How to Get ChatGPT and Perplexity to Recommend Your Business #

You get ChatGPT and Perplexity to recommend your business by becoming a name that third-party sources already repeat — directories, review sites, "best of" listicles, and credible industry pages — then making your own site easy to verify. That is the load-bearing insight most owners miss. These systems rarely invent a vendor out of thin air. They synthesize what the open web already says. If the open web never names you, the answer engines rarely will either.

I'm William Spurlock, an AI Solutions Architect and Fractional AI CTO who has spent 20,000+ hours building agentic systems and helping operators fix the gap between "we have a website" and "AI actually knows we exist." This pillar is the owner-facing action plan for one primary question: how do I get ChatGPT to recommend my business? Mechanics live in other posts. Here we stay on what you do this week, this quarter, and every month after.

The cost of inaction is concrete. When a buyer asks ChatGPT for a recommendation in your category and never hears your name, that lead does not bounce back to Google later for a fair rematch. It goes to the two or three competitors that do get named. You lose the conversation before your sales team ever sees it.

---

How Do I Get ChatGPT to Recommend My Business? #

Get named consistently on trusted third-party pages, keep your business facts identical everywhere, and publish clear pages that answer the exact questions buyers ask AI. ChatGPT does not run a "pay for placement" marketplace. You cannot buy a guaranteed slot. What you can do is raise the odds that when the model searches and synthesizes, your name is already in the evidence pile.

Think in revenue terms, not SEO hobbies. A named recommendation in ChatGPT is a soft referral: the buyer asked for options, the system answered with a shortlist, and your competitors just got a free warm intro. Your job is to become shortlist-worthy.

The three layers that actually move recommendations #

1. Third-party proof — Other sites naming you as a real, specific option in a category and place. This is the biggest lever.
2. Entity consistency — Same legal name, phone, address, service area, and category across your site, Google Business Profile, directories, and schema. An entity here just means "the machine-readable identity of your business as one distinct thing in the world."
3. Answer-ready pages — Pages that state who you serve, where, what you sell, pricing ranges or packages if you can publish them, and proof (reviews, case results, credentials) in plain language.

If you only polish your homepage and ignore the first layer, you are polishing the wrong surface. In my AI-visibility work, the owners who stay invisible almost always have a decent site and almost no third-party footprint that AI systems can cite.

What "recommended" looks like in practice #

When someone prompts ChatGPT with variations of:

• "best [service] near [city]"
• "who should I hire for [job] in [industry]"
• "recommend a [business type] for [use case]"

…a recommendation means your brand appears in the answer as a named option, ideally with a reason ("known for X," "serves Y area," "strong reviews for Z"). Citation of your URL is nice. Being named is the money event. Buyers remember names. They click names. They call names.

What you cannot control (say this out loud to your team) #

Controllable	Not controllable
Consistency of NAP (name, address, phone)	Exact shortlist for every user and phrasing
Presence on review and directory surfaces	Buying a guaranteed ChatGPT placement
Publishing clear service + location pages	A single ranking dashboard like classic SEO
Earning mentions on "best of" and industry pages	Identical answers across ChatGPT, Perplexity, Gemini, and Copilot
Monthly prompt testing and iteration	One-and-done "set it and forget it" results

Results vary by phrasing, by whether search is on, by the user's location signals, and by what sources the system retrieves that day. Treat this like reputation management with a measurement loop — not like buying ad inventory.

Owner checklist you can start this week #

• Audit your name in AI today. Open ChatGPT (with search/browsing available on your plan) and Perplexity. Ask five buyer-style prompts in your category. Screenshot the answers. That is your baseline.
• Fix identity mismatches. If your site says "Acme HVAC LLC," Google says "Acme Heating," and Yelp says "Acme Air," you are training machines that you are three different companies.
• Pick five third-party targets. One Google Business Profile cleanup, two directories, one review push, one "best [category] in [city]" or industry list opportunity. Put owners' names next to each target.
• Publish one answer page. A single page that states: who you are, what you sell, who you serve, where, proof, and a clear CTA. No fluff paragraphs. Facts first.
• Add Organization / LocalBusiness schema. Machines read structured facts faster than marketing copy. A JSON-LD example is later in this post.

For the deeper "how the models choose" explanation, read how ChatGPT and Perplexity actually decide which businesses to recommend. This pillar stays on the owner playbook: what to do.

Map your buyer prompts before you change anything #

Before you rewrite pages or pitch listicles, write down the actual sentences a buyer would type when they are ready to spend money. Owners often test vanity prompts ("Tell me about my brand") and then conclude AI "doesn't work." That is the wrong test.

Use this prompt inventory worksheet:

Buyer situation	Prompt they type	Revenue moment
New in town, needs a vendor fast	"best [service] near [city] open now"	Emergency / urgent hire
Comparing two options	" [competitor] alternatives in [city]"	Displacement opportunity
Budget-conscious but quality-aware	"affordable [service] with good reviews in [city]"	Mid-market conversion
Niche constraint	"[service] for [industry / building type]"	Specialty premium
Trust check	"who do locals recommend for [service]"	Reputation-driven pick

If your site and third-party profiles never speak to those situations in plain language, ChatGPT has nothing specific to recommend you for. Generic businesses get generic silence.

