Free Playbook
The AI Visibility Playbook
A complete, actionable checklist for getting ChatGPT, Claude, Perplexity, and Gemini to actually mention your brand - from technical crawlability to ongoing monitoring.
Your progress
0 of 23 completed
Saved automatically in this browser. Ran a scan already? Re-check your technical items.
Technical foundation
AI crawlers and answer engines have to be able to reach, fetch, and parse your site before anything else on this list matters. This is exactly what AskRank's free Website Scan checks - start there to see where you stand.
Why it matters. ChatGPT (via GPTBot), Claude (via ClaudeBot), Perplexity (via PerplexityBot), and Google's AI Overviews (via Google-Extended) each send a named crawler. If robots.txt blocks or omits them, your content is invisible to that engine no matter how good it is.
How to do it
- Open yourdomain.com/robots.txt in a browser.
- Make sure there is no "Disallow: /" rule under User-agent: * or under GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Applebot-Extended specifically.
- If your CMS or CDN (Cloudflare, Vercel) has a bot-blocking or "AI scrapers" toggle, explicitly allow these four crawlers even if you block generic scrapers.
- Re-check after any CDN/WAF change - security tools often add new bot rules silently during upgrades.
How to verify. Fetch /robots.txt with curl -A "GPTBot" yourdomain.com/robots.txt and confirm no Disallow rule matches your page paths.
User-agent: GPTBot
Allow: /
User-agent: ClaudeBot
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: Google-Extended
Allow: /Why it matters. llms.txt is an emerging convention (proposed 2024, gaining adoption through 2026) that gives AI systems a clean, curated map of your most important pages in plain Markdown - instead of them guessing from a messy sitemap or nav menu.
How to do it
- Create /llms.txt at your domain root with an H1 (your product name), a one-line summary, and a bulleted list of links to your most important pages (product, pricing, docs, key blog posts).
- Keep it short and curated - 10-30 links, not a dump of your entire sitemap.
- Optionally publish /llms-full.txt with the full text content of those pages inlined, for engines that fetch it directly instead of crawling each URL.
- Link to it from your footer or docs so it is discoverable, and list it in robots.txt as a Sitemap-style hint if your platform supports that.
How to verify. Load yourdomain.com/llms.txt directly in a browser and confirm it renders as valid Markdown with working links.
# Acme
> Acme helps SaaS teams track AI visibility across ChatGPT, Claude, and Perplexity.
## Docs
- [Getting started](/docs/getting-started)
- [API reference](/docs/api)
## Product
- [Pricing](/pricing)
- [Features](/product)Why it matters. AI crawlers use your sitemap the same way search engines do: as the primary discovery list. A stale sitemap (missing new pages, listing 404s) or duplicate URLs under different canonicals fragments the "credit" your content gets in an engine's index.
How to do it
- Confirm /sitemap.xml exists, returns 200, and is referenced in robots.txt.
- Make sure it auto-regenerates on publish (most frameworks/CMSs do this out of the box) rather than being a stale, manually-maintained file.
- Add a self-referencing <link rel="canonical"> on every page, especially ones reachable via multiple URLs (with/without trailing slash, tracking params, http/https).
- Remove or 410 old pages instead of leaving thin, outdated duplicates live.
How to verify. Open sitemap.xml, spot-check 5 URLs for a 200 status, and confirm the canonical tag on each matches the URL you would want indexed.
<link rel="canonical" href="https://yourdomain.com/pricing" />Why it matters. Organization, Product/SoftwareApplication, and FAQPage schema give AI engines an unambiguous, machine-readable summary of who you are, what you sell, and common questions answered on the page - reducing the chance of a hallucinated or outdated description.
How to do it
- Add an Organization schema (name, logo, url, sameAs links to your social/Crunchbase/Wikipedia profiles) site-wide, typically in the layout <head>.
- Add Product or SoftwareApplication schema on your pricing/product pages with name, description, and offers/pricing.
- Add FAQPage schema wherever you have a real FAQ section (see the Content section below) - it directly mirrors what LLMs quote.
- Validate every schema block - broken JSON-LD is worse than none, since it can get the whole page skipped by strict parsers.
How to verify. Paste the page URL into Google's Rich Results Test or schema.org's validator and confirm zero errors.
