How SMPL Measures AI Visibility for Local Businesses

What we test

The 7 engines we test

We test 7 leading AI answer engines systematically. Each engine draws from a distinct mix of signal sources and weights them differently. That is why a business can be fully visible on one engine and completely absent on another, even with the same underlying entity data. Knowing which engines you are missing from, and why, requires testing each one independently.

• Business profile and directory signals: whether the engine can accurately locate and describe your business from the directory and mapping sources it queries for local recommendations.
• Review platform signals: how your reputation data appears in the review ecosystems each engine draws from, and whether those signals are complete and consistent.
• Structured data markup: whether each engine can extract your entity information cleanly from your website, including business type, location, hours, and service area.
• Web indexing and citation signals: how broadly and accurately your business is described across the indexed web that engines use to cross-reference and confirm entity data.
• Geographic and location accuracy: whether your service area, address, and location data are consistent across all signal sources each engine consults.
• Entity consistency: whether your business name, address, phone number, and category match exactly across all signal sources. Inconsistencies suppress citation even when presence is strong.
• Answer-readiness: whether your website content is structured to be cited as a direct answer to the local queries your customers are asking. Engines do not cite content that buries the answer.

The full report names each of the 7 engines individually, shows what each one says about your business, and ranks the gaps by impact.

How we test

The query types we run

We test 3 query types for every client, because each type reveals a different layer of visibility.

• Transactional queries (the "I need help now" query)
  The customer is ready to call. They want a name and a phone number.
  Example: "emergency bee removal Anaheim CA same day"
  This is where phone number accuracy and directory presence decide whether you get the call. AI engines surface a business name, address, and phone number directly in the answer. If that information is wrong or missing, the lead goes to a competitor.
• Informational queries (the "I'm trying to understand my situation" query)
  The customer is not ready to call yet. They want to know what they are dealing with.
  Example: "how do I know if I have a bee colony in my wall vs a swarm"
  This is where FAQ content and educational blog posts create visibility. AI engines quote pages that answer the specific question. A business whose site answers this question in plain, readable HTML becomes the cited source.
• Comparison queries (the "help me pick" query)
  The customer is evaluating options. They want to know who is best and why.
  Example: "best licensed bee removal company Orange County CA"
  This is where credentials, reviews, and verifiable claims determine who gets named. AI engines apply a filter here: they look for specific evidence (a license number, a review count, years in operation) before recommending one business over another.

What we record per engine

For every query on every engine, we record:

• Citation presence: Is the business named in the AI answer at all?
• Recommendation position: Is it the first business named, second, or listed alongside several others?
• NAP accuracy: Is the name, address, and phone number in the AI answer correct? A wrong number in the answer means a lost call.
• Evidence cited: What source did the engine use to justify the recommendation? (Yelp listing, website FAQ, Google Business Profile, Foursquare data, a specific page on the site.)

We run every query live, across all 7 engines, at the time of the audit. We do not use cached results or proxy tools. What we show you is what your customer sees.

How we score

The scoring framework

We score each business across 24 dimensions on a 0-to-120 scale. The score is a baseline (where you stand today) and a target (what moves the number and by how much).

• Platform presence (roughly 25 points): Whether the business has claimed and completed the platforms each AI engine depends on: Google Business Profile, Yelp, Foursquare, and Bing Places. These are the entity anchors. Without them, the engine has nothing to resolve the business against.
• Technical signals (roughly 35 points): The structured data on the website: schema markup (LocalBusiness, Service, FAQPage), canonical tags, server-rendered content, sitemap completeness. These are the signals that tell crawlers what the business does, where it operates, and what questions it answers.
• Content signals (roughly 35 points): The quality and structure of the page content itself: city-specific pages, FAQ answers written for AI citation, opening sentences in direct-answer format, pricing information, hyper-local vocabulary. A page that cannot be quoted in an AI answer because it has no quotable sentences is a page that does not exist for AI purposes.
• Authority and trust signals (roughly 25 points): Review volume and recency, verifiable credentials (license number, years in operation, named owner), third-party citations, and E-E-A-T signals.

The resulting score is not an estimate. It reflects what we found on each dimension, with the evidence documented.

What Covered / Partial / Missing means

• Covered: The signal is fully implemented and working as intended. Example: A business with a verified Google Business Profile that matches its website NAP exactly, has 50+ reviews, and is actively managed scores Covered on GBP alignment.
• Partial: The signal exists but has a gap that limits its effectiveness. Example: A business with FAQPage schema on its website, but the answers are loaded by JavaScript rather than rendered in the static HTML (meaning AI crawlers that do not execute JavaScript cannot read them) scores Partial on FAQ crawlability.
• Missing: The signal is absent entirely. Example: A business with no Foursquare listing scores Missing on Foursquare, and as a result has no local entity anchor on the highest-volume AI answer engine for local queries regardless of how strong everything else is.

What happens next

What we do with the results

Every finding lands in one of 3 buckets based on who needs to act on it.

• Engineering tasks: Changes to the website's technical structure: adding or correcting schema markup, moving schema from JavaScript-rendered to static HTML, implementing canonical tags, completing the sitemap, fixing duplicate headings. One developer, one sprint, measurable impact.
• Content tasks: Changes to what the pages say: rewriting city-page opening sentences to direct-answer format, adding city-specific FAQ questions with static HTML answers, adding pricing context to service pages, strengthening the founder and credential narrative. One writer, one brief, permanent signal improvement.
• Off-site claims: Actions taken outside the website: claiming and completing Foursquare, claiming Bing Places, auditing NAP consistency across directories, building or correcting the Google Business Profile, managing review solicitation. These are the platform actions no amount of website work can substitute for. They are also the fastest wins. A Foursquare claim takes an hour and changes local AI eligibility immediately.

We deliver a scored baseline, the full 24-dimension scorecard with evidence, and a prioritized fix list with owner, effort, and projected score impact for every item. You know what to fix, who fixes it, and what it will move.

The process

How an audit works, step by step

1. Define the query set: We map the service categories and geographic combinations your customers use when asking AI engines for your type of business.
2. Run queries across all 7 engines: Each query runs live at audit time. We capture the actual AI answer, not a simulated or cached result.
3. Record raw AI answers: We document what each engine returned: who was cited, what information was shown, what source was referenced, and whether your business appeared.
4. Score across 24 dimensions: Each dimension gets a Covered / Partial / Missing rating with the evidence noted. Platform presence, technical signals, content signals, and trust signals all scored.
5. Assign gaps to owner buckets: Every finding is classified: engineering task, content task, or off-site claim. Each bucket has a clear owner and an effort estimate.
6. Deliver scored baseline and priority fix list: You receive the full scorecard, the prioritized fix list with projected score impact per item, and the verbatim AI answer transcripts. Nothing is summarized away.

Common questions

Methodology questions answered