The way businesses are discovered has changed in ways that most agencies have not yet caught up to.
For decades, the path to visibility followed a familiar pattern: rank on Google, appear in search results, earn a click. The agencies built to serve that model focused on keywords, backlinks, and technical SEO fundamentals. Those fundamentals still matter. But they no longer tell the full story.
AI search systems — ChatGPT, Google AI Overviews, Perplexity, Claude, Gemini — do not work like traditional search engines. They do not return a list of ranked links and leave the user to decide. They synthesize information, form a conclusion, and deliver a direct answer. Sometimes that answer includes a business recommendation. Often it does not.
Whether your business is included in that recommendation depends on something more fundamental than keyword density or domain authority. It depends on whether AI systems can identify your business, understand what you do, trust the information they have found, and judge you as credible enough to recommend.
This is AI Visibility — and it requires a different set of competencies than traditional SEO. Research Firefly has conducted across hundreds of business audits confirms that most agencies are not yet equipped to address it.
The questions in this article are ones Firefly Web Labs hears regularly from prospective clients. They are also the questions every business should ask before choosing a partner to manage their digital presence in an AI-driven environment. Some probe technical depth. Others test strategic thinking. All are designed to separate agencies that have genuinely adapted to the current landscape from those still operating on 2019 assumptions.
Ask these questions. Listen carefully to the answers. What you learn will tell you more than any sales presentation.
1. Do you include schema markup implementation and llms.txt optimization in your technical audits?
This question is a reliable early signal. The answer will tell you whether an agency operates at the level of surface optimization — title tags, meta descriptions, page speed — or whether they understand how machines actually read and interpret a website.
Schema markup, implemented as JSON-LD in the page's <head>, is the primary mechanism by which a business communicates structured, machine-readable information to AI systems and search engines. A well-implemented schema layer tells AI what your organization is, what services you offer, where you are located, who you serve, and how those elements relate to each other.
The most important schema types for most businesses include:
- Organization — establishes your business as a recognized entity with a defined identity
- LocalBusiness — communicates location, hours, service area, and contact information
- Service — describes individual offerings clearly enough for AI to categorize and retrieve them
- FAQPage — surfaces common questions and answers directly into AI-generated responses
- Article — signals content type, authorship, and publication context for editorial content
- BreadcrumbList — communicates site architecture and content hierarchy
Beyond schema, a serious technical audit must evaluate entity relationships — the connections between your organization, its people, its services, its locations, and its industry context. Weak or missing entity signals are one of the most common reasons businesses are invisible in AI-generated answers despite strong keyword rankings. Firefly's Structured Data and AI Discovery study documented this pattern consistently across audited sites.
llms.txt is a newer, still-evolving standard that allows websites to provide AI systems with a structured summary of their content and intent. While not yet universally adopted by AI platforms, its inclusion represents a forward-looking technical posture worth having. Similarly, robots.txt configuration — which controls what AI crawlers can access — must be reviewed to ensure you are not inadvertently blocking the systems you want to be discovered by.
Crawlability matters as much for AI systems as it does for traditional search bots. If a page cannot be reached, indexed, or parsed, it cannot contribute to your visibility. An agency that cannot speak fluently to all of these elements is not prepared to optimize for the environment that currently governs AI discovery.
2. How do you track our citation share across ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews?
AI citation measurement is one of the most significant unsolved challenges in digital marketing today. It is also one of the most important. Any agency claiming to improve your AI visibility must have a coherent answer to this question.
Citation share refers to how frequently your business is mentioned, referenced, or recommended when users query AI systems on topics relevant to your industry, location, or services. It is distinct from traditional search rankings because it operates across multiple platforms simultaneously, each with different underlying models, training data, and retrieval architectures.
A credible measurement approach includes several components:
Baseline establishment. Before any optimization begins, an agency should query each major AI platform with representative questions — the kind your prospective customers are asking — and document which businesses are mentioned. This creates a measurable starting point.
Competitor comparison. Citation share is a relative metric. Knowing you appear in 20% of relevant AI responses means little unless you also know competitors appear in 60%. Competitive benchmarking gives context to the data.
Platform-specific tracking. Different AI systems behave differently. Google AI Overviews tend to prioritize websites with strong traditional SEO signals. Perplexity emphasizes recently published, source-cited content. ChatGPT and Claude draw from training data and real-time web retrieval depending on the query. Our Inside AI Overviews, Inside ChatGPT, and Inside Perplexity research series documents exactly how each platform retrieves and weights information. Each requires distinct visibility strategies, and each should be tracked separately.