The "named recommendation" economics #

You do not need a fantasy ROI spreadsheet. You need a back-of-napkin model your ops brain trusts:

1. Estimate how many category recommendation prompts happen in your market each month (even a rough guess is fine).
2. Assume a shortlist of three names per answer.
3. Assume one of those three gets contacted.
4. Apply your normal close rate and average job value.

When you are never on the shortlist, your share of that contact stream is zero. When you appear on one-third of relevant prompts, you start participating in a referral channel you currently donate to competitors. That is the business case for AI visibility work — not "rank better for fun."

---

How Does ChatGPT Decide Which Websites to Mention or Recommend? #

ChatGPT recommends businesses by synthesizing training knowledge plus, when search is available, retrieved web results — then naming the options that look most consistent, credible, and relevant to the prompt. It is not scrolling Google the way a human does and picking the blue-link #1. It is assembling an answer from evidence.

Two plain definitions that matter for owners:

• Retrieval means the system looks up current web pages related to the question instead of answering only from memory.
• Grounding means the answer is tied to those retrieved sources so the model is less likely to invent a vendor that is not supported by evidence.

When search is off or the query is generic, ChatGPT may fall back to well-known brands burned into training data. That is why national chains show up more often than the excellent local shop with no public footprint. When search is on, the shortlist becomes more sensitive to what directories third-party pages and your own site currently say.

What the model is looking for (business translation) #

Signal the system favors	What it means for your P&L
Repeated third-party mentions	Independent sites already treat you as a real option
Clear category + location match	You show up for "near me" and "in [city]" prompts
Consistent facts across sources	Fewer "are these the same company?" doubts
Review volume and recency on major surfaces	Social proof the model can cite without guessing
Specific, extractable claims on your site	"Serves Dallas plumbers" beats "we care about excellence"
Structured data (schema)	Machines can read your identity without interpreting poetry

I am deliberately not turning this into a model-internals lecture. You do not need transformer architecture to fix a revenue leak. You need to become easy to verify and hard to confuse with a competitor.

Why "ranking #1 on Google" is not the same job #

Classic SEO still matters — strong pages and links help you appear in the evidence set. But ChatGPT can name a business that is not #1 if that business is clearly described on multiple trusted pages. Conversely, a #1 page that is thin, contradictory, or blocked from crawlers can still lose the AI shortlist. If you want a fast self-check on whether your site is even readable to these systems, run the 15-minute AI visibility audit before you spend another dollar on ads.

A practical mental model for owners #

Imagine ChatGPT as a junior analyst with a 30-second deadline:

1. Understand the buyer's ask (category, place, constraints).
2. Pull a handful of sources that look relevant.
3. Prefer names that appear in more than one credible place.
4. Prefer sources that agree on the basics (who / where / what).
5. Write a short answer that sounds helpful and avoid inventing companies that the sources never mentioned.

Your strategy falls out of that model: get into the sources, stay consistent, make extraction easy.

Relevance filters owners accidentally fail #

Even when you appear somewhere on the web, ChatGPT can still skip you if the match is weak:

• Wrong category language. You call yourself a "solutions partner." Buyers ask for a "bookkeeper." The model matches the buyer's words.
• Wrong geography. You serve three counties but never name them. A competitor names each city on their GBP and site.
• Wrong customer type. You want commercial accounts but every public page talks about homeowners.
• Outdated evidence. Your best case study is from 2019 with no date refresh. Fresher competitor pages win the retrieve-and-synthesize pass.
• Conflicting offers. Directory A says you do roofing only. Site says you do roofing and solar. GBP says solar only. The analyst (the model) gets confused and picks a cleaner entity.

Fix the match before you chase more volume. Ten perfect, consistent mentions beat fifty messy ones.

What "good evidence" looks like in a retrieved page #

When ChatGPT's search pulls a page, pages that help you get named usually have:

• Your exact business name in the title or first screen of content
• A category noun buyers use ("plumber," "immigration lawyer," "B2B payroll")
• A place noun ("Austin," "Queens," "nationwide for remote teams")
• A reason to exist on the shortlist (specialty, rating summary, credential, notable client type)
• A link or clear identity match back to your real site or profile

If a PR mention says "a local contractor" without naming you, it does almost nothing for recommendations. Insist on the name.

---

Why Does ChatGPT Never Mention My Business When I Ask About My Industry? #

ChatGPT skips you because the public evidence for your business is thin, inconsistent, or missing from the surfaces it trusts — not because your product is bad. Most "we're invisible" cases I see are identity and proof problems, not product problems.

Here is the uncomfortable version: if you ask ChatGPT about your industry and it names three competitors, those competitors are already documented as options somewhere the system can use. You are not. That silence is expensive. Every unanswered prompt in your category is a silent RFP you never received.