{"@context":"https://schema.org","@type":"Organization","name":"Acme","url":"https://acme.com","logo":"https://acme.com/logo.png","sameAs":["https://www.linkedin.com/company/acme","https://www.crunchbase.com/organization/acme"]}Why it matters. Most AI crawlers fetch raw HTML with a short timeout and do not execute heavy client-side JavaScript. Slow servers, render-blocking scripts, and content that only appears after hydration are effectively invisible to them.
How to do it
- Server-render (or statically generate) the primary content of every page that should be quotable - do not rely on client-side JS to paint the text.
- Use real semantic tags (h1-h3, p, ul/ol, table) instead of div soup - this is what lets an LLM correctly parse structure like "steps" or "comparison rows".
- Keep Time to First Byte low (under ~600ms) and avoid render-blocking third-party scripts above the fold.
- One clear H1 per page, a logical H2/H3 outline below it - mirrors how AI engines chunk a page for retrieval.
How to verify. Run "curl -s yourdomain.com/page | grep -c '<h1'" - if it returns 0, your H1 is client-rendered and crawlers may miss it. Confirm with view-source (not DevTools Elements) that the real text is present.
view-source:https://yourdomain.com/pricing - the plan names and prices should be visible in the raw HTML, not only after JS runs.Content built for AI answers
AI answer engines favor content shaped like an answer: a direct definition, a scoped comparison, a numbered list, a cited fact. Restructure your best existing pages around these patterns before writing anything new.
Why it matters. FAQ blocks are the single most-cited content shape in ChatGPT and Perplexity answers - the question/answer format maps almost 1:1 onto how a user phrases their own prompt, making it trivial for the model to lift and cite.
How to do it
- Add 5-8 genuine questions to your product, pricing, and top comparison pages - pull them from real support tickets, sales calls, and Reddit/forum threads about your category.
- Answer each in 2-4 direct sentences: state the answer first, then one sentence of context. Avoid marketing fluff in the answer itself.
- Pair the visible FAQ with FAQPage JSON-LD (see Technical foundation) so it is both human- and machine-readable.
- Update FAQs quarterly as pricing, features, or competitor names change - stale FAQ answers get cited and hurt trust.
How to verify. Ask ChatGPT or Perplexity one of your FAQ questions phrased naturally and check whether your answer (or a close paraphrase) appears, ideally with a citation.
Q: "Does [Product] support single sign-on?" A: "Yes - SAML and OAuth SSO are available on the Team and Enterprise plans, configurable from Settings > Security."Why it matters. When someone asks an AI assistant "what is [your category]" or "what is [your product]", the model needs a clean, quotable definition. Pages that bury this in narrative prose get paraphrased loosely or skipped for a competitor's clearer one.
How to do it
- Open your product/category page with a single bolded sentence that defines the term the way a glossary would: "[Term] is a [category] that [core function], used by [audience] to [outcome]."
- Keep the definition under 40 words and free of adjectives that only make sense to someone who already knows your product.
- Repeat the same core definition consistently across your homepage, about page, and category pages - inconsistent self-descriptions confuse extraction and invite the model to average across competitor definitions instead.
How to verify. Ask an LLM "what is [your product]" with no other context and compare its answer to your own definition sentence - they should be nearly identical.
"AskRank is an AEO/GEO visibility tracker that shows indie SaaS founders how often ChatGPT, Claude, Perplexity, and Gemini mention their brand."Why it matters. "Best tools for X" and "[A] vs [B]" are some of the highest-volume prompt patterns across ChatGPT and Perplexity. If you never publish this content yourself, AI engines fill the gap entirely from third-party review sites and Reddit threads you do not control.
How to do it
- Write one honest comparison page per real alternative prospects consider (not just favorable ones) - name real strengths and limitations on both sides.
- Write at least one "best [category] tools for [use case]" roundup that includes yourself alongside genuine competitors - a listicle that only mentions your own product reads as untrustworthy and gets down-weighted.
- Use real comparison tables (rows/columns, not prose) - table markup is one of the easiest structures for an LLM to extract accurately.
- Refresh pricing and feature rows at least quarterly - AI engines that cache or embed stale comparison data will misquote you otherwise.
How to verify. Ask an AI assistant "[your category] vs [specific competitor]" or "best [category] tools" and see whether your comparison page or roundup gets cited as a source.
