--- title: "An AI Visibility Audit Template Agencies Can Reuse" slug: "ai-visibility-audit-template-for-agencies" category: "agencies" canonical_path: "/articles/agencies/ai-visibility-audit-template-for-agencies" meta_title: "AI Visibility Audit Template for Agencies — Prime" meta_description: "A reusable AI visibility audit template for agencies: discovery questions, a 20-prompt starter structure, an effort/impact scorecard, and how to scope a pilot without overstating certainty." author: "Bob Generale" reviewer: "Alex Mannine" date: "2026-08-06" last_updated: "2026-08-06" read_time: "14 min" keywords: - AI visibility audit template for agencies - pre-sale audit - prompt starter structure - effort impact scorecard - pilot scope featured_image: "/brand/articles/agencies/ai-visibility-audit-template-for-agencies.png" featured_image_alt: "A grid of eight small amber squares feeding by thin lines into one larger hollow graphite square across a slim divider" og_image: "/brand/articles/agencies/ai-visibility-audit-template-for-agencies.og.png" cta_mid_headline: "Run the audit against real engine answers, not guesses" cta_mid_body: "Prime AI Visibility takes the prompt set from this template and runs it across ChatGPT, Perplexity, Gemini, and Claude, so your pitch shows the prospect exactly what the engines say about them today." cta_mid_button: "Run the agency audit template" cta_bottom_headline: "Turn a template into a scoped pilot" cta_bottom_body: "Use the effort/impact scorecard to propose a pilot the prospect can say yes to — a fixed prompt set, a baseline, and a short list of owned quick wins." cta_bottom_button: "Start the audit" --- # An AI Visibility Audit Template Agencies Can Reuse An AI visibility audit template for agencies is a reusable pre-sale and onboarding instrument that turns a prospect's curiosity into a scoped engagement: a set of discovery questions, a fixed sample of buyer prompts run across the engines, a competitor and source gap read, a factual-risk check, an effort/impact scorecard, and a recommended pilot. It shows a prospect what the engines say about them today — without promising a ranking. ## AI visibility audit template for agencies: the short answer 1. **Diagnose before you propose.** The audit's job is to make a prospect's current AI answers visible, so your recommendation is evidence-led rather than a pitch deck of assumptions. 2. **Prioritize with a formula, not a feeling.** Rank each finding by effort and impact using word ratings, so the quick-wins list is defensible and the pilot scope is obvious. 3. **Scope honestly.** Show what the engines do now and what work would follow, and never overstate certainty — no engine publishes its selection logic. ## Who this is for This is for agency owners and new-business leads who want to open (or win) a conversation about AI visibility with a prospect or a newly signed client. It is deliberately narrow: a pre-sale and onboarding audit, not the ongoing delivery loop (that lives in [the agency delivery operating system](https://primeaivisibility.com/articles/agencies/how-agencies-boost-client-ai-visibility)) and not the monthly reporting cadence (covered in [defensible client AI visibility reporting](https://primeaivisibility.com/articles/agencies/agency-client-ai-visibility-reporting)). ## What an agency audit must include A pre-sale audit has one job: replace assumptions with observations so both sides can scope real work. To do that credibly it must contain six things, and skipping any one weakens the pitch: - **A prospect-specific prompt sample** — real buyer questions, not head keywords, so the findings are about their buyers. - **Observed engine answers** — what ChatGPT, Perplexity, Gemini, and Claude actually return today, captured with conditions. - **A competitor and source gap read** — who is named instead, and which sources the engines lean on. - **A factual-risk flag** — any answer that describes the prospect inaccurately, prioritized by potential harm. - **A prioritized quick-wins list** — findings ranked by effort and impact, so the first month is obvious. - **A recommended scope** — software-only, agency-led, or hybrid, with an honest boundary on what each can claim. ## The Pitch-to-Proof Audit The Pitch-to-Proof Audit is an eight-module framework designed to move a prospect from "I wonder how we show up" to "here is a scoped pilot I can approve." Run the modules in order; each feeds the next. 1. **Commercial questions.** What outcome does the prospect actually want — demos, shortlisting, local walk-ins? The audit is prioritized against this, not against a generic visibility ideal. 2. **Prompt sample.** Build a fixed sample of buyer prompts from the prospect's funnel (see the starter structure below). Freeze it so the audit is reproducible. 3. **Competitor and source gap.** Run the sample, record who is named instead of the prospect, and note which sources the engines cite — the gap and its causes in one pass. 4. **Factual risk.** Flag any answer that states something wrong about the prospect. A factual error that could mislead a buyer is a higher priority than a missing mention. 5. **Quick wins.** Identify the fixes that are fast, safe, and high-impact — a wrong location, a missing category page, a broken canonical the engines are tripping over. 6. **Required owners.** Name who would do each fix: content, technical, entity, or reputation. This is where "agency-led" versus "client-led" scope becomes concrete. 7. **Effort/impact score.