AI answers to Japanese-language questions. ChatGPT is the default; Gemini (real UI) is also available. Perplexity's real UI is not measured.

Two levels. First, a no-signup check you can try on the spot, up to 5 times in total (max 3 per day, no credit card needed). If you want to keep measuring, sign up for the free plan and get weekly auto-tracking of 1 question for your first 14 days, plus up to 15 checks a month. After day 14 the automatic weekly runs pause -- everything measured stays viewable, and manual checks continue within the same monthly allowance. Even on the free plan you get the core features: brand mentions and visibility into cited sources (the sites the AI referenced).

No. The company name or question you enter is never published in a form that says "someone measured this." The service does show a ranking of brands frequently named in AI answers, but only names the AI itself repeatedly surfaced across many answers make it in, and it's aggregated so there's no way to tell which user measured which brand. A single check is never exposed as-is.

Not necessarily. In fact, there's real value in finding that out. Knowing your current starting point -- that you're not yet being picked up by AI -- gives you somewhere to start thinking about where to place information so it gets picked up. The result screen also shows a next step to help with exactly that. Knowing where you stand is the first move.

The result screen lists next steps derived from that answer, each with an impact and effort guide -- things like getting your information onto a site the AI already cited, or building a comparison page that highlights the differences with a competitor. Start with whichever feels easiest from the top of the list. Measuring the same question again later lets you confirm whether it worked.

We cannot yet express accuracy as a single number. What we can publish is reproducibility -- how much a repeat measurement of the same question agrees with the last one. We publish agreement rates only once we have at least 5 questions and 30 consecutive pairs; below that we show the counts and no percentage. Miss rate, false-positive rate and alias resolution for brand detection are not measured yet because we do not have a hand-labelled ground-truth set, and we publish no estimate in their place. The measurement captures the real, web-searching ChatGPT answer, including body text and citation structure. AI answers shift with phrasing and time, so we recommend measuring several questions on an ongoing basis rather than judging from one result.

• View observed reproducibility

There's a free plan plus three paid plans, all shown in Japanese yen with nothing added to the listed price at checkout. Free includes weekly auto-tracking of 1 question for your first 14 days (it pauses after that; your results stay and manual checks continue) and up to 15 checks a month with no credit card required. Starter is ¥2,980/month for 5 weekly questions and up to 100 checks, Pro is ¥9,800/month for 20 weekly questions and up to 300 checks, and Agency is ¥29,800/month for 100 weekly questions and up to 1,000 checks.

• View pricing plans

Yes. The free plan includes weekly auto-tracking of 1 question for your first 14 days and up to 15 checks a month with no credit card required, and gives you the core features -- brand mentions and visibility into cited sources -- at no cost. After day 14 the weekly runs pause; your results stay and manual checks continue. We'd suggest starting free and considering a paid plan once you outgrow the volume.

Yes. There's a free check that needs no account and no credit card, up to 5 times in total (max 3 per day). Enter one question you want to check and we put it to the real ChatGPT, showing how you and your competitors appear in the answer. Start there to see how you look today.

Yes. You can move up or down as your measurement volume changes. It's fine to start small and expand as needed.

Paid plans are normally billed monthly by credit card via Stripe. For invoice payment, annual contracts, agency use, or a formal quote, we're happy to discuss individually -- reach us at support@uitruth.app and we'll advise based on your situation.

You can register and track multiple shops or brands in parallel within a single account (Pro and above suits ongoing multi-brand tracking). We don't currently offer separate logins per member or granular view/edit permissions. If you're considering team or multi-location use, contact us at support@uitruth.app.

UITruth pulls in the real answers ChatGPT actually gives, and analyzes how your brand and your competitors' brands appear within it -- specifically, brand mentions, cited websites, related questions (fan-out), local results for place-name questions, products and pricing, and ads. It's a system for repeating what you'd otherwise have to do by hand: opening ChatGPT and recording the answer yourself.

UITruth sends the same question a real user would type to the real, web-searching ChatGPT, and pulls the body text and citation structure of the answer that comes back as-is. We measure against the answer a user sees on screen, not the one a summarization API returns.

This result is captured with ChatGPT's web search turned on. Content and citations can differ from a plain answer without web search. UITruth measures under "search on" conditions to stay close to how people actually research things. The result screen also states this plainly with a "Captured with web search on" badge.

ChatGPT (real UI) is the default, and Gemini (real UI) is also available. Perplexity is not supported. If the supported set changes, the product will say so.

• View free tools

Because the current retrieval method is limited to one AI. UITruth captures the body text and citation structure of web-searching ChatGPT answers to Japanese-language questions. Gemini and Perplexity are not measured. If we widen coverage, the product will say so.

Yes, if there's a spelling or naming variation. Even if your official name is a full corporate name, the answer might just use a short version of it. We recommend registering common short names or service names as aliases -- doing so makes them easier to catch automatically when they show up in an answer.

Not currently -- automatic tone judgment (positive vs. negative) is out of scope. What UITruth measures is the visibility side: how much you were mentioned, which sources were used, and how you compare with competitors. That said, the AI's raw answer is shown as-is on the result screen, so you can read the context yourself and judge the framing firsthand.

