Why AI Recommends Your Competitor (And What To Do About It)

You searched for a plumber in your city on ChatGPT. It named three companies. Yours wasn't one of them.

This is happening to service business owners every day - and most of them have no idea why. They've done everything "right": optimized their Google Business Profile, collected reviews, kept their website current. But when a potential customer asks an AI assistant for a recommendation, a competitor shows up instead.

The rules changed. Here's what's actually driving AI recommendations, and what you can do to show up where the leads are going.

---

The Problem With Everything You've Read About Local SEO

Most content about local SEO was written for a Google world - one where ranking is about backlinks, page speed, keyword density, and GBP completeness. Those signals still matter for traditional search. But AI search works differently.

When someone asks ChatGPT "who's the best HVAC company in Phoenix," they're not getting a ranked list of websites. They're getting a synthesized recommendation built from what the AI has ingested, weighted by signals that don't appear in your Google Analytics dashboard.

The businesses that consistently show up in AI recommendations share five characteristics. Most of them have nothing to do with your website.

---

The 5 Trust Signals AI Search Engines Actually Use

1. Citation Consistency Across Data Aggregators

AI language models are trained on web data - and that data includes business directories, industry listings, local data aggregators, and news mentions. When your business name, address, and phone number appear consistently across dozens of sources, the AI has high confidence that your business is real, established, and correctly described.

The problem: most service businesses have inconsistent NAP data. Your phone number changed two years ago, but 40 listings still show the old number. Your business name is "Smith HVAC" on Google, "Smith Heating and Cooling" on Yelp, and "Smith H&C LLC" on Angi. To a human, these are obviously the same business. To an AI training on scraped web data, they're three different entities with conflicting information.

ChatGPT and Perplexity both underweight businesses with inconsistent citations because inconsistency signals low reliability - the same way a human reader would distrust a source that can't keep its own facts straight.

What to do: Audit your citations across the top 40 data aggregators. Priority platforms: Google Business Profile, Yelp, Apple Maps, Bing Places, Foursquare, and the aggregators that feed everything else (Infogroup, Acxiom, Localeze). Every listing needs identical NAP data, the same business category, and the same service area description.

2. Review Volume Combined With Response Rate

Reviews are not just a social proof signal for humans - they're data about your business's relationship with its customers. AI models trained on review content use both volume and pattern analysis.

Here's what most guides miss: raw review count matters less than review velocity and response rate. A business with 200 reviews collected over five years, with 40% of them unanswered, ranks lower than a business with 80 reviews collected in the past 18 months, all of them responded to within 24 hours.

The mechanism: consistent review acquisition signals that a business is actively operating. Consistent response signals that the business engages with customers. Both are markers of a legitimate, ongoing operation - exactly what AI systems want to recommend.

Different AI engines weight this differently. Google's AI Overview leans heavily on GBP review signals it has direct access to. Perplexity pulls from multiple review platforms (Yelp, Google, Facebook) and synthesizes across them. ChatGPT relies more on training data, which means older, well-reviewed businesses sometimes outrank newer ones - a gap that's closing as models update more frequently.

What to do: Build a review acquisition sequence that fires automatically after job completion. The goal is consistent velocity, not a spike followed by silence. A plumber getting 4-6 new reviews per month outperforms one who ran a campaign and got 30 reviews in a week two years ago.

3. Schema Markup Implementing Local Business Vocabulary

This is the most technical of the five signals and the most commonly skipped.

Schema markup is structured data embedded in your website's code that tells machines - including AI crawlers - exactly what your business does, where it operates, what its hours are, and how it's categorized. Google invented the vocabulary. AI systems read it.

Most service business websites have either no schema or outdated schema that doesn't include newer local business properties. Specifically:

• areaServed: The specific cities, neighborhoods, or zip codes you operate in. Not "Phoenix Metro Area" - actual named entities.
• hasOfferCatalog: Your services as a structured list, each with a name and description.
• aggregateRating: Your review score marked up so AI doesn't have to guess.
• openingHoursSpecification: Your actual hours, including emergency and after-hours availability.

When an AI model is choosing between two plumbers in Scottsdale and one has detailed schema telling it exactly which zip codes that plumber serves, what services they offer, and their verified rating - and the other has none of that - the answer is predictable.

What to do: Implement LocalBusiness schema on your homepage and service pages. Use Google's Rich Results Test to verify it's reading correctly. This is a one-time technical task with compounding returns as AI models reindex.

4. Topical Authority in Local Content

AI search doesn't just recommend businesses - it recommends businesses it considers authoritative on the problem the user is describing.

When someone asks Perplexity "my furnace is making a clicking noise, who should I call in Denver," it's not just looking for HVAC companies in Denver. It's looking for HVAC companies in Denver that have demonstrated knowledge about furnace problems. A company with a post titled "Why Your Furnace Clicks and When It's Dangerous" that mentions Denver neighborhoods, describes the diagnostic process, and references local climate factors is going to outrank a company with a generic "Contact Us" page.

This is why the skyscraper approach to content works for local service businesses. Most page-one results for local service queries are either vendor blogs with shallow tips or national aggregators with no local specificity. A locally-authored piece that actually answers the technical question, with real service area context, is genuinely better content - and AI can tell.

What to do: Write one authoritative piece per major service category per quarter. Not "5 Signs You Need a New Furnace" - something specific enough that a homeowner with the problem would read it twice. Mention specific local factors: water hardness, climate extremes, local code requirements. This is topical authority that compounds over time.