Recommendation share. Beyond being cited as a source, recommendation share measures how often AI systems actively suggest your business when a user asks for a vendor, provider, or service recommendation. This is the metric with the most direct commercial value.
Trend reporting. Measurement only creates value when tracked over time. Visibility improvements after technical or content changes should be reflected in citation data within a measurable window.
No measurement framework is perfect. AI systems are not fully transparent about their retrieval logic. But an agency that cannot describe a structured approach to this question is essentially asking you to invest in visibility improvements without a way to verify whether they are working.
3. What is your strategy for digital PR and earning off-site authority?
AI systems do not form opinions about businesses in isolation. They synthesize what has been written, published, linked to, and referenced across the web. If your business exists primarily on your own website and nowhere else, AI systems have only one source to evaluate — and single-source corroboration is a weak foundation for trust.
Digital PR is the mechanism by which a business builds its presence across the broader information landscape. It is not about traditional media placements for brand awareness. In the context of AI visibility, it is about creating a web of corroborating signals that allow AI systems to recognize your business as a real, credible, and reputable entity.
The channels and tactics that matter most include:
Industry directories and associations. Listings in established industry directories, professional associations, and trade organizations create corroborating entity signals that reinforce your business's identity and category.
Chambers of commerce and local organizations. For businesses with a geographic service area, memberships and citations from local business organizations carry particular weight for location-based AI recommendations. Firefly's Local Authority Signals in AI Search study documents how geographic corroboration directly influences recommendation frequency.
Media mentions and editorial coverage. When reputable publications reference your business, they signal to AI systems that independent sources consider your business worth discussing. The quality of the publication matters significantly more than volume.
Podcasts and expert commentary. Appearances as a subject-matter expert — on podcasts, in webinars, in industry roundtables — build topical authority signals that AI systems recognize as indicators of genuine expertise.
Guest articles and contributed content. Publishing analysis or expert commentary in established industry publications does double work: it builds topical authority and creates backlinks that reinforce domain trust.
Resource pages and curated lists. When other businesses, publications, or organizations list your company as a recommended resource, those citations function as third-party endorsements that AI systems factor into trust assessments.
The underlying principle is corroboration. An agency should be able to explain not just what off-site tactics they use, but why each one contributes to the trust architecture that AI systems rely on. Tactics without strategic rationale are activity without direction.
4. How is AI Visibility different from traditional SEO?
Traditional SEO was built around a specific transaction: a user enters a query, an algorithm ranks pages by relevance and authority, the user receives a list of links. Success was measured by rank position and organic traffic.
AI Visibility operates on a fundamentally different model. AI systems do not return ranked lists. They generate synthesized answers, drawing from multiple sources, applying probabilistic judgment about what is accurate and trustworthy. The user receives a conclusion, not a menu. Firefly's research into AI Visibility vs. Traditional SEO quantifies the growing divergence between the two systems.
This changes what optimization means at a structural level.
From ranking to recognition. In traditional SEO, you compete for a position. In AI search, you compete for recognition — specifically, recognition as an entity that AI systems understand well enough to describe, categorize, and, eventually, recommend. A business that AI cannot identify clearly will not appear in AI-generated answers regardless of how well its pages rank.
From keywords to entities. Traditional SEO optimized for the presence of specific keywords. AI systems operate on semantic search principles, interpreting meaning rather than matching strings. This requires that businesses be defined as clear entities with consistent attributes — not just pages with keyword-rich text.
From traffic to trust. Traditional SEO success was measured by traffic volume. AI Visibility success is measured by trust — whether AI systems treat your business as a credible source worth citing and a credible provider worth recommending. These are different goals that require different methods.
From campaigns to infrastructure. SEO campaigns could be run and paused. AI visibility is built through durable infrastructure — structured data, consistent entity signals, corroborated authority — that compounds over time and degrades when neglected.
This distinction matters practically because agencies that repackage traditional SEO as "AI optimization" are solving the wrong problem. The question is not whether your pages rank. The question is whether AI systems recognize, understand, and trust your business.
5. What is GEO and why does it matter?
Generative Engine Optimization (GEO) refers to the practice of optimizing content and digital infrastructure so that AI systems that generate answers — rather than simply returning links — are more likely to include, cite, or recommend a business.
