The Ask-Engine Playbook: Get Named in ChatGPT, Then Convert the Click
Buyers stopped typing fragments into Google and started asking ChatGPT who to hire. The companies that get named are the ones whose sites answer that question in one extractable sentence, then prove it with decision pages, schema, inbound citations, and a visit that loads fast enough to finish the job.

Article Summary
- Ask-engine optimization is the work of getting named when a buyer asks ChatGPT, Claude, Gemini, Perplexity, or Copilot who to hire.
- Lock one buyer and one problem, then repeat that sentence on the homepage, product pages, FAQs, and comparison pages so the model can slot you.
- Decision pages (versus, alternatives, best-for, how to switch) are the fastest path into AI answers because they already contain the reasoning the model needs.
- Schema, an
llms.txtmap for large language models, and inbound citations are how machines resolve that you are a real business with a real category. - Speed is the remaining click. Xeal ships look, feel, and load time with proprietary delivery methods so the visit you still get converts.
The ask-engine window
Search engine optimization still decides who gets the open-web clicks that remain. SparkToro and Similarweb put that number at 232 clicks out of every 1,000 US Google searches in early 2026. You still fight for those clicks. You also fight for the name inside the answer.
Ask-engine optimization is Xeal's name for that second fight. A buyer does not type "LMS software." They ask, "What is the best LMS for a 400-student private school that needs parent messaging?" Claude, Gemini, and ChatGPT write a shortlist. One or two names survive. The rest of the category becomes a footnote.
Generative engine optimization is the citation and recommendation layer. Answer engine optimization is the one-line Siri read. Search engine optimization is the ranking slot. Local search is the map pack, Apple Maps, Bing Places, and the reviews those surfaces read. The generator on this site builds a blueprint for all four, plus the speed of the page that has to finish the booking.
PRO TIP Ask ChatGPT, Claude, Gemini, and Perplexity the same buyer question this week. Screenshot the names. Check again after you ship the pages in this playbook. The protocol for doing that without arguing from vibes is in We Asked Four AI Engines to Recommend a Business.
Lock the questions buyers actually ask
Keyword SEO still captures people who are browsing. Ask-engine queries usually fire when someone is choosing. That is a different intent, and it needs a different page map.
Act like your buyer. Type the questions they would type. Note the follow-ups the model suggests. Write down which products appear in every answer. Those repeating prompts are the lock. Your site has to become the key.
Family the questions by stage, not by keyword volume:
- Best-for: "What is the best [category] for [specific user]?"
- Fit: "Does [category] work for [constraint: size, budget, compliance, location]?"
- Compare: "Is [you] better than [competitor] for [job]?"
- Alternatives: "What are the alternatives to [competitor]?"
- Switch: "How hard is it to switch from [competitor]?"
- After: "What happens after I sign up / book / ship the device?"
- Local: "Who is open now for [emergency job] near [city]?"
If your content only answers the first question, you appear occasionally. If it answers the sequence, models start treating you as the reliable option across related prompts. That repetition is how recommendations compound. Run the mapping in the ASK-ENGINE blueprint generator. It writes the prompt list from your buyer, category, city, and competitors.
Lock one sentence, then repeat it
Broad category language fits hundreds of products. "Project management platform." "CRM tool." "Marketing agency." ChatGPT's job is to give an accurate answer. It will not confidently name a company it cannot slot.
Use this formula and put it on the homepage in the first screen:
[Category] for [specific user] who need to [specific outcome].
Examples that machines can lift:
- "Project management for remote product teams of 40 to 200."
- "CRM for early-stage B2B startups that sell in a 30-day cycle."
- "Emergency plumbing for Owasso and north Tulsa homeowners."
Then use the same wording on the product page, the comparison pages, the FAQs, the Google Business Profile, Apple Business Connect, Bing Places, and the llms.txt file. Humans can tolerate five different pitches. Models read five pitches as five different companies. Language alignment is how the association forms. Once it forms, recommendations start to stick.
The clearer the sentence, the easier it is for an LLM to put your name in the answer.
Make every money page extractable in 20 seconds
Most SEO content is written like a magazine feature. Long introductions. Buried conclusions. Models need to extract. If a human cannot understand the page in under 20 seconds, the model will skip you or hedge.
Lead with the answer. Use headings that match the question. State tradeoffs in the open. Kill the 400-word throat-clear. Bing's webmaster guidelines say the same thing in older language: semantic HTML, useful titles, logical heading hierarchy, key information early, so content can be retrieved, grounded, and cited.
This is why a restructure of what you already have often beats a publishing calendar. Most companies already have a homepage, service pages, and FAQs. They are just hard to lift. The 40-point scorecard in How to Get AI Search Engines to Recommend Your Business is the audit. The generator turns that audit into a named URL list.
