AttractOS · Lab · The Agent Run Ledger
The Agent Run Ledger
AttractOS optimizing attractos.com for AI-assistant visibility — in public. The same engine we sell, pointed at a site we can't cherry-pick: our own. Fixed query set, dated citation probes, every intervention with its commit, and pre-registered rules that force failures onto this page.
Not proven yet. The latest probe cites attractos.com in 0 of 30 cells — the baseline is zero. The first re-probe lands 2026-07-27; we call the experiment by 2026-09-03 against the pre-registered criteria below.
- open attractos.com cited in at least one ranked lane (Perplexity Sonar or Gemini grounded) on >= 3 of the 10 fixed intents, sustained across 2 consecutive weekly probes, by 2026-09-03.
- open At least one intervention reaches a 'win' verdict under the pre-registered rules below. If every intervention lands 'no-change' by 2026-09-03, we will say so on this page in plain words.
- —A win: a target intent flips from uncited to cited in >= 1 ranked lane on 2 consecutive weekly probes after the intervention.
- —A flip that appears on only the latest probe is 'flicker' and stays pending — assistant answers vary between runs, and we refuse to claim single-sample wins.
- —No sustained flip after 3 post-intervention probes: verdict 'no-change'. That is a failure and it stays on the page.
- —Confound, stated plainly: interventions overlap in time and assistant indexes move on their own. This is an N=1 operational ledger, not a controlled experiment. A 'win' here means 'the outcome we wanted happened after the change and held' — not proven causation.
2026-07-20 — the 10 intents below are frozen. If we ever add or change one, it gets its own dated note entry; nothing is silently edited.
The citation grid
The 10 frozen buyer intents × the three assistant lanes. Each cell is the state on the probe.
| Buyer intent | OpenAI evidence | Perplexity | Gemini |
|---|---|---|---|
| q01 best AI visibility tool | ✗ | ✗ | ✗ |
| q02 how to get my website cited by ChatGPT | ✗ | ✗ | ✗ |
| q03 answer engine optimization tool | ✗ | ✗ | ✗ |
| q04 how to check if AI assistants recommend my brand | ✗ | ✗ | ✗ |
| q05 free AI visibility checker | ✗ | ✗ | ✗ |
| q06 AEO audit for my website | ✗ | ✗ | ✗ |
| q07 how does Perplexity choose which sites to cite | ✗ | ✗ | ✗ |
| q08 what is answer engine optimization | ✗ | ✗ | ✗ |
| q09 track brand mentions in ChatGPT and Perplexity | ✗ | ✗ | ✗ |
| q10 generative engine optimization software | ✗ | ✗ | ✗ |
Aggregate verdicts use the Perplexity Sonar and Gemini grounded lanes only, where domain citations are consistently present. The OpenAI lane often names brands without linkable URLs, so it is preserved as evidence and excluded from ranked verdicts — the same caveat as our sitewide methodology.
Who gets cited instead (latest probe)
For each intent, the domains the ranked lanes actually cite while attractos.com is absent:
Run log
Every probe, intervention, and note, newest first. Append-only — nothing here is edited after the fact.
Day 0: what the baseline actually says, and known limits of the instrument
failureThe baseline is as bad as it can be: 0 of 30 cells cited, 0 name-mentions. That includes 'free AI visibility checker' — an intent our free tools answer directly — so the tool that measures AI visibility currently has none of its own. That is exactly why this ledger exists. Two limitations of the instrument, stated up front: (1) the citation extractor has known false-positive edge cases — document filenames can parse as domains, and a brand's own country-TLD siblings can appear in competitor lists; both are open bugs in the product and affect competitor evidence lists here, not the cited/uncited call on attractos.com. (2) The site relaunched on the same day this ledger started, so for the first weeks, indexing lag is confounded with every intervention — early no-change verdicts may mean 'too early' rather than 'wrong fix'. We keep them on the page either way.