The seven most common reasons you get skipped #

1. No third-party shortlist presence. Zero listings on the directories buyers and AI systems both use. No "best of" mentions. No industry association pages.
2. Thin or generic website copy. "Quality service since 1998" does not tell a model which city, which services, which customer type, or why you beat the next option.
3. NAP chaos. Different names and addresses across the web fracture your entity.
4. Reviews live only in private inboxes. Screenshots in your office do not count. Public review surfaces do.
5. Service area is implied, never stated. Humans infer. Models need the city and radius in text.
6. Site is hard to crawl or extract. Heavy client-only rendering, empty shells, or blocked important pages can hide the facts. (Fix the site mistakes that hide you from AI search — that is a separate deep dive.)
7. You are asking the wrong test prompts. Brand vanity searches ("tell me about Acme LLC") are not buyer searches. Buyers ask for categories and outcomes.

The "competitor got named, I didn't" diagnosis #

Run this 20-minute exercise with your marketing lead or yourself:

Step	Action	Pass / Fail signal
1	Ask ChatGPT 5 category prompts without your brand name	Note every named competitor
2	Google each competitor + your city/category	See if they appear on listicles, directories, Reddit threads
3	Search your brand on the same surfaces	Empty results = root cause found
4	Compare Google Business Profile completeness	Missing hours, categories, photos, Q&A = easy fix
5	Check if your site states services + location in plain HTML text	If the facts only live in images/PDFs, extractability is weak

In my experience, owners want a technical silver bullet. The actual fix is usually: get named where the competitors are already named, then make your own pages unambiguous.

Industry-by-industry failure patterns #

Different industries fail AI recommendations for different boring reasons:

Industry pattern	Typical invisibility cause	First fix
Home services	Incomplete GBP + weak review velocity	GBP + weekly review asks
Professional services	No association pages; site is résumé fluff	Credentials + practice-area pages with city
Local retail / hospitality	Menus and hours trapped in images/PDFs	Text hours, menu items, location facts
B2B SaaS	G2/Capterra empty; no category listicles	Review platforms + "best tools for X" mentions
Healthcare / regulated	Thin public pages out of caution	Compliant service pages + directory accuracy
Trades with multiple DBAs	Entity split across names	Pick a canonical public name and align everything

If you recognize your row, start there. Do not copy a SaaS playbook onto a plumbing company or the reverse.

Cost of inaction, said plainly again #

If ChatGPT and Perplexity keep naming three rivals and never you, you are funding their sales pipeline with your silence. Ad spend cannot fully patch that. Ads buy attention when someone is already searching your keywords. AI recommendations intercept the decision earlier — at the "who are my options?" moment. Staying invisible there means you keep paying for the bottom of the funnel while someone else owns the shortlist.

That cost compounds. Competitors who win the shortlist get the call, the review, the next mention, and a stronger entity next month. You do not just lose today's lead — you fall further behind the evidence pile the models will read tomorrow.

---

How Perplexity Differs From ChatGPT — and What That Changes for You #

Perplexity is more citation-forward and source-visible; ChatGPT is more conversational and may name options with lighter visible sourcing depending on mode. For an owner, that changes where you invest, not whether you invest.

Perplexity's product habit is to show sources beside answers. That makes being a citable page more important: clear titles, dated facts, author identity where relevant, and pages that answer one question cleanly. ChatGPT can still recommend you from a blend of search results and known entities, and the user may not click through to inspect every source. Both reward third-party proof. Perplexity simply makes the citation layer more obvious to the human reading the answer.

ChatGPT vs Perplexity: source behavior for business owners #

Factor	ChatGPT (with search available)	Perplexity
User experience	Conversational shortlist, variable visible citations	Answer with prominent source links
What "winning" looks like	Being named as a recommended option	Being cited and/or named with a clickable source
Sensitivity to third-party lists	High — listicles and directories feed the shortlist	High — plus stronger preference for pages it can quote
Best owner move	Earn mentions + keep entity consistent	Earn mentions + publish citation-ready pages
Local business angle	Strong when GBP + local directories + reviews align	Strong when local pages are specific and quotable
Failure mode	Names national brands from memory if evidence is thin	Skips you if no crawlable, specific sources support the claim
Measurement	Prompt tests + screenshots over time	Prompt tests + whether your URL appears in citations

If your goal is "get Perplexity to cite my site as a source," that is a sibling playbook — how to get Perplexity to cite your website as a source goes deep on citation mechanics. For this pillar, remember the shared rule: third-party naming still does most of the heavy lifting for recommendations.

What to do differently for each system #

For ChatGPT recommendations

• Prioritize being listed as a named business on directories, review sites, and "best X in Y" pages.
• Make sure your category and city appear in the same sentence on your site ("Dallas commercial HVAC repair for multi-family properties").
• Build review velocity on the platforms buyers in your industry actually use.

For Perplexity citations and mentions

• Publish pages that answer one buyer question with a clear first sentence.
• Add dates, specifics, and proof that a model can quote without hallucinating a number you never published.
• Earn links and mentions from pages Perplexity already trusts in your niche (industry publishers, local news, associations).