PRO TIP Paste your homepage into ChatGPT and ask, "In one sentence, who is this for and what problem does it solve?" If the model hedges, your positioning is still a slogan. Fix the sentence on the page before you buy more content.
Build the pages that help people choose
Ask-engine recommendations fire at the moment of choice. Informational blog posts still have a job. They are rarely the page the model cites when someone is picking a tool, a shop, or a clinic.
High-leverage decision URLs:
/[competitor]-alternatives//[you]-vs-[competitor]//best-[category]-for-[job]//how-to-switch-from-[competitor]/- Honest pricing, process, and "what happens after you book" pages
Write them fairly. Name the tradeoff. If you only praise yourself, the model treats the page as an ad and looks for a third-party source. If you help the buyer decide, your page becomes a natural citation. This is the same pattern we use in the B2B SaaS playbook and the product recommendation article. Comparison pages are how shopping engines resolve attributes. Service businesses need the same format for jobs, not SKUs.
Cover the whole decision, not one blog topic
Buyers ask a sequence. What should I use. How does it compare. How hard is the switch. What happens after I pay. Models prefer brands that show up at more than one of those steps.
Map the journey before you publish another "thoughts on the industry" post. The generator writes a page inventory by stage: discovery, qualification, comparison, risk, onboarding, and proof. Two blog posts a month was a staffing compromise. Coverage of the questions that close the deal is the job. That argument is in Two Blog Posts a Month Is a 2018 Answer.
Schema, llms.txt, and inbound proof
A beautiful page that only a human can parse is invisible to the systems that now write the first impression. Three machine-layer jobs sit under every content plan.
Schema
Mark the entity so Google, Bing, and answer engines can resolve you as one business. Local companies use LocalBusiness (or a subtype). Nationwide brands use Organization. Service pages get Service. Products and SaaS get Product, Offer, or SoftwareApplication. FAQs get FAQPage. Add sameAs to the profiles that already mention you. Citations are entity resolution now, not a map-pack trick. That is the whole point of Citations Aren't for the Map Pack Anymore.
llms.txt is for models
llms.txt is a map at the root of your site for large language models. ChatGPT, Claude, Gemini, Perplexity, Copilot, Grok, Llama, Mistral, phone assistants, on-device models, and more. Ship a clean file that states who you are and which pages to cite. Google Search Central said on 15 June 2026 that Search itself ignores the file, including its generative features. Expected. That file was built for the models that write the answer, not for the ten blue links fewer people click. See ours at xeal.net/llms.txt. The generator drafts yours from the same sentence you locked on the homepage.
Inbound links and earned mentions
Models look for third-party corroboration. A site that only talks about itself is a brochure. Inbound work in 2026 is source-tier work: editorial mentions, industry directories, review platforms, association pages, and partner citations that use the same name, address, phone, and category. That is how the entity graph holds still. Practical tactics live in Build Links for Free and 100+ Backlink Ideas. The high-leverage version is getting quoted, which is Get Quoted, Get Cited and the publicity service.
Local search is still a stack, not a Google listing
If you serve a city, the ask-engine layer sits on top of a local stack you cannot skip. Google Business Profile. Apple Business Connect for Apple Maps and Siri. Bing Places for Copilot. Name, address, and phone consistency. Review recency, because reviews are training data now. Situation pages for "open now," emergency, and the job the buyer actually has.
Local SEO advice aimed at a single-location bakery will hurt a national mail-in shop. Geography is a decision, not a default. That fork is in You're Not a Local Business. The generator asks you local, regional, national, or global before it writes the map-pack work.
Speed is how the remaining click pays
Getting named is half the job. The visit that remains has to open on a real phone, hold still, and make the next step obvious. Google's practical thresholds are still the ones operators should manage: Largest Contentful Paint at or under 2.5 seconds, Interaction to Next Paint at or under 200 milliseconds, Cumulative Layout Shift at or under 0.1.
Slow pages lose the click you paid years to earn. They also lose Quality Score if you run ads, which is Quality Score Is a Discount. Third-party tags that measure the page often make the page worse, which is The Measurement Tax.
Xeal treats look, feel, and speed as one delivery system. We use proprietary methods to ship marketing sites that stay fast as content grows: lean HTML, intentional media, a cache policy that does not fight itself, and a publishing path that does not drag a plugin tax onto every URL. The public version of the argument is The Flat-File Comeback and the 2026 Small Business Speed Index. Xeal ships the implementation.