The term "generative engine" distinguishes AI answer platforms like ChatGPT, Perplexity, Claude, and Google AI Overviews from traditional keyword-matching search engines. These systems do not retrieve and rank; they synthesize and generate. The optimization strategies appropriate for each are meaningfully different.
GEO matters because generative engines are rapidly absorbing the queries that once drove organic search traffic. Firefly's Great Click Collapse research documents how zero-click search behavior is accelerating — users are getting answers directly from AI systems without visiting websites at all. A growing percentage of users are turning to AI systems for product recommendations, service provider identification, and local business discovery: the exact queries where appearing in an AI-generated answer creates direct commercial value.
A GEO strategy typically addresses several layers:
Content structure. AI systems prefer content that is clearly organized, answers specific questions directly, and demonstrates a point of view. Thin, hedged, or keyword-stuffed content performs poorly in AI retrieval.
Topical depth. Generative engines reward topical authority — the signal that a source covers a subject comprehensively and consistently, rather than touching it superficially. A business with ten well-developed service pages, a coherent knowledge base, and consistent publishing history signals more depth than one with a homepage and a contact form.
Entity clarity. AI systems must be able to identify what your business is, what it does, who it serves, and why it is credible before they will include it in a generated answer. GEO builds the entity foundation that makes that identification possible.
Citation-worthiness. Generative engines favor sources that other credible sources reference. Digital PR, editorial coverage, and industry citations all contribute to a business's citation-worthiness in AI-generated responses.
6. What is AEO and how does it relate to GEO?
Answer Engine Optimization (AEO) focuses specifically on optimizing content to appear as the direct answer to a user's question — rather than as a link the user must click to find the answer.
AEO predates the current AI search moment. Its early expression appeared in featured snippets, knowledge panels, and voice search answers in traditional Google results. The discipline asked: if a user asks a specific question, how do we structure our content so that Google selects our answer?
That same discipline now extends across every AI platform. When a user asks ChatGPT "Who is the best accountant in Phoenix for small businesses?" the answer that appears is the result of a retrieval and synthesis process that AEO can meaningfully influence.
The relationship between AEO and GEO is complementary rather than redundant:
AEO focuses on the question-answer unit. It asks: for specific questions users are asking, is our content structured to be selected as the authoritative answer? This involves FAQ schema, clear question-and-answer formatting, and direct, specific language.
GEO focuses on the generative system's broader trust model. It asks: does the totality of our digital presence — content depth, entity signals, corroboration, structured data — position us as a source a generative AI should draw from?
A business that invests only in AEO may structure its FAQ content well but lack the entity depth and topical authority that AI systems use to select which sources to trust. A business that focuses only on GEO may have strong infrastructure but poorly structured content that is difficult for AI to retrieve in response to specific queries.
The most effective AI visibility strategies address both layers simultaneously. Firefly's AI Visibility Optimization practice is built on this integrated model.
7. Can AI recommend businesses that are not ranked #1?
Yes — and this is one of the most practically important facts for businesses to understand about the current landscape.
Traditional search results are ordered. Position one gets the most clicks; position ten gets a fraction of that. The entire logic of SEO was built around competing for the highest rank.
AI systems do not operate on this logic. When a generative engine recommends a business, it is not translating a ranked list into prose. It is synthesizing a judgment based on the totality of signals it has processed: entity clarity, corroborated reputation, demonstrated expertise, content depth, and trust indicators. Firefly's research on why rankings no longer mean reach documents this decoupling in detail.
A business that ranks seventh organically but has exceptionally clear entity signals, consistent off-site citations, strong review sentiment, and well-structured service content may be recommended by AI systems far more frequently than the business occupying position one.
This has significant strategic implications:
AI visibility and search ranking are separate competitions. A business can improve its AI recommendation frequency without necessarily moving its organic rankings, and vice versa.
Newer businesses can compete. A well-established competitor with strong domain authority may outrank a newer business on every keyword. But if that newer business has invested in entity optimization, structured data, and off-site authority signals, it may appear in AI-generated recommendations more consistently.
Recommendation quality matters more than recommendation frequency. AI systems recommend businesses in context. A business that appears in AI responses to highly specific, high-intent queries is generating more commercial value than one that appears in broad, low-intent answers.
The implication for agency selection is direct: an agency focused exclusively on improving your organic rankings is optimizing for a metric that does not govern AI recommendations. These are related but distinct systems, and they require distinct strategies.