** Rate every finding on effort and impact using word ratings, and sort. The top-left quadrant — low effort, high impact — is your pilot. 8. **Pilot scope.** Propose a bounded first engagement: a frozen prompt set, a baseline, and the quick-wins list, with a clear boundary on what a pilot can and cannot prove. The framework's discipline is that a recommendation only appears after an observation. That is what separates a Pitch-to-Proof Audit from a template full of assumptions, and it is why the underlying [step-by-step AI visibility audit method](https://primeaivisibility.com/articles/ai-visibility/how-to-run-an-ai-visibility-audit) is worth reading alongside it. ## Discovery questions for the client Before running a single prompt, ask the prospect these. Their answers shape the prompt sample and keep the audit aimed at a commercial outcome. - What decision do your best customers make right before they buy, and what do they ask while making it? - Which competitors do you lose to, and where do you think buyers first hear about them? - Which of your facts must be exactly right in any answer — locations, certifications, pricing model, service area? - Which engines do your buyers actually use, as far as you can tell? - What would a win look like in 90 days, stated as a business outcome rather than a metric? - Who on your side can approve content, provide facts, and respond to reviews? The last question matters more than it looks: it determines whether the engagement is agency-led, client-led, or hybrid, and therefore what you can honestly promise. ## A 20-prompt starter structure by intent Below is a **starter structure**, organized by intent — not a universal benchmark. It is a shape to fill with the prospect's real category, competitors, and locations; the numbers are slots, not a score. | # | Intent theme | Prompt shape | |---|---|---| | 1–4 | Category education | "what is [category] and how does it work"; "how do I choose a [category] provider" | | 5–9 | Comparison | "best [category] for [segment]"; "top options for [use case]"; "[prospect] vs [competitor]" | | 10–13 | Use case | "how do I solve [buyer problem] with [category]" | | 14–16 | Objection | "is [category] worth it for [segment]"; "downsides of [approach]" | | 17–18 | Local (if relevant) | "best [category] near [city]"; "[category] in [region]" | | 19–20 | Brand direct | "is [prospect] any good"; "who is [prospect] and what do they do" | Fill each slot with the prospect's specifics, freeze the set, and run it. The value is in the intent coverage — spotting whether the prospect is absent across an entire stage of the journey, not whether they missed a single question. A weak, keyword-shaped prompt set produces a misleading audit no matter how carefully you run it, which is why the discovery step comes first. Regulated and vertical prospects need extra prompt families and risk handling. If you are auditing a healthcare client, adapt the sample using the [healthcare AI visibility audit method](https://primeaivisibility.com/articles/healthcare/healthcare-ai-visibility-audit), which adds audience separation and privacy-safe logging; for a retail or DTC client, the [e-commerce AI visibility audit](https://primeaivisibility.com/articles/ai-visibility/ecommerce-ai-visibility-audit) adds catalog, availability, and offer prompts. ## Audit scorecard and action-priority formula Score every finding with word ratings, never invented numbers. Two axes drive the priority. | Finding | Effort | Impact | Owner | Priority | |---|---|---|---|---| | Wrong location in answers | low | high | technical/entity | do first | | No mention on comparison prompts | high | high | content/reputation | pilot core | | Mentioned but not recommended | medium | high | content | pilot core | | Missing owned citation | medium | medium | content | schedule | | Outdated fact on one engine | low | medium | content | quick win | **Action-priority rule.** Sort by impact first, then by effort within each impact band. Low-effort / high-impact findings become the quick-wins list you can ship in the first cycle; high-effort / high-impact findings define the pilot's core work; everything else is scheduled. This keeps the proposal honest: the prospect sees exactly why each item sits where it does. For a shared definition of the terms you rate — mention, recommendation, citation — point the client to the [Prime AI Visibility glossary](https://primeaivisibility.com/glossary) so the scorecard reads the same way to everyone. ## How to scope software-only, agency-led, and hybrid engagements The audit's owner column tells you which engagement shape fits. Name the boundary plainly for each. - **Software-only.** The prospect has an internal team that can execute. Prime AI Visibility provides the measurement, baseline, and gap classification; the client's team does the fixing. Best when the constraint is *visibility*, not *capacity*. - **Agency-led.** The prospect wants the work done. Your agency owns the delivery loop and reports against the baseline. Best when the constraint is capacity and the client trusts you to execute. - **Hybrid.** The prospect has some capacity but needs specialist help — technical remediation, digital PR, entity clarification. This is the most common shape, and it is where a managed partner earns its place. When a client needs the fixes performed at a level beyond in-house capacity, Percepture provides [white-glove AI-search remediation](https://percepture.com/services/geo-services/) that executes against the audit's findings. *Disclosure: Prime AI Visibility and Percepture have a commercial relationship. Prime provides visibility intelligence and diagnosis; Percepture provides managed implementation. Recommendations and comparisons use the criteria shown on this page.