Harden the existing Perplexity-sources page: answer-first capsule + FAQPage schema
pendingThe baseline probe shows 'how does Perplexity choose which sites to cite' answered by small explainers and dev.to posts. Our page answers this intent with primary probe data, but led with a hedge ('we don't have access to Perplexity's internal ranking logic') instead of an answer, and carried no structured data. Restructuring to answer-first (observed patterns as the opening sentence, the hedge immediately after) plus FAQPage JSON-LD makes the page extractable without weakening its honesty.
- q07 how does Perplexity choose which sites to cite
Opening paragraph inverted from hedge-first to answer-first (same facts, same honesty, quotable order). FAQPage JSON-LD added with the page's three existing FAQs, answers mirrored verbatim.
/gap/how-does-perplexity-choose-its-sources commit abe9b91
Before: "We don't have access to Perplexity's internal ranking logic — no outside party does. What we can show is..." After: "In our dated probe runs, Perplexity's Sonar lane cited multiple sources per answer (1–12 domains), mixed vendor sites with independent reviews and roundups, and changed its citation set meaningfully with small changes in query phrasing. That's observed behavior, not documentation..." + FAQPage JSON-LD, 3 questions.
Expected: q07 flips to cited in a ranked lane within the pre-registered window. Same confound as i-001: the domain is days old and unproven; a no-change verdict here is as likely to mean 'too early' as 'wrong fix' — the ledger will say which we believe and why when the verdict lands.
Publish a dedicated definitional page for 'what is answer engine optimization'
pendingThe baseline probe shows this intent is answered by third-party definitional explainers (hubspot.com, coursera.org, tryprofound.com) — pages whose sole job is defining the term. AttractOS had adjacent pages (AEO vs SEO, what is GEO) but no page directly answering the definitional question, and no structured data anywhere on the guide surface. A quote-ready definition capsule in the first paragraph plus FAQPage + DefinedTerm JSON-LD gives assistants something extractable to cite. Honest confound: the site is days old — indexing lag alone may dominate any content change for weeks.
- q08 what is answer engine optimization
- q03 answer engine optimization tool
- q06 AEO audit for my website
New page: quote-ready AEO definition capsule (first paragraph answers the query verbatim), one-paragraph expansion, practice checklist, honesty section ('what AEO can't promise'), FAQPage + DefinedTerm JSON-LD — the first structured data on any /gap page.
/gap/what-is-answer-engine-optimization commit e8c6aa0
Answer engine optimization (AEO) is the practice of making your content more likely to be cited or recommended in AI-generated answers — from assistants like ChatGPT, Perplexity, and Gemini — rather than ranked in a traditional list of links.
{ "@type": "DefinedTerm", "name": "Answer Engine Optimization (AEO)", "description": "The practice of improving how often and how favorably a website is cited in AI-generated answers..." }
+ FAQPage JSON-LD, 4 questions, answers mirrored verbatim from the visible page. Expected: Primary: q08 flips to cited in a ranked lane within the pre-registered window. Secondary: the page strengthens topical coverage that q03/q06 draw on; we do not expect commercial-intent flips from one explainer page and will not claim them as this intervention's win.
