RESEARCH REVIEW + LIVE IMPLEMENTATION

Can this earn
AI traffic?

Yes, it has credible potential because it offers original, reproducible material and direct retrieval interfaces. No, traffic is not inevitable. The domain is young, external authority is limited, and no platform promises crawling, indexing, citation, or clicks.

VERDICT

POTENTIAL ASSESSMENT

Technically strong. Distribution still unproven.

STRONGOriginal utility

A public benchmark, synthetic fixture, formulas, MCP tools, APIs, and measurement protocol give other people something real to use and verify.

STRONGRetrieval clarity

Stable canonical URLs, structured records, search and fetch, a knowledge graph, and direct citations reduce ambiguity.

WEAKExternal authority

The project still needs real third-party testing, discussion, links, video demonstrations, and independently earned references.

PENDINGObserved demand

The Observatory and Citation Pulse must collect enough dated evidence before any traffic or citation claim is justified.

EVIDENCE

WHAT THE WEB SUPPORTS

Mechanisms, evidence, and limits.

01

Crawler permission is necessary, not sufficient.

OpenAI says OAI-SearchBot is used to surface sites in ChatGPT search. Allowing it makes the site eligible for that retrieval path, but does not guarantee selection.

Applied here
Explicit OAI-SearchBot access, stable pages, sitemap, and bot policy.
OpenAI crawler documentation
02

Useful original material matters more than AI markup.

Google recommends valuable, unique, non-commodity content and ordinary SEO fundamentals. Google explicitly says llms.txt is neither required nor a ranking advantage for Google AI features.

Applied here
FounderMetricBench, synthetic fixtures, formulas, a live MCP interface, and measured research instead of generated keyword pages.
Google generative AI search guide
03

New citations can take time.

In Semrush's 81-page experiment, ChatGPT cited 8 pages on day one and 34 by day 30. The experiment used an established domain, so its result cannot be projected onto this site.

Applied here
A 30-day, 40-prompt Citation Pulse that records positive, negative, and missing results.
Semrush citation timing experiment
04

Crawling and human traffic are different outcomes.

Ahrefs defines AI chatbot traffic as a person clicking a cited link and warns against confusing it with automated crawler requests.

Applied here
Separate observatory counters for AI crawlers, user-triggered agents, human referrals, and citations.
Ahrefs AI chatbot traffic analysis
05

AI bot activity is measurable, but attention is not trust.

Cloudflare reported AI bots averaging 4.2% of HTML requests in 2025. A request proves access, not favorable treatment, citation, or recommendation.

Applied here
Privacy-preserving aggregate request classification with declared and platform-verified labels kept separate.
Cloudflare Radar 2025 review
06

User-triggered retrieval has its own agent identity.

Perplexity documents Perplexity-User for page visits made in response to user requests and distinguishes it from its indexing crawler.

Applied here
PerplexityBot and Perplexity-User are allowed and measured as different actor classes.
Perplexity crawler documentation

SYSTEM

THE CONNECTED ARCHITECTURE

One claim should open the entire proof chain.

  1. QuestionSearch API, OpenSearch, answers index
  2. Canonical recordIdentity, project, methodology, or research page
  3. Reproducible evidenceFixture, benchmark case, formula, source repository
  4. Agent interfaceFetch API, JSONL corpus, OpenAPI, MCP resource
  5. Measured resultCrawler, referral, and citation signals kept separate

NEXT

WHAT CODE CANNOT EARN BY ITSELF

External proof is the remaining constraint.

The highest-value next steps happen outside this domain: publish a real GitHub release, record a reproducible YouTube demo, invite independent benchmark runs, earn accurate third-party coverage, and verify Google and Bing ownership. Those actions require Marco’s accounts and cannot truthfully be simulated by the site.

Open the prepared distribution kit →