5. Third-Party Mentions and Entity Associations

The fifth signal is the hardest to manufacture and the most powerful: your business being mentioned positively in third-party content.

When local news sites, industry publications, chamber websites, neighborhood blogs, and local directories mention your business by name - especially in the context of solving a specific problem - AI models register those as trust signals. You're not just a business that claims to be good. You're a business that other sources have found worth mentioning.

This is why backlink strategy matters even in an AI search world - not because AI counts backlinks the way Google's old PageRank did, but because the presence of your business name in credible third-party content trains the model to associate your business with legitimacy and local relevance.

Sponsoring the local youth sports league, being interviewed for the local business journal, earning a mention in a "best of" roundup from a regional publication - these all create the kind of third-party entity associations that AI search treats as social proof.

What to do: Identify 10 local or industry publications that cover your category. Pitch one story per quarter: a project that solved an unusual problem, a community initiative you supported, a trend you're seeing in your market. One genuine mention in a credible local source is worth more than 50 directory submissions.

---

Why ChatGPT, Perplexity, and Google AI Give Different Answers

If you've tested all three, you've noticed they don't always name the same businesses. Here's why:

ChatGPT (GPT-4 and newer): Trained on a corpus with a knowledge cutoff, which means its recommendations are influenced by historical web data. Businesses with long, consistent track records of web presence tend to do well. The model doesn't browse the web in real-time (unless using tools), so a business that recently did everything right may not see results for months.

Perplexity: Actively crawls and retrieves current web content before generating answers. This means recently updated content, fresh reviews, and new third-party mentions have faster impact. Perplexity citations show you exactly what sources it's drawing from - these are the specific pages you need to appear in.

Google AI Overview: Has access to Google's full index, GBP data, and review database in real-time. Of the three, it's most closely correlated with traditional local SEO signals - but it also layers in semantic understanding that means genuine relevance beats keyword stuffing.

The implication: optimizing for one isn't enough. The businesses that appear consistently across ChatGPT, Google AI and Perplexity are the ones that have covered all five trust signals.

---

The Competitive Teardown: What Your Competitor Did That You Didn't

Take the HVAC company that shows up when you search your city on ChatGPT. Run it through the five-signal checklist:

• Consistent NAP across 80+ directories? Almost certainly.
• Review velocity of 4-8 per month with consistent responses? Check their recent reviews.
• LocalBusiness schema on their site? Right-click, view source, search for "schema.org".
• Service-specific content with local context? Browse their blog or service pages.
• Third-party mentions? Search their business name in quotes alongside your city.

The gap is almost never one thing. It's the compound effect of all five signals being consistently maintained over 12-18 months. That's the frustrating part - and the encouraging part. You're not behind because your competitor did something clever. You're behind because they started earlier on fundamentals you can begin today.

---

How to Measure Your Current Position

Before fixing anything, establish a baseline. Three measurements matter:

1. AI mention frequency: Search your business category plus your city in ChatGPT, Perplexity, and Google AI. Are you in the results? For which queries? How often compared to your top competitor? Do this across 10-15 different query phrasings.

2. Citation consistency score: Audit your listing accuracy across the key aggregators and surface the specific discrepancies - wrong phone number, inconsistent name format, missing service categories - that are suppressing your visibility. FoundScore's local presence audit does this automatically and shows you exactly where the gaps are.

3. Review velocity: Calculate your average reviews per month for the last 12 months. Compare it to your top local competitor's pace. The gap between those two numbers is your benchmark.

These three numbers tell you which signal to fix first. Citation consistency is typically the fastest fix - one afternoon of corrections can close gaps that have been working against you for years. Review velocity is the longest build. Content authority sits in the middle: one good piece takes a week to write and 3-6 months to reach full effect.

---

The Next 90 Days

If you started today:

Weeks 1-2 - Citation cleanup: Identify and correct every inconsistent NAP listing. This is the single highest-leverage technical task because it affects how every AI engine reads your business's basic identity. FoundScore's audit will surface every discrepancy; fix them top-down by aggregator authority.

Weeks 3-4 - Schema implementation: Get LocalBusiness schema on every key page. Include areaServed with actual named locations, hasOfferCatalog with your real services, and openingHoursSpecification. Verify with Google's Rich Results Test. This is a one-time task that keeps working indefinitely.

Month 2 - Review velocity: Launch an automated review acquisition sequence tied to job completion. Set a minimum target: 4 new reviews per month, every month. Respond to every review within 24 hours - even the negative ones, especially the negative ones.

Month 3 - Content authority: Publish one authoritative piece of content in your most competitive service category. Not a tip list - a real answer to a real question your customers ask. Make it locally specific. This is the beginning of your topical authority stack.

At the end of 90 days, run your baseline measurements again. If you've executed consistently, you'll start appearing in AI recommendations you weren't in before.

---

The Bottom Line

AI search isn't replacing local SEO - it's raising the bar on what it means to have an established, trustworthy local presence. The businesses showing up in AI recommendations aren't doing something exotic. They're doing the fundamentals at a consistently higher level than their competitors.

FoundScore exists to audit exactly this gap - to show you, specifically, which of the five signals are holding you back and what the fix looks like. The audit takes minutes. The gap, once you see it, is impossible to unsee.

Your competitor is showing up because they built trust across multiple signals over time. There's no shortcut to that. But there's also no mystery. You know what to do.

Start with the audit. The rest follows.