For both

• Keep NAP identical.
• Ship Organization / LocalBusiness schema.
• Re-test monthly with a fixed prompt set (template later in this post).

A simple split of weekly effort #

If you have five focused hours a week for AI visibility, try this split until your scorecard says otherwise:

Hours	ChatGPT-weighted work	Perplexity-weighted work
2	Directories, GBP, reviews, listicle pitches	Same third-party work (shared foundation)
1.5	Service + location page clarity for naming	Answer-first pages built to be quoted
1	Competitor shortlist monitoring	Citation checks on your target queries
0.5	Prompt testing screenshots	Prompt testing + note which URLs got cited

Notice that most of the hours are shared. You are not running two unrelated marketing programs. You are running one evidence program with two readouts.

When to prioritize one engine over the other #

• Prioritize ChatGPT naming if your buyers are consumers or SMB owners who casually ask ChatGPT for "who should I use."
• Prioritize Perplexity citations if your buyers are researchers, consultants, or technical evaluators who click sources before they call.
• Prioritize both equally if you sell a considered purchase above a few thousand dollars — those buyers often check more than one assistant.

As of mid-2026, I tell clients to measure both monthly even if one engine dominates their audience. The work overlaps too much to pretend you can ignore either forever.

---

The Third-Party Surfaces That Drive Recommendations (And How to Get Named) #

A large share of the work is off your website: directories, review platforms, listicles, associations, and forums are the surfaces answer engines reuse when they shortlist businesses. Your site is necessary. It is not sufficient.

This is the section where owners either lean in or stay stuck. Polishing H1 tags while your Google Business Profile is half-empty is a classic miss. AI systems need independent confirmation that you are a real option.

Surface map: where recommendations are born #

Surface	What it is	Why AI cares	Owner action this month
Google Business Profile	Your local identity card	Strong local entity signal + reviews	Complete categories, services, hours, photos, Q&A
Vertical directories	Industry-specific listings (legal, medical, home services, SaaS catalogs)	Category-matched third-party naming	Claim/optimize the 2–3 directories buyers already use
Review sites (Yelp, Angi, G2, Capterra, etc.)	Public proof of delivery	Corroboration that customers exist	Ask for reviews with a simple weekly process
"Best X in Y" listicles	Publisher roundups	Direct shortlist feedstock	Pitch or earn inclusion with a clear differentiator
Industry associations / chambers	Membership pages	Credibility + entity corroboration	Get listed with correct NAP and link
Local news / PR	Mentions in articles	Independent narrative about you	Offer a newsworthy angle, not a press-release dump
Reddit / forums	Real-user threads	Conversational evidence (noisy but influential)	Earn mentions via genuine helpful participation; never spam
Partner / vendor pages	"Certified partner" or "featured provider" lists	B2B trust transfer	Activate partner marketing you already paid for

How to get named on each surface without wasting a quarter #

1. Google Business Profile (do this before anything fancy)

• Primary category must match how buyers ask AI for help.
• Secondary categories for real services only — do not keyword-stuff.
• Services list with plain descriptions.
• Weekly photo updates beat a one-time dump from 2019.
• Respond to every review. The response text is more public evidence.

2. Directories that matter in your niche

Stop chasing 50 directories. Pick the ones that already appear when you search your category. If ChatGPT or Perplexity cite a directory in your niche today, that directory is on your target list. Claim the listing, match NAP, add categories, add a short description that states city + service + customer type.

3. Review velocity as an operating habit

Reviews are not a branding nice-to-have. They are third-party sentences that say you delivered. Set a weekly quota: ask five recent happy customers. Make the ask specific ("Would you share how the install went on Google?"). Track count and average monthly. Stale review profiles lose to active ones.

4. Listicles and "best of" pages

These pages are recommendation catnip. Find the ones ranking or getting cited for "best [service] in [city]" and "top [category] for [use case]." Then earn a spot with:

• A one-page pitch: who you serve, proof, what makes you different in one sentence
• Willingness to provide a quote, photo, or data point the publisher can use
• A unique angle (specialty niche, response time, guarantee, credential) — not "we're passionate"

5. Forums and Reddit (handle carefully)

AI systems do read public forum text. That does not mean you should drop links in every thread. The winning pattern is earned mentions: answer questions helpfully under a real identity, let other people name you when you deserve it, and fix public misinformation about your brand when it appears. Spam gets ignored by humans and eventually by systems that downweight low-quality sources.

6. Digital PR that serves AI visibility

If you invest in PR, optimize for mention + accurate entity facts, not vanity impressions alone. A niche industry article that names your company, city, and specialty can outperform a vague national mention. Ask every placement to include your canonical name, city, and specialty in plain text — not a vague "local provider" nod.