/compare/yourproduct-vs-competitor and /best-[category]-tools, each with a feature/pricing table and an honest "best for" line per entrant.Why it matters. LLMs strongly prefer to quote concrete, falsifiable claims ("tracks 50-500 prompts across 4 LLMs", "$19/mo Starter plan") over vague marketing language ("industry-leading", "powerful"), because concrete numbers are easier to extract and lower-risk to repeat.
How to do it
- Replace vague superlatives on your homepage and product pages with specific numbers: pricing, plan limits, response times, supported integrations, dataset sizes.
- Attribute non-obvious stats to a source (your own data, a named study) so the model can cite it with confidence instead of treating it as an unverifiable claim.
- Keep a single source of truth for frequently-cited numbers (e.g. pricing) and update it everywhere at once - contradictory numbers across pages actively suppress citation.
How to verify. Grep your own site for "leading", "best-in-class", "powerful" and replace at least the homepage instances with a real number or fact.
"Runs 50-500 of your prompts across ChatGPT, Claude, Perplexity, and Gemini on a schedule" instead of "the most powerful AI visibility platform".Why it matters. AI engines weigh recency and demonstrated expertise similarly to how Google's E-E-A-T (Experience, Expertise, Authoritativeness, Trust) guidelines work - an unattributed, undated page reads as lower-confidence than a dated one with a named author.
How to do it
- Add a visible "last updated" date to evergreen guides and comparison pages, and actually update them on that cadence.
- Attribute articles to a real named author with a short bio and (ideally) an Author schema/sameAs link to their LinkedIn.
- Cite your own primary data or research where you have it (usage stats, survey results) - original research is exactly the kind of source LLMs like to attribute.
- Fix or remove outdated claims (old pricing, deprecated features, dead competitor names) the moment they go stale - do not let old content quietly rot while new pages are being written.
How to verify. Spot-check your 10 highest-traffic pages: each should have a visible date and, where relevant, a named author byline.
"Updated July 2026 by Jane Doe, Head of Growth" under the H1 of a guide, linked to a short author bio.Register your presence where AI reads from
A large share of what AI assistants "know" about a company does not come from your own website at all - it comes from a small set of third-party sources they trust and re-crawl often. Get your business properly listed on each one.
Why it matters. Wikidata is a structured knowledge graph that many LLMs and Google's Knowledge Graph draw entity facts from directly (founding date, category, key people, official website). A missing or wrong Wikidata entry is a common source of a model confusing you with an unrelated company of the same name.
How to do it
- Check wikidata.org for an existing item for your company name first - do not create a duplicate.
- If none exists and you meet basic notability (real product, real users, some independent coverage), create an item with instance-of (e.g. "software" or "business"), inception date, official website, and country.
- Link your Wikidata item from your official site's Organization JSON-LD sameAs array, and vice versa where the Wikidata item has a "official website" property.
- Wikipedia notability is a much higher bar than Wikidata - do not attempt a Wikipedia article until you have genuine independent press coverage; a promotional, unsourced Wikipedia page usually gets deleted and can hurt more than help.
How to verify. Search "[your company] wikidata" and confirm an item exists with your correct official website property (P856).
Wikidata item type "instance of: software company", properties: official website, inception, country, LinkedIn ID.Why it matters. Crunchbase and LinkedIn are two of the most-crawled sources of "who is this company" ground truth for AI training and retrieval - a thin or outdated profile here directly feeds an outdated AI answer about your funding, team size, or category.
How to do it
- Claim or create your Crunchbase profile: correct company description, category tags, founding date, website, and funding info if applicable.
- Complete your LinkedIn Company Page in full: tagline, about section (mirror your "what is X" definition from the Content section), industry, size, website, specialties.
- Keep both in sync with your actual current positioning - a Crunchbase description that describes your product from two pivots ago is actively misleading an AI summarizer.
- Post product updates to your LinkedIn Company Page periodically - active pages get re-crawled more often than dormant ones.
How to verify. Open both profiles in an incognito window and read them as a stranger would - do they match what your homepage says today?
LinkedIn About: "AskRank tracks how often ChatGPT, Claude, Perplexity, and Gemini mention your brand, so indie SaaS founders can measure and improve their AI visibility."Why it matters. Google Business Profile feeds Google's Knowledge Panel and increasingly informs Google AI Overviews for brand and local queries - relevant even for a SaaS company if you have a registered business address or serve a specific market.
