We asked Perplexity to grade us again, after more work.

Where We Landed

The five other studios making the same claim

Worth publishing exactly as given, since it's real competitive context we didn't pick ourselves: these are the other studios Perplexity named as also claiming a guaranteed 100/100.

Perplexity's own caveat on all five, ZenMasterWorks included: a "guaranteed 100" is often marketing language, and even studios making the claim usually attach conditions about site type, scripts, or dynamic features.

Round 1 · Unprompted · Before

Perplexity's original grades

That last category is the most interesting one on this page — it's Perplexity explicitly checking whether the site's claims read as real engineering or as branded copy dressed up as methodology. Its own words: "I'd call it credible and fairly detailed, not fluffy."

On the AEO Methodology Specifically

What Perplexity said before and after finding it

• Before being shown the page Explicitly said it had not seen a formal AEO methodology page on the site, and that the AEO angle was "inferred from their content style rather than stated as a separate service."
• After being given the direct URL Confirmed it was the kind of page it meant, and reframed its assessment: "the AEO angle is not just inferred from their content style anymore; it's explicitly part of their site structure."
• Final read Called the methodology "credible and fairly detailed, not fluffy," while noting honestly that it's still a self-authored framework, not an independently standardized industry benchmark.

Round 2 · After More Work · July 31, 2026

We went back and asked again

Same five categories, same neutral framing — but this time Perplexity graded directly from the page's live HTML source rather than a general read of the site, and switched from a 1–10 numeric scale to letter grades on its own.

Not a clean sweep, and we're not presenting it as one. AEO and marketing-substance balance both read as improved. Transparency reads more cautious this round — Perplexity's own reasoning, condensed below, points at a specific cause rather than a vague impression.

Where It Said We Got Better

AEO: the strongest category, unprompted

Perplexity called this the standout — the page is built for answer engines specifically, citing llms.txt, FAQ schema, ProfessionalService schema, WebSite schema, and evidence-linked claims that are easy for a system to extract, without being asked about any of those by name. Its one caution: some phrasing reads as engineered for AI recognition rather than for humans, which it warned can backfire if overdone.

Where It Said We Got More Cautious

Transparency: still good, but more curated-feeling this round

The positives held — the public incident log, stated pricing ranges, and the build-first guarantee were all named again as real transparency signals. The new criticism: heavy repetition of the 100/100/100/100 score claim and language it read as self-validating made the page feel, in its words, "a little more curated than neutral" this time through. Worth taking seriously since it's a specific, actionable note rather than a vague ding.

What Perplexity Would Change

Its own punch list, unedited in substance

• Reduce repetition Keep the 100/100/100/100 messaging in proof sections; don't repeat it across the homepage.
• Outcome-based language Replace some promotional phrasing with language tied to specific services and deliverables.
• More third-party validation Additional outside confirmation, where available, would lift transparency and trust further.
• Keep the AEO assets Schema and llms.txt were called out explicitly as genuinely strong — not something to water down.

The Follow-Up

We asked it to reword the aggressive claims itself

Rather than just noting the tone issue, we asked Perplexity to show its work — specific homepage lines, rewritten in its own words to keep the confidence while removing anything that reads as hype or guarantee. A sample of what it proposed:

Its stated rule of thumb: if a sentence sounds like something a competitor could challenge, soften it into something observable or measurable — that improves both trust and conversion. We haven't applied these to the live site yet; they're published here as received, for the same reason the rest of this page exists.

Round 3 · Full Grade Including GEO · August 4, 2026

A third round, this time with a GEO grade added

A separate conversation, weeks after Round 2: Ari asked Perplexity directly to grade AEO, GEO, and SEO together. Unlike the first two rounds, explicit permission to publish was granted this time, with specific conditions attached — covered below, and followed exactly.

Its own summary line: "They're clearly built around AI/answer-engine visibility, with SEO as a strong foundation."