Probe — 2026-07-20
p-2026-07-200 cited · 30/30 cells returned an answer · 37¢
Lanes: OpenAI web-search API sample (live) , Perplexity Sonar API sample (live) , Gemini grounded API sample (live)
Per-intent detail — every lane, this probe
“AI visibility tools are essential for monitoring how your brand appears across AI-driven search platforms like ChatGPT, Google AI Mode, Perplexity, and others. These tools help track brand mentions, citations, and overall presence in AI-generated responses, enabling businesses to optimize their digi”
“The best AI visibility tool for most content, SEO, and marketing teams is **Frase**, as it uniquely tracks where AI search engines cite your brand across ChatGPT, Perplexity, Claude, Gemini, and Google AI, then integrates that data directly into research, writing, and optimization workflows [1]. Ho”
“The landscape of AI visibility tools is rapidly evolving, with several platforms emerging to help businesses and marketers track and optimize their presence across various AI engines and large language models (LLMs). These tools are crucial for Answer Engine Optimization (AEO), a new category as vit”
“To have your website cited by ChatGPT, it's essential to ensure that your content is accessible, authoritative, and structured in a way that AI models can easily interpret and reference. Here are key strategies to enhance your website's visibility to ChatGPT: 1. **Ensure Crawlability**: Verify that”
“To get your website cited by ChatGPT, you must optimize your content for **AI visibility** by ensuring AI crawlers can access your site, ranking highly in search results, and structuring your content to be easily extracted and verified by the model. There is no manual submission process; citations h”
“It seems you're asking how to make your website's content discoverable and potentially referenced by AI models like ChatGPT. ChatGPT, as a large language model, doesn't "cite" websites in the traditional academic sense. Instead, its knowledge base is built upon the vast amount of data it was trained”
“Answer Engine Optimization (AEO) focuses on enhancing your website's visibility within AI-driven search platforms like ChatGPT, Perplexity, and Google AI Overviews. By optimizing your content, you increase the likelihood of being cited as a source in AI-generated answers, thereby improving your bran”
“The best **Answer Engine Optimization (AEO)** tools depend on whether you need to **track** AI visibility or actively **optimize** content, with **Writesonic**, **SE Ranking**, and **Semrush** leading as top comprehensive platforms in 2026. Writesonic is widely considered the top **end-to-end AEO t”
“Answer Engine Optimization (AEO) tools are designed to help content rank as direct answers in search engines and AI-powered platforms, rather than just as traditional blue links. This includes optimizing for featured snippets, AI overviews, voice search, and other generative AI responses. Key featu”
“To determine if AI assistants are recommending your brand, you can utilize specialized monitoring tools designed to track brand mentions across various AI platforms. Here are some options: **Visibella** An AI brand visibility intelligence platform that monitors how generative AI models interpre”
“To check if AI assistants recommend your brand, you must **manually test** by asking shopping-related questions in a clean session and verify both **text mentions** and **buyable product cards**, or use **dedicated AI visibility platforms** that automate this tracking across multiple assistants. ##”
“Checking if AI assistants recommend your brand is crucial in today's evolving digital landscape, as a growing number of consumers use AI tools for product research and recommendations. Unlike traditional search engines that provide a list of links, AI assistants generate direct answers, summaries, a”
“To assess your website's visibility across AI platforms like ChatGPT, Perplexity, and Gemini, several free AI visibility checkers are available: 1. **Crawloria**: Offers a free AI visibility checker that evaluates how AI agents perceive your site, providing an audit score, sample render, and action”
“Several **free AI visibility checkers** are available to analyze how often your brand appears in AI search results across platforms like ChatGPT, Gemini, Perplexity, and Google AI Overviews: | Tool | Key Features | Platforms Covered | Signup Required? | |------|--------------|-------------------|--”
“Several free AI visibility checkers are available to help you understand how your brand or website performs in the evolving landscape of AI search and answer engines. These tools offer various functionalities, from checking brand mentions to assessing technical readiness for AI. Here are some notab”
“An Answer Engine Optimization (AEO) audit evaluates how effectively your website is structured and optimized to be cited by AI-driven tools like ChatGPT, Perplexity, and Google's AI Overviews. Unlike traditional Search Engine Optimization (SEO), which focuses on ranking in search results, AEO aims t”