How to do it
- Claim or create your Google Business Profile with the correct category, website, and description.
- Fill in every available field (hours, description, services) rather than leaving a bare-minimum listing - thin profiles rank and get surfaced less.
- Respond to any reviews that appear - review responses are themselves a signal of an actively-managed, legitimate business.
How to verify. Search "[your company name]" in Google and confirm a Knowledge Panel appears on the right with your correct details.
Category: "Software company", with website, description, and logo all populated.Why it matters. Perplexity and ChatGPT frequently cite G2, Capterra, and Product Hunt directly when answering "best tools for X" and "is [product] good" prompts - these sites are trusted third-party sources exactly because they are not written by the vendor.
How to do it
- Claim your product listing on G2 and Capterra with an accurate category, description, and screenshots.
- Launch (or re-launch a major update) on Product Hunt - the launch page itself becomes a durable, frequently-cited page even after launch day.
- Systematically ask satisfied customers for reviews on G2/Capterra/Trustpilot after a good support interaction or renewal - a handful of detailed, specific reviews outweighs many generic five-star ones.
- Respond to every review, positive or negative - active engagement is itself a trust signal these platforms (and by extension, AI summarizers) weigh.
How to verify. Search "[your product] reviews" on Perplexity and check whether G2/Capterra/Trustpilot show up as cited sources in the answer.
A G2 profile with 10+ reviews averaging 4.5+ stars and a filled-out "Categories" field matching how buyers actually search.Why it matters. Curated tool directories (including AskRank's own toolpicker.com dataset) and independent "best of" roundups are exactly the kind of aggregated, third-party list an AI engine reaches for when asked to recommend options in a category.
How to do it
- Submit your product to category-relevant directories (e.g. toolpicker.com and similar AI-tool/SaaS directories) with an accurate, specific description - not generic marketing copy.
- Identify the top 10-20 "best [category] tools" articles ranking for your category today and reach out to the authors with a specific, honest pitch for inclusion (what you do differently, who it is for).
- Prioritize directories/roundups that are themselves well-established and frequently updated - inclusion in a dead, unmaintained list has little effect.
How to verify. Search "best [your category] tools" and check whether you appear in the top few organic and AI-generated results within 60-90 days of outreach.
A submission to a SaaS directory with category, one-line pitch, logo, and pricing tier clearly filled in - not a placeholder listing.Why it matters. News and established blog coverage carries high trust weight for AI engines' source ranking - a handful of genuine press mentions do more for how confidently a model describes your company than a large volume of low-authority backlinks.
How to do it
- Pitch a specific, newsworthy angle (a real data point, a genuine launch, a contrarian take on your category) to journalists and newsletter writers who already cover your space - not a generic "we launched" release.
- Write guest posts for respected blogs in your category rather than low-quality guest-post-for-link mills; one placement on a well-known industry blog outweighs a dozen on unknown sites.
- Track every mention you earn and add the strongest ones as a simple "As seen in" strip with logos on your homepage - it also reinforces the same signal to visitors and to any model reading the page directly.
How to verify. Search "[your company]" in Google News or Perplexity and confirm at least a few independent (non-owned) results appear.
A quoted mention in a relevant industry newsletter, linked back to your site, plus the newsletter's name added to an "As seen in" homepage strip.Monitor and keep improving
AEO/GEO is not a one-time project - model behavior, competitor content, and your own product all change continuously. This is what AskRank is built to do for you on an ongoing schedule.
Why it matters. Manually checking "does the AI mention us" a handful of times gives you a snapshot with high variance - LLM answers are non-deterministic, so a real read needs repeated sampling across the actual prompts your buyers ask, on a schedule.
How to do it
- Set up your brand and 50-500 real buyer prompts in AskRank (an AI-generated starting set from your domain, editable) covering the questions and comparisons your buyers actually ask.
- Let AskRank poll ChatGPT, Claude, Perplexity, and Gemini on a schedule and track your Visibility Score, position, and sentiment over time rather than one-off spot checks.
- Review your dashboard weekly, not just after a change - visibility drifts even without you touching anything, as competitors publish and models update.
How to verify. Your AskRank dashboard shows a non-zero Visibility Score with a trend line building up over at least 2-3 weeks of polling.