Recommendation levers ranked by effort vs payoff #

Lever	Typical effort	Payoff for recommendations	Notes
Fix NAP + GBP completeness	Low	High	Fastest win for local/service businesses
Review generation system	Low–medium	High	Compounds monthly
Claim top 3 niche directories	Low	Medium–high	Stop at quality; ignore junk directories
Answer-ready service pages	Medium	Medium–high	Needed for citations and clarity
Organization / LocalBusiness schema	Low	Medium	Cheap insurance for extractability
Earn 1–2 listicle / association mentions	Medium–high	High	Often the difference vs competitors
Full site redesign "for AI"	High	Variable	Only after identity + third-party gaps are closed
Buying random guest posts at scale	Medium–high	Low–variable	Easy to waste money on junk domains
Hoping ads alone fix AI shortlists	Ongoing cost	Low for this job	Ads and AI visibility solve different moments

My take, after watching owners spend in the wrong order: do not start with a redesign. Start with identity consistency and third-party naming. Redesigns are for conversion after you are findable.

Outreach scripts that do not sound like spam #

When you pitch a listicle editor, association directory manager, or local publisher, keep it short and useful:

Subject: Quick addition for your [City] [Category] roundup

Hi [Name] —

I'm [Your Name] with [Business], a [one-line specialty] serving [city/area].
I noticed your [article/directory] currently lists [Competitor A] and [Competitor B].

We might be a fit because:
- [Specific differentiator buyers care about]
- [Proof: years, credential, review count, niche]
- [Service area clarity]

Happy to send a 3-sentence blurb, logo, and a customer-ready proof point.
No fluff needed on your side — just accurate listing details.

Thanks,
[Name]
[Phone] · [Site] · [GBP or primary profile link]

You are not begging for a favor. You are reducing the editor's work to add a correct option. That is how inclusions actually happen.

How to prioritize which third-party targets first #

Score each opportunity from 1–5 on three axes, then multiply:

1. Buyer overlap — Do your customers already trust this surface?
2. AI reuse likelihood — Does this surface already appear in ChatGPT/Perplexity answers in your niche?
3. Attainability — Can you realistically get listed or mentioned in 30–60 days?

A directory with a 5 on AI reuse and a 2 on attainability might still beat a random blog with a 5 on attainability and a 1 on AI reuse. Chase evidence the models already drink from.

---

Make Your Website Easy to Verify (Schema, Pages, and Facts) #

Your website should read like a fact sheet a skeptical analyst can trust in under a minute: who, what, where, for whom, with proof. That is how you support recommendations that third-party sources start.

The minimum page set for recommendation support #

Publish or clean up these pages so each one answers a single job:

1. Home — Category, location, primary offer, proof, CTA.
2. Services — One page per major service with outcomes and who it is for.
3. Locations / service area — Cities and radii stated in text.
4. About — Real people, credentials, years operating, what you will not do.
5. Proof — Case studies, before/after metrics you can stand behind, review embeds.
6. FAQ — The exact questions buyers ask AI, answered in two to four sentences each.

Every page should include the business name and primary city or market in natural language. Not stuffed. Just present.

JSON-LD schema example (Organization / LocalBusiness) #

Schema means machine-readable markup that states your business facts in a standard format (schema.org). Below is an encouraged pattern — adapt the fields to your real data. Do not invent reviews or ratings.

{
  "@context": "https://schema.org",
  "@type": "LocalBusiness",
  "@id": "https://www.example.com/#business",
  "name": "Example Service Co",
  "url": "https://www.example.com/",
  "image": "https://www.example.com/images/storefront.jpg",
  "telephone": "+1-555-555-0100",
  "email": "hello@example.com",
  "description": "Example Service Co provides commercial HVAC repair and maintenance for multi-family properties in Dallas–Fort Worth.",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "123 Main Street",
    "addressLocality": "Dallas",
    "addressRegion": "TX",
    "postalCode": "75201",
    "addressCountry": "US"
  },
  "geo": {
    "@type": "GeoCoordinates",
    "latitude": 32.7767,
    "longitude": -96.7970
  },
  "openingHoursSpecification": [
    {
      "@type": "OpeningHoursSpecification",
      "dayOfWeek": [
        "Monday",
        "Tuesday",
        "Wednesday",
        "Thursday",
        "Friday"
      ],
      "opens": "08:00",
      "closes": "17:00"
    }
  ],
  "areaServed": [
    {
      "@type": "City",
      "name": "Dallas"
    },
    {
      "@type": "City",
      "name": "Fort Worth"
    }
  ],
  "priceRange": "$$",
  "sameAs": [
    "https://www.facebook.com/exampleserviceCo",
    "https://www.linkedin.com/company/example-service-co",
    "https://www.yelp.com/biz/example-service-co-dallas"
  ]
}

Place JSON-LD in your site in the manner your CMS or developer supports. The business win is simple: fewer opportunities for the model to misread your identity.

On-page writing rules that help AI name you correctly #

• Lead every key page with a direct sentence: who you help + what you do + where.
• Prefer specifics over adjectives. "2-hour emergency response in Dallas zip codes 75201–75219" beats "fast, friendly service."
• Keep claims consistent with your directory listings.
• Put important facts in HTML text, not only in images.
• Update dated proof ("As of August 2026…") when numbers change.