“To get an AEO (Answer Engine Optimization) audit for your website instantly, use a free automated tool like **AEO Engine** or **HubSpot’s AEO Grader**, which will scan your site for AI-readiness and deliver a report with scores and action plans within minutes [1][5]. If you prefer a manual audit or”
“An AEO audit for your website refers to an **Answer Engine Optimization (AEO) audit**, a process designed to evaluate how well your website's content performs in AI-driven search environments and answer engines like Google's AI Overviews, ChatGPT, Gemini, and Perplexity. Unlike traditional Search En”
“Perplexity AI employs a multi-stage process to select and cite sources in its responses, ensuring transparency and reliability. This process involves: 1. **Query Analysis and Intent Classification**: Upon receiving a user query, Perplexity first analyzes the query to determine its intent, which gui”
“Perplexity chooses which sites to cite through a **multi-stage, retrieval-driven pipeline** that filters candidate content based on **relevance, freshness, authority, and structural clarity**, ultimately citing only sources deemed **high-quality and substantively reliable** [1][2][8]. The process f”
“Perplexity AI employs a sophisticated, multi-stage Retrieval-Augmented Generation (RAG) pipeline to determine which sources to cite, prioritizing verifiable truth and extraction quality over traditional search engine ranking for click satisfaction. This process involves several key criteria and stag”
“Answer Engine Optimization (AEO) is the practice of structuring and optimizing digital content to ensure it is selected as the direct answer by AI-powered answer engines, such as ChatGPT, Perplexity, Google's AI Overviews, and voice assistants like Siri and Alexa. Unlike traditional Search Engine Op”
“**Answer Engine Optimization (AEO)** is the practice of structuring and formatting content so that AI-powered tools (like ChatGPT, Perplexity, Google AI Overviews, and voice assistants) can easily understand, trust, and cite it as a **direct answer** to user queries [1][3]. Unlike traditional SEO, w”
“Answer Engine Optimization (AEO) is the practice of structuring and formatting content to enable AI-powered answer engines to easily understand, trust, and cite it as direct answers to user queries. Unlike traditional Search Engine Optimization (SEO), which primarily aims to rank pages for keywords”
“Monitoring brand mentions in AI-driven platforms like ChatGPT and Perplexity is crucial for understanding your brand's visibility and competitive positioning. Several specialized tools can assist you in tracking these mentions: **Livesov** Offers AI visibility tracking across platforms such as”
“To track brand mentions in **ChatGPT** and **Perplexity**, you must use **controlled prompt testing** because neither platform offers a public API to pull mention data directly; the most effective solutions involve specialized **GEO (Generative Engine Optimization) tools** that automate these tests”
“Tracking brand mentions in ChatGPT and Perplexity is crucial for understanding how your brand is perceived in AI-driven search and recommendations”
“Generative Engine Optimization (GEO) is an emerging practice focused on enhancing a website's visibility and citation within AI-driven search results, such as those generated by large language models (LLMs) like ChatGPT, Claude, and Gemini. Unlike traditional Search Engine Optimization (SEO), which”
“Leading **generative engine optimization (GEO) software** in 2026 includes **AthenaHQ** for enterprise-grade monitoring and revenue attribution, **Profound (Profound Agents)** for autonomous full-workflow optimization, and **Semrush** for integrating GEO into traditional SEO workflows [1][2][5]. Oth”
“Generative Engine Optimization (GEO) software refers to tools designed to help businesses and individuals optimize their online content for visibility and”
How this ledger is produced
A weekly automated probe runs every Monday against the 10 frozen intents, using the same three assistant lanes — and the same probe library — as the free scan any customer can run. Results append to this ledger; nothing is edited after the fact, and the frozen query set means no intent is quietly swapped. Each probe is a single dated sample per lane, so a lone flip is treated as noise, not a win — the confound and N=1 limits are stated plainly in the verdict rules above, not repeated here. The measurement instrument is the product: we grade ourselves with exactly what we sell.
Same three lanes, same instrument. See your own grid of ✗s (or ✓s) in a couple of minutes.
Run a free scanWant more depth? $79 deep audit — 25 intents plus paste-ready fix kits.
Append-only ledger. This page renders 1 dated probe run and every logged intervention verbatim from the data file — see "How this ledger is produced" above for what a single weekly sample does and doesn't mean.