8. How do AI systems decide which businesses to trust?
Trust, in the context of AI systems, is not a single signal. It is a composite judgment assembled from dozens of data points across multiple dimensions. Understanding how that judgment is formed is foundational to building an effective AI visibility strategy.
Entity recognition. Before trust can be evaluated, identity must be established. AI systems use entity recognition to determine whether your business is a known, defined entity in their model of the world. Businesses with inconsistent NAP data (name, address, phone), conflicting descriptions across platforms, or minimal third-party presence are effectively unknown entities to many AI systems. Firefly's Entity Clarity and Visibility Performance study quantifies how ambiguity directly costs businesses recommendation opportunities.
Consistency. AI systems flag inconsistency as a reliability risk. If your business is described differently on your website, your Google Business Profile, your LinkedIn page, and third-party directories, the AI cannot determine which description is authoritative. Consistency across all owned and cited properties signals reliability.
Corroboration. No single source — including your own website — is sufficient for trust. AI systems evaluate whether independent, reputable sources confirm what you claim about your business. This is why off-site authority matters: it provides the corroboration that elevates a claim from self-assertion to verified fact.
Reputation signals. Reviews, ratings, and reputation data across Google, Yelp, industry-specific platforms, and social channels contribute to AI trust assessments. Consistent positive sentiment from real customers is a meaningful signal. Unaddressed negative reviews or a thin review profile raise uncertainty.
Demonstrated expertise. E-E-A-T — Google's framework for evaluating Experience, Expertise, Authoritativeness, and Trustworthiness — maps closely to how many AI systems evaluate content and source credibility. Businesses whose content demonstrates direct subject-matter expertise, and whose authors or principals have verifiable professional credentials, are weighted more heavily.
Authority signals. Authority signals — backlinks, mentions in reputable publications, citations from industry organizations, appearances in authoritative directories — all contribute to the authority dimension of AI trust assessment. Authority is not the same as trust, but it is a prerequisite for it.
The Firefly approach structures these dimensions as four interdependent signal layers: Recognition, Corroboration, Trust, and Expertise. A business cannot shortcut through these layers. Each one depends on the preceding one.
9. What are entity signals?
An entity, in the context of AI and semantic search, is a uniquely identifiable thing — a business, a person, a place, a concept — that can be distinguished from all other things. Entity signals are the data points that help AI systems correctly identify, categorize, and understand what that entity is.
For a business, entity signals include:
Core identity signals. Business name, address, phone number, website URL, and business category — consistently applied across all properties. These are the minimum foundation; without consistency here, all other signals are weakened.
Descriptive signals. How your business is described — the language used to characterize your services, your expertise, your audience, and your differentiation — must be semantically consistent even if not identical across properties. AI systems use natural language processing to extract meaning, and contradictory descriptions create interpretive uncertainty.
Relational signals. The connections between your business and other recognized entities — your industry category, your location, the associations you belong to, the platforms where you are listed — all contribute to a richer entity profile that AI systems can reason about. The concept of a digital entity footprint captures this well: the total shape of your business's presence across the web.
Knowledge Graph presence. Google's Knowledge Graph is one of the most important entity databases in the world. Businesses that achieve Knowledge Panel status — earned through consistent signals, strong corroboration, and sufficient public interest — have a measurable advantage in AI visibility. This is not a status most businesses can acquire quickly, but it is one a competent agency should be working toward.
Structured data signals. Schema markup, when implemented correctly, provides AI systems with explicit, machine-readable entity definitions. It is the clearest possible signal: rather than asking AI to infer what your business is, you are stating it directly in a format designed to be parsed and retained.
Entity signals are not a one-time optimization. They require ongoing maintenance. Business information changes; competitors introduce noise into the information landscape; new platforms emerge. An agency serious about AI visibility treats entity signal maintenance as continuous, not periodic.
10. How important is structured data for AI search?
Structured data is the most direct communication channel a business has with the machines that govern AI discovery. It is not the most important factor in AI visibility, but it is the most controllable one — and agencies that treat it as optional are leaving a significant advantage on the table. Firefly's Schema & Structured Data practice exists precisely because most businesses have never had schema implemented correctly.
In practice, structured data does several important things for AI visibility:
It disambiguates identity. AI systems encounter thousands of businesses with similar names and similar services. Schema markup provides explicit identifiers — URLs, address data, industry classifications — that allow AI to distinguish your business from similar ones unambiguously.
It communicates relationships. Advanced schema implementation can describe the relationships between entities on your site: the organization that employs these people, offers these services, is located at this address, and holds these credentials. That relational web is more informative to AI systems than any individual piece of data.
