Search is no longer one results page. A buyer may discover a problem on social, research it on Google, ask an AI system to summarize options, compare a shortlist, and convert on a site or through an agent. The practical job is to make your business findable, answerable, referenceable, comparable, and easy to choose.
This guide adapts the useful funnel in James Dooley's modern search optimization stack, then checks it against current platform documentation. The acronyms are a planning model. They are not six proven algorithms, and they should not become six disconnected strategies.
The stack in one minute
- SEO — get found: make useful pages crawlable, indexable, relevant, and competitive in traditional search.
- SMO — get discovered: distribute useful ideas where your market researches and talks.
- AEO — be the answer: publish direct, supported passages that resolve specific questions.
- GEO — get referenced: make your claims, entities, and evidence clear enough to retrieve and cite.
- DEO — get chosen: expose the facts needed to compare your offer against a user's constraints.
- SXO — convert: turn qualified discovery into a useful next step and a measurable business result.
The shared foundation is more important than the labels: clear entities, useful content, accurate data, independent evidence, and trust.
What the platforms actually document
Google: foundational SEO still applies
Google says its generative AI features are rooted in core Search ranking and quality systems. Its AI features may use retrieval-augmented generation and query fan-out, but there are no special technical requirements or magic AI markup required to appear. A page still needs to be accessible, indexable, and eligible to show a snippet.
Source: Google's guide to optimizing for generative AI features and AI features and your website.
OpenAI: search crawling and training are separate controls
OpenAI documents OAI-SearchBot for surfacing sites in ChatGPT search and GPTBot for content that may be used to improve foundation models. A site can allow the search bot while disallowing the training bot. That distinction matters: “available to ChatGPT search” and “available for model training” are not the same setting.
OpenAI also says ChatGPT search referral URLs include utm_source=chatgpt.com, giving site owners a concrete referral metric to track.
Source: OpenAI crawler documentation and Publishers and Developers FAQ.
Perplexity: crawler eligibility is not a citation promise
Perplexity documents PerplexityBot for its search index and Perplexity-User for user-triggered fetches. It recommends allowing the bot and its published IP ranges, including through a WAF. That makes content accessible to the retrieval system; it does not guarantee that a page will be cited for a particular answer.
Source: Perplexity crawler documentation.
Structured data: describe visible facts, do not invent them
Structured data can make page meaning more explicit, but Google requires markup to represent the page's visible content and does not guarantee a search feature. Use schema to mirror facts a visitor can verify — organization identity, service details, offers, price, availability, and policies — not to manufacture authority.
Source: Google's structured data guidelines.
llms.txt: useful option, not access control
llms.txt is an open community proposal for giving agents a concise, curated map of a site. It can be useful, especially for documentation-heavy sites. It is not a replacement for robots.txt or sitemap.xml, and no major search platform documents it as a ranking requirement. Treat it as an optional navigation artifact, not a visibility score multiplier.
Source: the llms.txt proposal.
Layer 1: SEO — make the evidence retrievable
SEO remains the retrieval layer: technical accessibility, indexation, helpful content, internal links, descriptive metadata, site reputation, and a usable page experience. If a search system cannot discover or understand a page, the downstream answer and recommendation layers have nothing reliable to work with.
Ship
- Indexable pages that match real customer questions and decision criteria.
- Server-rendered, crawlable facts with stable URLs and descriptive internal links.
- Sitemaps, canonical URLs, accurate titles, and intentional snippet controls.
- Useful original material: first-party data, examples, expert methods, or documented experience.
Measure
- Indexed coverage and crawl failures.
- Query impressions, position, clicks, and qualified organic conversions.
- Backlinks and earned references to the specific evidence you published.
Layer 2: SMO — earn discovery and corroboration
SMO is best treated as a parallel discovery and evidence-distribution layer. Publish where your audience actually learns: professional networks, video, communities, forums, newsletters, podcasts, or specialist publications. The goal is not to spray identical posts everywhere. It is to put verifiable expertise into the conversations buyers already use.
Third-party discussion can corroborate what a company says about itself, but there is no defensible universal formula that turns social mentions into AI citations. Measure the observable path instead.
Measure
- Qualified reach, saves, discussion, and referral visits.
- Independent brand mentions and links from relevant sources.
- Assisted conversions from those sources.