Entity consistency checklist (print this) #

Before you call the website "done," verify these match character-for-character where possible:

Field	Website	GBP	Top directory	Schema	Social / sameAs
Legal / public name
Primary phone
Primary address
Primary category
Service area statement
Primary URL
Hours

One mismatched phone number has wrecked more local AI visibility projects than a missing blog ever did.

What not to put in schema #

• Fake aggregate ratings
• Services you do not actually offer
• Cities you do not serve
• Old phone numbers "for tracking" that conflict with your public number
• Keyword-stuffed business names ("Best Dallas Plumber Acme LLC")

Schema is a truth format. If you lie to machines, you train a fragile entity that collapses the first time sources disagree.

---

Your 30 / 60 / 90-Day Implementation Sequence #

Ship identity fixes and third-party presence in the first 30 days, citation-ready pages and listicle outreach by day 60, then measurement, iteration, and harder PR wins by day 90. Sequence beats intensity. Owners who try to do everything in one weekend usually ship nothing that compounds.

Days 1–30: Stop the bleeding #

Goal: Become a consistent, findable entity with a baseline measurement.

Week	Owner actions	Done when
Week 1	Run 10 ChatGPT + 10 Perplexity prompts; screenshot; list named competitors	Baseline folder exists
Week 1	NAP audit across site, GBP, top directories; fix mismatches	One canonical name/address/phone
Week 2	Complete Google Business Profile (categories, services, photos, Q&A)	Profile completeness ~100%
Week 2	Claim/optimize top 3 niche directories	Listings live with correct NAP
Week 3	Launch weekly review-ask process (5 asks/week)	First new reviews arriving
Week 3	Add LocalBusiness / Organization schema	Schema validates against your real facts
Week 4	Publish or rewrite 1 core service page + 1 service-area page	Pages state who/what/where in sentence one
Week 4	Re-run the same 20 prompts; note any movement	Month-1 delta documented

Budget reality for month one: mostly time and process, not a huge media buy. The expensive mistake is skipping this and buying a redesign.

Days 31–60: Earn the shortlist #

Goal: Get named on at least one high-quality third-party surface and make your site citation-ready.

• Build a target list of 15 "best of," association, and industry pages that already mention competitors.
• Pitch 5 of them with a one-page differentiator + proof pack.
• Publish 3 FAQ-style pages that mirror buyer prompts ("How much does X cost in [city]?", "How to choose a [provider] for [use case]?").
• Align blog or resource content to those same questions if you have a content engine — answer-first, not fluff-first.
• Continue review velocity without pause.
• If you are local-first, deepen city coverage the right way: real service pages for real service areas, not doorway spam. Name the cities and neighborhoods you actually cover, with proof that belongs to those places.

Days 61–90: Compound and operationalize #

Goal: Turn AI visibility into a monthly operating cadence, not a project that dies after launch week.

• Land or advance 1–2 earned mentions (listicle, association, partner page, or local news).
• Expand schema only where it matches reality (FAQ schema on real FAQ pages, Product/Service where appropriate).
• Build a simple internal dashboard: prompt set, date, whether you were named, whether you were cited, competitors named, notes.
• Assign an owner (marketing lead, VA with a checklist, or you) who runs the monthly test on the same calendar day.
• Decide what to stop doing. If a directory never appears in AI answers and never sends leads, stop feeding it.

30 / 60 / 90 scoreboard #

Horizon	Must-have outcomes	Nice-to-have
30 days	Clean NAP, strong GBP, 3 directories, schema live, baseline prompts	First new reviews
60 days	3 answer pages live, 5 pitches sent, review cadence stable	First listicle reply or inclusion
90 days	Monthly test ritual, 1+ earned third-party mention, clear competitor gap list	Repeat citations on Perplexity for a target query

Once the 90-day foundation is in place, convert the scoreboard into a recurring monthly ops checklist so the work does not depend on launch-week energy.

Roles and ownership so this does not die #

AI visibility fails in businesses where "marketing will handle it" means nobody owns the calendar. Assign names:

Workstream	Suggested owner	Cadence
NAP / GBP maintenance	Ops manager or owner	Weekly 30 minutes
Review asks	Frontline staff with a script	5 asks/week
Directory claims	VA with a checklist	One-time + quarterly audit
Listicle / PR outreach	Marketing lead or founder	2 pitches/week in days 31–90
Answer page publishing	Founder + writer/designer	1 page/week in days 31–60
Monthly prompt testing	Marketing lead	Same day each month
Decision meeting	Founder	30 minutes after each monthly test

If you are a solo operator, that table still helps — it becomes a time-block list, not a headcount fantasy. Block the hours or the work will lose to whatever fire is loudest.

Budget ranges without fake precision #

I will not invent a universal price. What I can say from client work:

• Low cash / higher time: You can execute the 30-day layer with sweat equity if you already have a workable site.
• Medium cash: Paying for review software, a VA for directory cleanup, and light PR outreach usually beats paying for a full rebrand first.
• Higher cash: Rebuild the site for answer-readiness after identity and third-party gaps are clear — so the new site has the right pages, schema, and proof modules on day one.