It surfaces FAQ content. FAQPage schema is one of the most commercially valuable schema types for AI visibility. AI systems — including those powering Google AI Overviews and Perplexity — regularly surface FAQ schema content in direct answers to user questions. A well-implemented FAQ schema layer means your content is formatted for AI retrieval in a way that prose alone cannot match.
It signals content type and context. Article schema, with proper authorship markup, tells AI systems that your content is editorial, attributed, and dated — signals associated with credible, fact-checkable content rather than promotional material. This connects directly to the E-E-A-T signals that AI systems evaluate.
The absence of structured data does not make a business invisible to AI. Many AI systems can infer information from well-written prose. But inferred information is less reliable than declared information, and AI systems operating under uncertainty will favor sources that remove that uncertainty. Structured data removes it.
11. What technical problems prevent businesses from being recommended?
Many businesses are technically invisible to AI systems not because their content is poor, but because their technical infrastructure creates friction that AI cannot navigate. Firefly has documented this consistently across studies of businesses that disappear from AI recommendations: the root cause is almost always technical, not editorial. An audit that does not identify and resolve these problems is not a complete audit.
The most common technical barriers to AI recommendation include:
Duplicate content. When the same or substantially similar content appears at multiple URLs — a problem common in e-commerce platforms, multi-location service businesses, and poorly managed CMS implementations — AI systems cannot determine which version is authoritative. The use of canonical tags is the standard resolution, but it must be applied correctly.
Weak information architecture. Sites with no clear hierarchy, no consistent internal linking logic, and no discernible topical organization are difficult for AI crawlers to navigate and interpret. AI systems favor sources with clear structure, because structure is itself a signal of intentionality and organization.
Poor internal linking. Internal links do two things relevant to AI visibility: they distribute authority through the site and they communicate semantic relationships between content. A services page that is never linked from blog posts, case studies, or resource pages exists in isolation — disconnected from the topical context that would make it more retrievable.
Missing or broken schema. Schema markup that was implemented incorrectly, that references outdated information, or that fails validation testing provides no benefit and can actively introduce confusion. Regular schema audits are essential.
Inconsistent NAP data. Name, address, and phone number inconsistencies across your website, Google Business Profile, Yelp, Bing Places, and other platforms undermine entity recognition. Even minor variations — "St." versus "Street," a suite number present in one listing and absent in another — create uncertainty that AI systems treat cautiously.
Thin or underdeveloped service pages. A service page with 150 words, no supporting detail, and no evidence of expertise does not give AI systems enough signal to confidently categorize or recommend the associated service. Service pages should be treated as entity definitions: specific, comprehensive, and evidence-rich.
Crawl blocks and indexing errors. Robots.txt misconfigurations, noindex tags applied incorrectly, and canonical errors can make pages invisible to AI crawlers even when they are visible to human visitors. This is a technical problem that requires technical resolution — and it is often the first thing Firefly's Site Audit & Strategy process uncovers.
12. Does every business need a new website?
No. This is one of the most common misconceptions in the industry, and it is one Firefly actively works to correct.
Most businesses with AI visibility problems do not have a website problem. They have an information architecture problem, a messaging problem, a technical optimization problem, or an entity signal problem. A new website built on the same structural and strategic weaknesses as the old one will reproduce those problems at greater expense and greater disruption.
The relevant question is not whether your website is new. It is whether your website clearly communicates what your business is, who it serves, what makes it credible, and what services it offers — in a format that both human visitors and AI systems can interpret and trust.
An honest assessment might find that your current website needs:
- A restructured information hierarchy that makes content easier to navigate and categorize
- Stronger, more specific service pages that function as genuine entity definitions
- Schema markup implemented throughout critical pages
- Internal linking that connects topically related content
- Cleaner, more consistent brand messaging that reduces interpretive ambiguity for AI systems
- UX improvements that reduce friction for human visitors converting to clients
These are architectural, editorial, and technical improvements. They can often be applied to an existing website at a fraction of the cost of a full rebuild — and they address the actual problem. Firefly's case studies on Broken Infrastructure, Strong Visibility and Legacy Reputation vs. AI Discovery illustrate how significant improvements are achievable without rebuilding from scratch.
There are circumstances where a new website is the right recommendation: when the existing platform fundamentally limits technical implementation, when the brand has evolved past what the current site can represent, or when the UX is so compromised that no amount of optimization will yield results. But those are specific circumstances, not a default prescription.