A 75-prompt set on the Pro plan polled 3x/day across 4 LLMs, with a weekly trend chart and per-LLM breakdown.Why it matters. A flat visibility score in isolation is hard to interpret - knowing that a named competitor is mentioned in 80% of the same prompts where you appear in 20% tells you exactly where the gap is and which of their pages/listings to study.
How to do it
- Add your 3-5 real competitors to AskRank's competitor tracking so every poll captures relative Share of Voice, not just your own mention rate.
- When a competitor outranks you on a specific prompt, look at what their site/listings do differently for that specific question (their FAQ, their comparison page, their G2 reviews) rather than guessing.
- Re-run this comparison monthly - competitive visibility shifts as competitors ship new content or listings.
How to verify. Your AskRank competitor chart shows a clear per-competitor Share of Voice breakdown, updated on the same schedule as your own polling.
A Share of Voice chart showing you at 34%, Competitor A at 41%, Competitor B at 25% across the same 75 tracked prompts.Why it matters. A visibility drop of 10%+ almost always has a fixable cause (a competitor launched new content, a model update changed how it weighs a source, one of your pages went down) - catching it within days instead of discovering it a quarter later at renewal time is the entire value of monitoring.
How to do it
- Turn on AskRank's Telegram and/or email alerts for your brand so a visibility drop over 10% notifies you within a day, not silently in a dashboard nobody checks.
- When an alert fires, check what changed: a competitor's recent content, your own site uptime/robots.txt, or a shift in which LLM providers are driving the drop.
- Treat every alert as a prompt to revisit this checklist for the specific section most likely to be the cause (technical foundation first if it is sudden and across-the-board; content/authority if it is gradual).
How to verify. Trigger a test alert (or wait for a real one) and confirm it reaches your configured Telegram chat or email within the expected window.
"Visibility dropped - Brand: Acme, LLM: Perplexity, Now: 22.0%, Was (7 days): 38.5%, Change: -16.5%" delivered to Telegram the same day it happens.Skip the guesswork on whether it's working
AskRank tracks your real visibility across ChatGPT, Claude, Perplexity, and Gemini on a schedule, so you can see which sections of this checklist actually moved your Visibility Score.
Frequently asked questions
Common questions about AEO/GEO and how to use this playbook.
AEO (Answer Engine Optimization) and GEO (Generative Engine Optimization) are the practices of optimizing content so AI assistants like ChatGPT, Perplexity, Claude, and Gemini mention and recommend your brand in their answers - as opposed to SEO, which optimizes for ranking in traditional search results pages. The two overlap heavily (crawlability, structured data, authority) but AEO/GEO cares about being quoted inside an AI-generated answer, not just ranking a blue link.
Technical fixes (robots.txt, llms.txt, structured data) can be re-indexed within days to a few weeks, since AI crawlers re-crawl frequently. Content and presence changes (FAQ sections, Wikidata, review platforms) typically take 4-8 weeks to show up in answers, since they depend on the next model update or retrieval refresh. Authority-building (press, guest content, Reddit presence) is the slowest, often 2-3 months - track it with AskRank rather than guessing.
Start with the Technical Foundation section - it is the prerequisite for everything else, and most items take under an hour each. Run the free Website Scan first to see exactly which technical items you are missing, then move on to Content and Presence & Registration, which tend to have the highest impact per hour invested for most SaaS companies.
AskRank's Website Scan automatically checks every item in the Technical Foundation section for you. The rest of the checklist (content, registrations, authority building) is manual work you or your team does - AskRank's core product is the monitoring layer: it tracks whether your efforts are actually moving the needle across ChatGPT, Claude, Perplexity, and Gemini, so you are not guessing.
Yes - these are exactly the kind of trusted, frequently-recrawled third-party sources AI engines pull entity facts and recommendations from, and most of them take under 30 minutes to set up or claim. For an early-stage company, an accurate Wikidata entity and a complete G2/Capterra profile often move the needle more than another blog post, because they are the sources models already trust by default.
Track your Visibility Score in AskRank before you start and again every few weeks as you work through sections. Because LLM answers vary run to run, a single manual check tells you very little - AskRank samples your real prompts across all 4 engines on a schedule so you see a real trend line, not noise.
Now measure whether it's working
AskRank runs your real prompts across ChatGPT, Perplexity, Claude, and Gemini on a schedule - so every item you check off here shows up as a real, trackable change in your Visibility Score.