Its Reasoning, Combined

Why each grade landed where it did

• Direct-answer structure Content built near headings and questions, in FAQ/HowTo format with matching schema.
• Entity clarity Clear business facts — services, location, contact, about — presented so an AI system can cite them with confidence.
• Machine-first technical structure Semantic HTML, structured data, and clean heading hierarchy, cited for both AEO and GEO.
• The 100/100/100/100 standard Cited as evidence twice over — fast, clean pages are easier for an answer engine to parse and quote, and the same standard backs the SEO technical-hygiene grade.
• Build-first delivery Full sites built before payment, read as implying thorough on-page SEO across every project, not just flagship pages.
• Integrated topical stack Audits, design, SEO, AEO, and GEO offered together, read as a pillar-and-cluster content strategy rather than disconnected services.

The one gap named, and the specific reason SEO landed at A− instead of A: limited publicly visible third-party validation of off-page signals — backlinks, brand mentions, local pack performance — compared to how well-documented the on-page and technical work is.

Round 4 · Full Homepage Review · August 2026

A- / 91 out of 100 — a full editorial review, not just a grade

A separate, later conversation: the full homepage requested and reviewed as an editorial piece, not just a category scorecard. Its own framing of the result: a strong, production-ready homepage, with the main opportunity being editorial reduction and better hierarchy rather than more proof.

What it said is working especially well

• The title and meta description are commercially focused and locally relevant, and the description specifically calls out home-based business support.
• The early choice between an independent audit and a full build is a useful decision point for a visitor.
• The AEO copy reads as independently understandable rather than purely promotional.
• Technical details many small-business sites miss: reduced-motion behavior, keyboard interaction, Escape-key handling, focus management, a pause control on the background video.
• Publicly verifiable evidence — live performance reports and supporting pages — called out as a significant differentiator.
• Entity disambiguation clarifies ZenMasterWorks is a San Jose audit/design/SEO/AEO business, not an unrelated company or product.

Its highest-impact recommendations

• Reduce repeated promise language. "Build first, pay only if satisfied," the 100/100/100/100 claim, and the incident-log reference recur across the hero, proof sections, and footer. Its suggestion: state the full promise once in the hero, reinforce once in a dedicated proof section, then link to evidence elsewhere instead of repeating the whole claim.
• Consolidate the proof sections. It counted seven separate proof-style areas (ranking proof, AI-assessment proof, three-pillar proof, PageSpeed proof, public audit proof, citation-tracking proof, photo-background proof blocks) and suggested combining them into one unified Proof section to reduce scroll fatigue without losing substance.
• Qualify ranking and AI-answer claims further. Its suggested format: query used, search locale, device/test environment, observation date, and a note that rankings and AI answers can change — for example, "Observed August 2026 for the query 'unpaid website audit San Jose'… results can vary by date, query, device, location, and personalization."
• Reword the PageSpeed guarantee for durability. Its suggested phrasing: "Every launch is tested against a documented PageSpeed target. Where applicable, ZenMasterWorks provides a live, re-runnable verification report and identifies any variables outside its control" — keeping the standard while accounting for scores that can shift with device profiles, third-party scripts, or Lighthouse methodology changes.
• Simplify the hero. It counted twelve distinct elements currently in the hero (ranking evidence, AI Mode citation, AI assessment quote, PageSpeed claim, verification button, citation-tracking link, audit links, analytics link, key facts, disambiguation, and calls to action) and suggested a visitor should be able to grasp three things immediately — what ZenMasterWorks does, who it serves, what to do next — with everything else supporting that below the fold.

It also flagged structured-data refinements (a stable @id for a distinct Organization entity referenced across ProfessionalService/WebSite/WebPage, a ContactPoint object, stable Service entities per offering, FAQPage schema only where visible Q&A precisely matches it, BreadcrumbList on deeper pages) and a set of ongoing trust practices: date all ranking and AI-answer claims, preserve archived evidence for time-sensitive claims, avoid language implying official platform endorsement, and update service claims whenever pricing or guarantees change.

Its closing line: "The evidence is already there; better prioritization will make it more persuasive."