An agency whose first recommendation is always a new website is optimizing for its own revenue, not your outcome.
13. How do you measure success?
This question separates agencies with disciplined performance frameworks from those managing to activity rather than outcomes. The metrics that matter for AI visibility are different from traditional SEO metrics, and any agency worth evaluating should be able to articulate them clearly.
A complete measurement framework for AI visibility includes:
AI citation frequency. How often your business appears in AI-generated responses to relevant queries — measured across platforms, tracked over time, and compared against a baseline and against competitors.
Recommendation share. Of the queries that produce a specific business recommendation (as opposed to general information), how frequently is your business the recommendation? This is the metric with the most direct relationship to commercial outcomes.
Search intent alignment. Are you appearing in response to high-intent queries — the ones where a user is evaluating providers — or only in response to informational queries with no commercial value? Volume without intent alignment is a vanity metric.
Technical health scores. Schema validation rates, crawl error rates, Core Web Vitals, mobile usability, and structured data implementation completeness. These are inputs to visibility, not outputs, but they must be monitored because degradation in technical health is the most common cause of visibility loss.
Entity signal consistency. Periodic audits of NAP consistency, Knowledge Graph status, and entity representation across third-party platforms. Entity signals degrade over time without maintenance.
Organic visibility trends. Traditional search visibility remains relevant as a correlated signal. Improvements in AI visibility often produce correlated improvements in organic rankings, and vice versa.
Conversion quality. Ultimately, the purpose of visibility is to generate qualified business. Tracking how AI-referred traffic converts — and at what rate — ties the visibility investment to business outcomes.
If an agency proposes measuring success exclusively by keyword rankings and organic traffic, they are applying a legacy measurement framework to a problem that has moved beyond it.
14. How long does AI visibility improvement usually take?
This question deserves an honest answer rather than an optimistic one.
AI visibility is not a campaign with a fixed end date. It is a compounding infrastructure investment. Some improvements produce results quickly; others take months to register across AI systems.
Technical changes produce the fastest results. Resolving crawl blocks, fixing broken schema, correcting NAP inconsistencies, and improving site architecture create conditions for better AI discovery. AI systems that regularly re-index the web may reflect these improvements within weeks.
Content improvements take longer. Publishing comprehensive service pages, developing a topic cluster structure, and building out FAQ content requires time both to produce and to be discovered, indexed, and weighted by AI systems. Expect three to six months before content investments produce measurable citation improvements.
Authority building is the longest investment. Digital PR, off-site citations, media coverage, and industry recognition build gradually. Individual placements may appear quickly, but the cumulative authority effect takes six to twelve months to register meaningfully in AI trust assessments.
Competitive context matters. A business entering a low-competition local market may see measurable AI visibility improvements in 90 days. A business competing nationally in a crowded professional services category may require eighteen months of sustained investment before citation share reaches competitive parity.
The honest framing is this: AI visibility is a permanent strategic asset, not a one-time fix. The businesses that invest consistently over time accumulate compounding advantages. Those that treat it as a short-term initiative tend to achieve short-term results — and watch those results erode when investment pauses.
15. Can AI optimization replace SEO?
No. This framing presents a false choice.
Traditional SEO remains essential for two reasons: first, search engines continue to drive significant traffic for most businesses; second, the signals that traditional SEO builds — crawlability, content authority, backlink profiles, technical health — are inputs into many AI visibility systems.
Google AI Overviews, for example, are generated from the same index that powers organic search. A page that cannot be crawled and indexed will not appear in AI Overviews, regardless of how well its schema is implemented. Perplexity draws substantially from web content that has been indexed and assessed for credibility by traditional search infrastructure. Our Inside AI Overviews research documents exactly how Google's retrieval pipeline connects traditional indexing to AI answer generation.
What has changed is the hierarchy of priorities and the completeness of the discipline required.
Traditional SEO, practiced in isolation, is increasingly insufficient. A business can rank on page one for every target keyword and still be absent from AI-generated recommendations if its entity signals are weak, its structured data is missing, and its off-site authority profile is thin. Firefly's SEO & Discovery Strategy practice is built on this integrated model.
The agencies best positioned to serve businesses in the current environment are those that practice AI visibility as an integrated discipline — one that includes traditional SEO fundamentals alongside entity optimization, structured data strategy, GEO, AEO, and authority building. These are not competing priorities. They are complementary layers of a complete digital visibility strategy.