Spend in the order of learning. Measure first. Then buy acceleration.

---

How to Test and Track Whether You Are Being Recommended #

You track AI recommendations with a fixed monthly prompt set, screenshots, and a simple scorecard — because there is no official ranking dashboard. Anyone selling you a magic "ChatGPT rank #1" product is selling comfort, not a control panel that OpenAI publishes for local businesses.

Build your monthly prompt set #

Write prompts the way buyers talk, not the way marketers write keyword lists. Keep a master list of 20–30 prompts, then run a core 12 every month so comparisons stay honest.

Core prompt templates (customize the brackets):

1. Best [service] in [city]
2. Who should I hire for [job] in [city]?
3. Recommend a [business type] for [use case]
4. Top [service] companies near [neighborhood / zip]
5. [Service] for [industry] businesses in [city]
6. Alternatives to [well-known competitor] in [city]
7. Affordable [service] in [city] with good reviews
8. Emergency [service] near me (run while located in-market if possible)
9. What's the best [service] for [specific constraint]?
10. [Service] that specializes in [niche]
11. Compare [competitor A] and other [service] options in [city]
12. Who do locals recommend for [service] in [city]?

Run the same set in:

• ChatGPT with search available
• Perplexity (default search answer)
• Optionally a third engine you care about (Gemini, Copilot) if that is where your buyers are

Scorecard columns that matter #

Date	Engine	Prompt	Named? (Y/N)	Position in answer (1/2/3/other)	Cited URL?	Competitors named	Notes / source themes
2026-08-06	ChatGPT	Best …	N	—	N	A, B, C	Competitors on Angi + listicle
2026-08-06	Perplexity	Best …	N	—	N	A, B	Directory pages cited

Track rates over time:

• Name rate — % of prompts where your brand appears
• Citation rate — % where your URL is cited (especially on Perplexity)
• Share of voice — your name appearances ÷ all business-name appearances in the set
• Competitor displacement — prompts where you replaced a rival vs last month

Keep the scorecard boring and consistent. Fancy dashboards can wait until the monthly ritual is actually happening.

Testing rules that keep you honest #

• Same prompts, same day of month. Drift in wording destroys trend lines.
• Incognito / clean session when possible. Personalization and chat history can skew results.
• Do not overreact to one lucky answer. Look for 60–90 day trends.
• Separate brand queries from category queries. "What is Acme?" is vanity. "Best commercial cleaner in Austin" is revenue.
• Record whether search/browsing was on. Comparing search-on to search-off is comparing different jobs.
• Screenshot everything. Interfaces change. Your folder of dated screenshots is the audit trail.

What good progress looks like #

Month 1: you may still be invisible. That is a baseline, not a verdict.

Month 2–3: you start appearing on a subset of local or niche prompts where your third-party footprint is strongest.

Month 4+: name rate climbs on the prompts tied to surfaces you won (for example, after a listicle inclusion and a review surge). If nothing moves after 90 days of real third-party work, your targeting is wrong — wrong directories, wrong city framing, or a category mismatch — not "AI hates us."

Expanded monthly prompt pack (copy/paste and customize) #

Use this fuller set when you want a deeper quarterly audit. Still keep a fixed core of 12 for the monthly trend line.

Category + city
- Best [service] in [city] 2026
- Top-rated [service] companies in [city]
- Who is the best [service] near [landmark / neighborhood]?

Problem / outcome phrasing
- I need help with [problem]. Who should I call in [city]?
- Recommend a company that can [outcome] for a [customer type]
- What's the best option for [constraint: budget / timeline / specialty]?

Competitor-adjacent
- Alternatives to [competitor] for [service]
- Is [competitor] good for [use case], or who else should I consider?
- Companies like [competitor] in [city]

Trust / proof phrasing
- [Service] with the best reviews in [city]
- Highly recommended [service] for [niche]
- Which [service] do homeowners / business owners recommend in [city]?

B2B / specialty (if relevant)
- Best [service] for [industry] companies
- [Service] providers that understand [regulation / stack / building type]
- Who should a [role] hire for [job] in [region]?

Run them as a human buyer would. Do not add your brand name into the prompt unless you are testing brand comprehension separately.

Decision rules after each monthly test #

Result pattern	Interpretation	Next action
Still never named	Evidence footprint still too weak or mismatched	Double down on the surfaces competitors win on
Named on 1–2 local prompts only	Local entity improving	Expand reviews + one more city/service page
Named but wrong specialty	Category language mismatch	Rewrite categories and service descriptors
Cited on Perplexity, unnamed on ChatGPT	Citation pages working; shortlist proof still thin	Push listicles/directories harder
Named on ChatGPT, never cited on Perplexity	Brand mentions exist; your pages are not quotable	Strengthen answer-first pages and proof
Spikes then disappears	Fragile / thin evidence	Add corroborating sources so one page change cannot erase you

This is how you manage AI visibility like an operator instead of refreshing ChatGPT for dopamine.