Layer 3: AEO — make the answer extractable
Answer engine optimization asks a narrow question: can a person or system find a supported answer in the page without reverse-engineering the marketing copy?
Ship
- A direct answer immediately after a question-shaped heading.
- Definitions that name the entity and avoid vague pronouns.
- Explicit comparisons with criteria, tradeoffs, and disqualifiers.
- Claims linked to primary evidence, with dates and scope.
- Tables, lists, or steps only when they make relationships easier to parse.
FAQ markup is not AEO by itself. A weak answer with schema is still a weak answer, and markup must match visible content.
Layer 4: GEO — make the source worth referencing
Generative engine optimization concerns whether a retrieval-and-generation system can discover, understand, and use your material when constructing an answer. The controllable work is familiar: publish uniquely useful evidence, define entities consistently, make claims easy to verify, and earn independent corroboration.
Ship
- Original research, benchmarks, calculators, datasets, or documented processes.
- Named authors or reviewers with relevant experience.
- Stable pages with visible publication and update dates.
- Clear source links and precise language around uncertainty.
- Consistent organization, product, and service facts across owned profiles.
Measure
- A fixed, versioned query basket by engine, market, locale, and date.
- Brand mentions, linked citations, cited URLs, and competitor share in captured responses.
- ChatGPT referral sessions using documented UTM attribution.
- Downstream qualified actions, not citations alone.
Layer 5: DEO — make the offer comparable and selectable
Decision engine optimization is an emerging planning label, not an established industry standard or a documented ranking system. The underlying problem is real: an assistant cannot confidently compare or recommend an offer when basic constraints are missing.
Publish the facts a decision needs
- Fit: who the product or service is for — and who it is not for.
- Price: the actual price, a defensible range, or the variables that determine a quote.
- Availability: inventory, capacity, locations, delivery windows, or engagement start dates.
- Capabilities: features, integrations, methods, and constraints.
- Policies: cancellation, returns, privacy, warranties, and guarantees.
- Proof: sourced outcomes, reviews, credentials, and independently verifiable evidence.
Put those facts in visible HTML first. Add matching structured data where a supported vocabulary fits. Never hide the only useful facts inside an image, gated PDF, sales call, or invented schema field.
Layer 6: SXO — convert attention into a useful outcome
The search journey still reaches an interface: a page, form, booking flow, checkout, phone call, or agent-compatible action. Search experience optimization ensures the path is fast, accessible, credible, and proportionate to the user's intent.
Measure
- Qualified conversion rate by source and landing page.
- Form completion, booking completion, checkout completion, and error rate.
- Pipeline and revenue with honest attribution windows.
- Accessibility, performance, and task-completion failures.
A measurement model that does not lie
Do not collapse all six layers into one “AI visibility score” and pretend it proves live citations or commercial impact. Keep the evidence separate:
- Technical readiness: can documented crawlers access and render the content?
- Search visibility: is the page indexed, shown, and clicked?
- Answer coverage: does the site directly answer the target question set?
- AI observation: was the brand mentioned or cited in a captured, versioned query run?
- Decision readiness: are the attributes needed to compare the offer explicit and current?
- Business outcome: did qualified traffic become pipeline or revenue?
A crawler check is not a citation check. A citation is not a recommendation. A recommendation is not a conversion. Preserve that chain and you can improve the weak link instead of celebrating a vanity metric.
The 30-day implementation order
- Baseline: capture indexation, referrals, conversions, and a fixed multi-engine query basket.
- Fix retrieval: resolve crawl, WAF, rendering, canonical, sitemap, and internal-link issues.
- Clarify entities: make organization, offer, author, and policy facts consistent and visible.
- Close answer gaps: publish direct answers and comparisons for the highest-intent questions.
- Add decision facts: state fit, price, availability, constraints, and proof.
- Earn corroboration: distribute original evidence to relevant independent sources and communities.
- Remove conversion friction: test the next step on mobile, with assistive technology, and under failure states.
- Re-run the same evidence set: compare like with like and report unknowns honestly.
Bottom line
SEO is not dead. It is the retrieval foundation of a broader decision journey. The winning system does not chase every acronym independently. It publishes useful evidence once, makes that evidence accessible and easy to interpret, exposes the facts needed to choose, distributes it where buyers research, and measures the path all the way to a real business outcome.