16. How do reviews influence AI recommendations?
Reviews are a trust signal that AI systems treat with measurable weight, particularly for local and service businesses.
AI systems encounter reviews not just as sentiment data but as evidence of real-world performance. A business with 400 reviews averaging 4.8 stars across multiple platforms is providing AI systems with corroborating evidence of consistent customer satisfaction. That evidence contributes to the trust assessment that governs recommendation decisions.
Several dimensions of review data matter:
Volume. A thin review profile — even if uniformly positive — signals limited market presence. AI systems use review volume as a proxy for how established and widely experienced a business is.
Recency. Reviews from three years ago do not carry the same weight as reviews from the past ninety days. Active review acquisition is an ongoing operational discipline, not a one-time task.
Platform diversity. Reviews concentrated on a single platform are more susceptible to appearing manipulated. Consistent sentiment across Google Business Profile, Yelp, industry-specific platforms, Facebook, and others provides stronger corroboration than a single-platform profile.
Response behavior. How a business responds to reviews — particularly negative ones — is itself a trust signal. Thoughtful, professional responses to criticism signal that the business is accountable and customer-focused. Ignored or defensively handled reviews raise doubt.
Sentiment specificity. Reviews that describe specific experiences, mention specific services or staff members, and use language consistent with your business category contribute to entity signal strength. Generic praise has less AI signal value than specific, detailed feedback that confirms what your business actually does.
17. Does local SEO still matter?
Yes — significantly so, and more than many businesses currently recognize.
Local SEO creates the geographic and categorical data that AI systems use to generate location-based recommendations. When a user asks "who is the best physical therapist near downtown Dallas" or "top-rated Italian restaurant in Chicago's River North," the AI systems answering that question are drawing on exactly the signals that local SEO builds and maintains. Firefly's Local Authority Signals in AI Search study examined how geographic entity signals translate directly into recommendation frequency.
The components that matter most:
Google Business Profile. This remains the single highest-leverage local signal for most businesses. A complete, accurate, regularly updated Google Business Profile with consistent NAP data, professional photos, and active review management is a prerequisite for local AI visibility, not an enhancement.
Local citation consistency. Your business information across Yelp, Bing Places, Apple Maps, local directories, and industry platforms must be consistent with your Google Business Profile and your website. Inconsistency signals unreliability.
Geographic entity signals. Schema markup should explicitly communicate your service area, your city and region associations, and your physical location. These signals help AI systems categorize you accurately for location-based queries. The concept of local entity authority captures how this geographic signal layer operates.
Localized content. Service pages, blog content, and case studies that reference the geographic context of your work — the markets you serve, the communities you operate in, the regional issues relevant to your clients — create topical authority signals that AI systems use for location-based recommendations.
For multi-location businesses, these principles must be applied consistently across every location. Each location should function as a distinct entity with its own fully optimized digital presence, rather than being treated as a variation of the parent brand.
18. Should every business have an llms.txt file?
llms.txt is a proposed standard — modeled loosely on robots.txt — that allows websites to provide AI large language models with a structured, human-readable summary of their site's content and intent. The concept was introduced in 2024 and has gained interest in the AI optimization community.
The honest answer is: it depends, and the impact is currently limited.
llms.txt is not yet natively supported by all major AI platforms. Whether a given AI system will read and weight an llms.txt file depends on that system's crawling and retrieval architecture — and most have not published documentation confirming that they use it in any determinative way.
That said, there are reasonable arguments for implementing it:
It represents a forward-looking technical posture. If the standard gains adoption, businesses with llms.txt files already in place will benefit from early-mover advantage. The implementation cost is low.
It provides AI systems with structured, authoritative self-description. Even in cases where llms.txt is not determinatively weighted, providing a clear, curated summary of your site's content and purpose in a machine-readable format is consistent with good AI-readiness practice.
It signals intent. A well-crafted llms.txt file demonstrates that your business understands and is engaging seriously with the AI visibility landscape. That may have indirect value in establishing credibility as the standard matures.
However: implementing llms.txt should not be prioritized over foundational work. Schema markup, entity consistency, structured content, and off-site authority building all have demonstrably greater impact on AI visibility today. llms.txt is a worthwhile addition to a complete technical stack, not a substitute for it.
Any agency that identifies llms.txt as a primary AI visibility strategy has its priorities inverted.
19. How important are branded searches?
Branded searches — queries that include your company name or the names of your principal team members — are among the most important signals in AI visibility, and they are routinely underinvested.
