The full free AI audit: a 34+ slide report | Suparanku

A full audit — now free for everyone

We’ve opened the free AI brand audit to everyone. It is not a demo and not a “one-query check”: the free audit runs the same full research pipeline as paid scans, with the same methodology — just within the Free plan’s limits. Sign-up and the first scan require no card, and Free is a permanent plan, not a trial with a timer. What you get is not a number on a dashboard but a proper report deck: a 34+ slide PDF, generated automatically after onboarding and delivered to your inbox as a link.

Why it matters: more and more buying journeys start with a question to an AI assistant, not a search box. The audit shows how ChatGPT and Google AI Overviews answer about your brand, competitors and sources — and what to do about it. It is the first measurement of the closed loop of AI Representation Management, the discipline within which GEO, LLMO and AEO tactics operate.

What actually happens during the audit

Behind the word “audit” sit several parallel investigations — and the report makes the depth of each one visible:

1. Onboarding structures the business. Brand and spelling variants, topics and prompts — the questions real buyers ask (up to 15 prompts across 5 topics), competitors (up to 3 tracked). Plus market research: a business profile, the players and the choice criteria — these become report slides too.
2. The AI engine scan. Every prompt runs through ChatGPT and Google AI Overviews, three samples per prompt–engine pair, and the answers are analyzed across five axes: visibility, position in the answer, sentiment, competitors, sources. AI answers are unstable from run to run — independent SparkToro research found the odds of getting the same brand list in two answers are below 1 in 100. A single check is noise; the honest metric is frequency across many runs.
3. The site audit. A crawl of up to 1,000 pages: robots, structured data, meta tags, broken links, structure — every page and the whole domain get a 0–100 score (SRPS). Plus a deep check of the homepage, down to Lighthouse speed metrics.
4. Recommendations. The gaps from the scan and the audit become a prioritized action list with ready-made briefs.

Now the interesting part — how all of this looks in the report. Every screenshot below comes from a demo report for the fictional brand Mikazuki: the numbers, domains and competitors are all invented; only the structure of a real report is shown. The “Suparanku × Supasaito” header on the slides is partner co-branding: partners ship reports with their company name next to ours.

Slide 1: the whole picture on one page

Slide 1 of the demo report — the cover.

The report opens with a summary of three headline answers: how the site looks to AI crawlers (SRPS and Lighthouse scores), where you get found across the funnel — split into branded and unbranded queries, and in which topics competitors are ahead of you. On the left — what the report is based on: engines, topics × prompts × samples, pages crawled, competitors, mentions and sources. This is the slide you can show an executive instead of the whole deck: where the brand stands right now, on one screen, with pointers to the detail slides.

Slides 8–15: how your site looks to AI

The site-audit block answers the question “can AI even read you”: domain and robots, technical checks, structure and broken links, social signals — and homepage speed on its own slide.

Slide 14 — the homepage’s Lighthouse lab speed metrics.

Every metric — LCP, CLS, TBT, FCP, Speed Index, TTI — is plotted against PageSpeed thresholds: the green, yellow and red zones are visible at a glance, no interpretation needed. In the example above the site has healthy LCP and CLS, but Total Blocking Time sits in the red zone — heavy JavaScript is blocking the main thread. That’s a concrete task for a developer, not an abstract “speed score”.

Slide 21: where exactly you get found — and where you don’t

Next the report moves to AI visibility. The audit’s prompts are tagged by funnel stage, and visibility is computed per stage.

Slide 21 — visibility by funnel stage and the branded / unbranded split.

Here is the most common story made visible: in queries that already name the brand there is visibility (50% in the example), while in unbranded queries — where buyers choose among unknowns — it is just 6%, even though those queries make up 92% of the corpus. This is the gap the recommendations go on to work on.

Slide 26: whom AI cites when answering about your category

Slide 26 — the mix of cited sources and the top domains.

AI answers lean on sources, and the report shows them in full: how source types are distributed (corporate, owned, social, PR, reviews) and which domains get cited most. Your site is highlighted in the list — you see instantly where it ranks among dozens of source domains and who holds the positions above you.

Slides 28–29: competitors — where they lead, and why

The competitor block starts with an honest comparison by topic.

Slide 28 — share of AI answers with a mention: you against every tracked competitor, per topic.

For every topic you see what share of answers each brand takes and the gap in percentage points. Branded prompts are excluded from the comparison — it measures precisely the fight for unbranded answers.

Slide 29 — domains whose pages verifiably mention each brand.

And this is the answer to “why they’re ahead”: the domains AI draws its evidence about each competitor from. Every domain in the list is verified — its pages really do mention the brand; it’s not a model guess. In practice it’s a ready-made list of the venues where you should be present too.

Slide 32: what to do — by priority

Slide 32 — prioritized recommendations across three directions.

The report ends not with conclusions but with a plan: recommendations across three directions — context, technical, content — with priorities. Each one is tied to a concrete gap from the audit: missing sameAs links — there’s a task; AI recommends competitors in a topic where you’re absent — there’s a task; heavy JavaScript — there’s a task. The full lists and a ready brief for every recommendation live in the product.

You don’t even have to read the report yourself — we’re AI-friendly

Suparanku is AI-friendly in both directions: we don’t just measure how AI represents your brand — we’re open to AI agents ourselves. Through the Suparanku MCP server your assistant — ChatGPT, Claude, Gemini or any other MCP client — connects directly and pulls everything the audit contains on its own: site-audit data, AI-visibility metrics, and even the finished PDF report itself — an agent can request generation and receive a download link. It works with recommendations as tasks: fetches the list, takes items into work and marks them published — the loop turns without manual routine. MCP is available on every paid plan, including Starter.

How to get your report

1. Sign up — no card needed.
2. Complete onboarding: a few minutes to confirm the brand, topics and competitors.
3. The system runs the scan and the site audit, and once everything is ready it assembles the report itself and emails you the link.

The report is generated in your interface language — Japanese, English, Russian, Italian, Korean, Spanish, German, French and Portuguese are supported. The size depends on your brand’s data: the more prompts and data-backed sections, the more slides.

Honest about expectations

The audit doesn’t promise growth — it shows where you stand, why, and what to do next. AI Representation Management is a closed loop: context → measurement → recommendations → publishing → re-measurement. The free audit is the first turn of that loop; keeping the gap closed while models, competitors and content keep changing is what the subscription is for.