Use the generator, then decide who ships it
The ASK-ENGINE, GEO, SEO, and Local SEO website blueprint lives on the tools page. You enter the business, the buyer, the problem, the city, and the competitors. It writes:
- The positioning sentence to repeat everywhere
- The ask-engine prompt map by decision stage
- The decision pages and supporting SEO cluster, as named URLs
- The local stack, if you serve a place
- The schema types and an
llms.txtdraft - The inbound source tiers
- The speed spec
- A 90-day build order you can print and take to a meeting
Print the URL list. Run it in-house, or send it to Xeal. For content, placements, and technical wiring in 21 days, use the Gold Plan. For a site rebuilt to hold that content at speed, start with Super Fast Loading Websites plus Get Recommended by AI and Search.
Putting it together
Map the questions buyers ask AI. Lock one sentence. Structure pages so a model can lift them. Build the decision layer. Align the language. Cover the journey. Wire schema, llms.txt, and inbound proof. Make the visit fast.
Most companies will keep publishing blog posts, chasing keywords, and buying links the old way. Those jobs still matter. The decision layer moved. The companies that win are the ones easiest for a large language model to recommend, and easiest for a human to hire once the click arrives.
Generate the blueprint. Screenshot what the engines say this week. Then either ship the list, or send it to Xeal and we will.
Generate your website blueprint
Takes a few minutes. Prints. Downloads as Markdown. Bring the output to the call if you want Xeal to run the 90 days.
FAQ
What is ask-engine optimization?
Ask-engine optimization is the work of getting named when a buyer asks ChatGPT, Claude, Gemini, Perplexity, Copilot, or a phone assistant who to hire. It sits next to search engine optimization (the ranking slot), answer engine optimization (the one-line spoken answer), and generative engine optimization (the written recommendation with a citation). The ask is the prompt. The engine is the model. Your site has to be extractable, consistent, and corroborated, or the model names someone else.
Does Google Search use llms.txt?
No. Google Search Central said on 15 June 2026 that Search ignores llms.txt, including its generative features. Ship the file anyway. It is a map for large language models: ChatGPT, Claude, Gemini, Perplexity, Copilot, Grok, and the rest. We rank on Google with extractable HTML, a consistent entity, and public proof. We future-proof the models with the map. See xeal.net/llms.txt.
Should I still do traditional SEO and backlinks?
Yes. SEO wins the clicks that remain. Inbound links and mentions are how both Google and answer engines resolve that you are a real entity in a real category. The mix changed. Decision pages, schema, review recency, and speed moved up the list. Keyword-only blog calendars moved down. Keep earning links. Earn them from sources a model would trust, and point them at pages that answer a buying question.
How is this different from a keyword tool?
A keyword tool starts from fragments and volume. This generator starts from the questions a buyer types into an LLM when they are ready to choose, then writes the site architecture, schema, local stack, llms.txt draft, inbound tiers, and speed spec around that ask. Volume still matters for SEO. The named recommendation matters for GEO. You need both maps.
Can I implement the blueprint myself?
Yes. Print the plan: named URLs, schema types, and a 90-day order. Ship it in-house, or send the list to Xeal. Look, feel, and speed are where proprietary delivery methods matter, and that work is easier to buy than to staff.
Where does local SEO fit if I am a national brand?
If you do not serve a walk-in geography, skip city pages and the map pack. Keep Bing, Apple Business Connect only if you have real locations, and put the energy into category pages, comparison pages, and entity consistency. Forcing local SEO onto a national mail-in or SaaS brand creates conflicting signals. The generator asks geography first so the local section only appears when it should.
Sources
- SparkToro + Similarweb: 2026 zero-click study (232 open-web clicks per 1,000 US searches)
- Google Search Central, 15 June 2026: Search ignores llms.txt
- Bing Webmaster Guidelines: semantic HTML, sitemaps, Copilot retrieval
- web.dev: Core Web Vitals thresholds (LCP, INP, CLS)
- Apple Business Connect
- Xeal: llms.txt
- Xeal: ASK-ENGINE website blueprint generator
- Xeal: 40-point AI recommendation checklist
- Xeal: four-engine recommendation protocol
- Xeal: The Old Playbook Is Broken
- Xeal: citations as entity resolution
- Xeal: 2026 Small Business Speed Index
- Xeal: Get Recommended by AI and Search
What operators say after working with Xeal
Quotes from people who hired Tony and Xeal for websites, SEO, publicity, and strategy.
I have known Tony since 1991. I have always been impressed with his technical ability, creative talent, and verve. He has made an impact with his business, and I am privileged to be counted in his network. I would recommend any venture Tony undertakes.
Tony has a great eye for design. He has a breadth of knowledge for web design and internet marketing. His knowledge and skill are superior. I highly recommend him and his company to anyone serious about improving their internet and web presence. Our internet site has really done well with Tony's help.
Want help putting this into production?
If this article named a problem you already feel, tell me what is stuck. We will turn the playbook into systems your team can run.