Here is why they matter specifically for AI systems:
AI systems use branded search volume as a proxy for relevance and recognition. When many people search for your business by name, AI systems interpret that as evidence that your business is known and valued. It functions as a demand signal: proof that your brand has market recognition, not just web presence. This connects directly to the concept of training data visibility — whether your brand appears in the sources that AI models learned from.
A strong branded presence reduces AI uncertainty. When an AI system encounters your business name in a context that suggests a recommendation, it draws on everything it knows about that entity. A business with a rich branded search history, a populated Knowledge Panel, consistent NAP data, and strong review sentiment gives AI systems a complete picture. A business with no branded search history creates interpretive uncertainty.
Your principals' authority matters. The people behind a business — founders, lead practitioners, subject-matter experts — can carry significant entity authority in their own right. A founder who has spoken at industry conferences, published articles in respected trade publications, and has a consistent professional profile across LinkedIn, industry directories, and media appearances contributes meaningful entity authority signals to the business entity.
Branded content should be actively managed. The information that appears when AI systems encounter your brand name — your website, Google Business Profile, news mentions, reviews, social profiles — must be consistent, current, and credible. Outdated information, negative coverage without corresponding positive signals, or simply a thin brand profile all reduce the confidence with which AI systems will recommend your business.
20. Why should we choose Firefly Web Labs?
We will answer this question the way we answer all of them: honestly, with evidence, and without the pressure tactics that characterize lesser pitches.
Firefly Web Labs was built specifically for the AI visibility era. We did not adapt an existing SEO practice to accommodate AI; we built our methodology from first principles around the question of how AI systems discover, evaluate, trust, and recommend businesses. Our published research series — including the Inside The Models program covering Claude, ChatGPT, Google AI Overviews, and Perplexity — reflects the depth of analysis that informs every client engagement. That distinction matters.
Our methodology follows a structured sequence:
Audit. We begin with a comprehensive technical and strategic audit that evaluates your current AI citation share, entity signal strength, structured data implementation, content architecture, and off-site authority profile. We document baselines before we propose solutions, because measurement without a baseline is speculation.
Strategy. Based on audit findings, we develop a prioritized roadmap. Every recommendation is tied to a specific AI visibility objective, ranked by expected impact and implementation effort. We do not present a menu of services; we present a causal chain from where you are to where you need to be. Engagement details and pricing are available for review.
Entity Optimization. We resolve entity signal inconsistencies, implement and validate schema markup, update and align your business information across all relevant platforms, and work to establish your business as a clearly defined, recognizable entity in the data systems that AI models rely on.
Technical Foundation. We address the technical barriers that prevent AI discovery: crawl issues, indexing problems, architecture weaknesses, internal linking gaps, and content quality deficits that reduce retrievability.
Authority Signals. We build the off-site corroboration network that supports AI trust: digital PR, industry citations, editorial placements, and directory presence across the platforms that AI systems treat as reliable references.
Measurement. We establish regular reporting cycles that track AI citation share, recommendation frequency, technical health, and entity signal strength — not as a performance theater, but as the actual feedback loop that governs our strategic decisions.
Continuous Improvement. AI systems evolve. Search behavior evolves. Competitive landscapes shift. We maintain your visibility infrastructure as an ongoing strategic asset, not a completed project.
We work with businesses that are serious about their digital future. We do not promise shortcuts. We do not inflate projections. We report what we find, recommend what we can justify, and measure everything we do.
That is what a partnership with a competent AI visibility agency should look like. Review our AI Visibility FAQ and AI Visibility Hub for further background on our approach.
The agencies you evaluate this year are making a choice that will compound for years.
Choosing a partner who understands traditional SEO but not AI visibility is not a neutral decision. It is a decision to optimize for a system that is gradually becoming less relevant while the system that will govern future discovery goes unaddressed.
The questions in this article are not designed to be easy. They are designed to surface the difference between agencies that have adapted their thinking to a genuinely new environment and those that have adapted their language while keeping their old playbooks.
Do not settle for a conversation about keyword rankings and backlinks when the real question is whether AI systems recognize, understand, and trust your business enough to recommend it.
The future of discovery is built on clarity, corroboration, and credibility — not on gaming algorithms that are no longer the ones making the decisions.
Ready to understand exactly where your business stands? Schedule your AI Visibility Audit with Firefly Web Labs. We will tell you what we find, not what you want to hear.