---

A Direct Opinion on Where Owners Waste Money #

The biggest waste I see is buying more website animation before buying third-party proof. Pretty sites that nobody cites still lose the shortlist. Second biggest waste: spraying 100 directory submissions instead of owning the five surfaces your competitors already appear on. Third: treating AI visibility as a one-week agency package with no monthly prompt ritual.

If you only do four things after reading this pillar, do these:

1. Fix NAP + GBP.
2. Install a weekly review ask.
3. Earn one real third-party mention on a page competitors already use.
4. Run the monthly prompt set without fail.

Everything else is amplification.

---

Frequently Asked Questions #

How do I get Perplexity to cite my website? #

Publish specific, quotable pages and earn mentions from sources Perplexity already trusts — then make your facts easy to extract. Clear titles, a direct first-sentence answer, dated details, and consistent business identity matter more than clever slogans. The dedicated Perplexity citation playbook linked earlier in this post covers the page-level citation mechanics in more depth.

How does Perplexity decide which sources to use? #

Perplexity retrieves live web results for the query and favors pages that look relevant, specific, and corroborative enough to support the answer with citations. It is not a pure "highest Google rank wins" button. Third-party pages that already discuss your category — and your own pages when they are clear — both compete for those citation slots.

What is the difference between ChatGPT search and Perplexity search? #

ChatGPT search supports conversational answers that may name businesses with variable visible sourcing; Perplexity is built to show sources beside the answer as a default habit. For owners, that means ChatGPT work skews toward being named on trusted lists, while Perplexity work skews toward being a citable page and a named entity. You still need third-party proof for both.

How do I optimize my website for ChatGPT search mode? #

State your identity, services, and location in plain text, keep facts consistent with directories, add LocalBusiness schema, and create pages that answer buyer questions in the first two sentences. ChatGPT search mode still needs evidence from the web. A vague brochure site gives it nothing reliable to ground on when it retrieves results.

How long does it take to start showing up in ChatGPT recommendations? #

Many businesses see early movement within 60–90 days if they fix identity gaps and earn real third-party mentions — but there is no guaranteed timeline. National brands with years of public proof can appear faster. Thin local footprints take longer. Track monthly; judge on a quarter, not a weekend.

Can I pay ChatGPT or Perplexity to recommend my business? #

No — there is no official buy-a-recommendation placement for organic answers the way you buy a Google Ad. Ads products and sponsored placements, when they exist in an ecosystem, are a different channel from being organically named in an answer. Treat organic AI visibility like reputation and evidence, not like auction inventory.

Does Google Business Profile help ChatGPT recommend local businesses? #

Yes — a complete, consistent Google Business Profile is one of the highest-payoff local entity signals you can control quickly. It will not guarantee a ChatGPT mention by itself, but incomplete or conflicting GBP data is a common reason local businesses never make the shortlist.

What should I do if competitors are always named and I am not? #

Mirror their public evidence footprint: the directories, listicles, review platforms, and association pages where they already appear — then out-execute them on consistency and review velocity. Run the competitor diagnosis table earlier in this post. The gap is usually visible on the open web within an hour.

Do online reviews affect AI recommendations? #

Yes — public reviews are third-party sentences that confirm you deliver, and answer engines reuse that kind of corroboration when shortlisting. Private compliments do not count. Focus on the review platforms your buyers already trust in your category, and keep reviews recent.

Should I create separate pages for every city I serve? #

Create location pages only for areas you truly serve and can describe with real proof — not doorway pages that repeat the same paragraph with a city name swapped. Thin location spam can hurt trust. Specific service-area pages with genuine detail help both humans and AI systems place you correctly.

Will adding FAQ schema make ChatGPT recommend my business? #

FAQ schema helps machines extract Q&A cleanly, but it will not create recommendations by itself if no third-party sources treat you as an option. Use FAQ markup on real questions you answer on the page. Pair it with the third-party and identity work in this pillar. Schema is a clarity tool, not a shortlist cheat code.

How often should I re-test ChatGPT and Perplexity prompts? #

Re-test your fixed core prompt set once a month, and run a deeper pack once a quarter. Weekly testing usually creates noise and anxiety without better decisions. Monthly is enough to see whether reviews, listings, and mentions are changing outcomes.

Does entity SEO matter if I only care about recommendations? #

Yes — if AI systems cannot tell that your site, profiles, and mentions refer to one business, they struggle to recommend you confidently. Entity work is the boring consistency layer underneath naming: one canonical name, matching NAP, clear category, and corroborating third-party pages that all point at the same company.

---

If you want a second set of eyes on whether ChatGPT and Perplexity can even find a coherent version of your business, I can run an AI-visibility audit and map the exact gaps between you and the competitors currently getting named. That is the fastest path from "we think we're online" to "we know what to fix for recommendations." If you also need the website rebuilt so it is answer-ready for AIO/AEO — clear entities, schema, and pages AI can cite — that is the other half of the studio work I ship.