No -- there's no automatic fact-checking. UITruth doesn't judge whether an answer is correct, only visualizes which brands appeared and which sources they came from. That said, since we surface the sources the AI relied on, if inaccurate information is circulating you'll have a clue about which site's content to fix.

Right now, we measure ChatGPT's answers to questions asked in Japanese. English questions and English answers aren't supported yet. Think of this as a tool for checking your visibility within Japan, via Japanese-language answers.

Yes. Register the competitor brands you want to compare against, not just yourself, and you can see how often each of you appears for the same question (share of voice). That lets you gauge your own presence relatively, not in isolation. Parallel monitoring of multiple brands is more fully featured on Pro and above.

The number of brands you can register, and how many you can monitor at once, depends on your plan. Pro and above suits ongoing tracking of multiple brands; for large-scale use like an agency, consider the Agency execution allowance.

You can cancel anytime from the settings screen, and billing stops from the next renewal date. Paid plans renew automatically each month, but there are no special conditions attached to cancelling.

No -- you keep access through the end of the period you already paid for. Paid features stop as of the next renewal date.

Yes, within 7 days of your first payment we refund it in full for any reason (once per account). Just email support@uitruth.app. We process the refund within 5 business days and cancel the paid plan at the same time. After that window, and for the second payment onward, we generally don't refund or pro-rate given the nature of a digital service. See the Commercial Transactions Act notice for the full conditions.

Rights to data you enter -- brand names, keywords, and so on -- belong to you. We use it only as needed to provide and improve the service. See the Privacy Policy for details.

Being upfront here: because of how measurement works, the questions and brand or name you enter are sent to the real ChatGPT via our external measurement platform (DataForSEO), in order to measure how you actually appear there. We never use the data you enter to train our own AI. That said, handling at the destination -- external services like OpenAI -- follows their own respective policies. We want to be transparent that sending data externally is a necessary part of measurement, not something we hide. Our list of processors and our approach to cross-border transfer are described in the Privacy Policy.

Through our external measurement platform (DataForSEO), we send that question to the real, web-searching ChatGPT and retrieve the body text and citation structure of the answer that comes back. Rather than receiving a summarized answer via API, we pull the real ChatGPT answer itself, so citation ordering and related-question structure are captured too. What we take in is the answer's text and structure, not a screenshot. Keep in mind the AI's answer reflects the moment it was captured and can change with phrasing or over time.

Yes -- execution history is saved and can be reviewed later. Running the same question on a schedule lets you build a record of how your AI visibility changes over time.

Account information, input data, and measurement results are stored in our database platform (Supabase), hosted in Japan (Tokyo region). That said, we also use overseas services for measurement processing (DataForSEO), payments (Stripe), and hosting (Vercel), so some data may be handled in regions outside Japan through those services. In other words, not everything is stored domestically -- some information moves overseas depending on purpose. Our list of processors and our approach to cross-border transfer are described in the Privacy Policy; see there for specifics.

• View security details

CSV export is available -- use "Export CSV" on the measurement history screen (it opens directly in Excel). You can also print a single result as a one-page PDF from the measurement detail screen. An externally callable API is not available yet. If you'd find one useful, let us know at support@uitruth.app and we'll factor it into the roadmap.

Yes. On the free plan and above, we email you a weekly digest of your automatic measurements at your registered address (in weeks with no change, we bundle it and send once every 4 weeks instead). On the free plan the weekly auto-tracking runs for your first 14 days, so that is how long the digest arrives. If you'd rather not receive it, click the unsubscribe link in any of those emails, or turn it off any time from Settings > Email notifications. Once it's off, nothing further is sent. We do not send marketing newsletters.

Working to rank higher in a search engine is called SEO (search engine optimization). In the same way, working to make sure your brand appears correctly within a generative-AI answer is called LLMO (generative-AI optimization) -- sometimes also called GEO. UITruth handles the first step of LLMO: measuring how you appear right now.

Yes -- it's designed to be workable by one person, part-time. Start with the no-signup check (up to 3 a day), and move to the free plan (1 weekly question for your first 14 days, plus up to 15 checks a month) once you want to keep going. No need to overcommit.

A shop, company, or service name. Not limited to companies -- it covers place names, organization names, and individual or creator names too: anything you want to look for within an AI's answer.

A website or page the AI referenced while building its answer.

Related or expanded queries generated by the AI from the original question. They are not queries entered by users and do not indicate user interest or behavior.

Within the same question, the ratio of how often you and your competitors were each brought up.

Generative Engine Optimization: work intended to help a brand or its information be appropriately referenced in generative AI answers.

Answer Engine Optimization: structuring information so it can be found as a direct answer in search and AI answer surfaces.

LLM Optimization: work to make sure your brand appears correctly within a generative AI's answers. Think of it as the AI-era version of SEO. Difference from GEO: both name the same goal of being referenced appropriately in generative-AI answers, with little practical difference — LLMO is the common term in Japanese, GEO in English.

The question text you send to the AI. UITruth registers this question text in order to measure it.