* The diagnosis-versus-execution split is the same one the public OPTK program illustrates: Prime AI Visibility identified the opening; Percepture's strategy, content, and technical execution pursued the result, and software alone did not cause it. At pitch stage, say it exactly that way — it sets honest expectations and makes the hybrid scope obvious. The wider version of this split, across the full stack, is mapped in the category piece on [an AI visibility engine, automation, and services](https://primeaivisibility.com/articles/automation/ai-visibility-engine-marketing-automation-services), and the procurement trade-offs are compared in [an AI visibility engine versus a marketing automation platform](https://primeaivisibility.com/articles/automation/ai-visibility-engine-vs-marketing-automation). ## What can be shown in a pitch without overstating certainty A pitch built on this audit is powerful because it shows real answers — but power invites over-claiming. Hold these lines: - **Show, don't promise.** "Here is what Perplexity says about you today" is a fact. "We will get you recommended by Perplexity" is a claim no one can honor. - **State the variability.** Present findings as a bounded snapshot, not a permanent truth. - **Separate the ladder.** Distinguish mentions, recommendations, and citations so the prospect understands which one the work would target. - **Name the boundary.** Google's own guidance is clear that strong fundamentals and genuinely useful content remain the basis for eligibility, and that there is no special AI file or schema shortcut [[1]](#references) — say so, rather than selling a silver bullet. > **Measurement variability.** AI answers can vary by platform, model or product, search state, location, prompt wording, time, and repeated run. Results describe a defined observation method, not a permanent universal rank. ## Printable audit worksheet Copy this worksheet into a document and complete one per prospect. It is plain text so it prints cleanly and travels between tools. - **Prospect:** ____________________ **Date:** ____________ **Auditor:** ____________ - **Commercial objective (90-day, business terms):** ____________________________________ - **Priority audiences / personas:** _____________________________________________________ - **Frozen prompt set (attach the 20-slot table, filled):** ______________________________ - **Engines run (name, mode, location, date):** __________________________________________ - **Findings ledger:** for each prompt — mention (y/n), recommendation (y/n), owned citation (y/n), competitor named, factual accuracy (ok / flagged), source URLs, screenshot reference. - **Factual-risk flags (ordered by potential harm):** ____________________________________ - **Quick wins (low effort / high impact):** _____________________________________________ - **Effort/impact scorecard (word ratings):** ____________________________________________ - **Recommended scope (software-only / agency-led / hybrid):** ___________________________ - **Proposed pilot (frozen set + baseline + quick-wins list):** ___________________________ - **Honesty note stated to prospect:** results are a bounded snapshot; no ranking or citation is promised. ## A note for agency business development Separate from the client visibility audit, agencies often need to *find* the accounts worth auditing. That is a business-development task, not an AI-visibility measurement — keep the two clearly apart in any proposal. When you are building a target list of accounts to pitch this audit to, [source-backed B2B prospect intelligence](https://theleadseeker.com/) can help you identify and qualify companies. It is not part of the client's AI-visibility audit and should never be presented as one; it lives in your own new-business pipeline. ## Methodology and sources This article describes the Pitch-to-Proof Audit and a reusable, printable worksheet for agencies. The 20-prompt table is a starter structure organized by intent, explicitly not a universal benchmark, and all scorecard values are word ratings rather than invented numbers. The OPTK example is used only within its documented public scope: Prime AI Visibility identified the opening, and Percepture's strategy, content, and technical execution pursued the result — software alone did not cause it. AI answers vary by platform, model, search state, location, prompt wording, time, and run, so audit findings are bounded snapshots, and no ranking or citation is promised. This article was authored by Bob Generale; the methodology was reviewed by Alex Mannine, whose review scope is limited to measurement methodology and product claims. Disclosure: Prime AI Visibility and Percepture have a commercial relationship; Percepture provides managed implementation. The Lead Seeker is referenced only as a separate business-development resource, not as part of the client AI-visibility audit. > **Run the audit against real engine answers, not guesses** > > Prime AI Visibility takes the prompt set from this template and runs it across ChatGPT, Perplexity, Gemini, and Claude, so your pitch shows the prospect exactly what the engines say about them today. > > **[Run the agency audit template](https://app.primeaivisibility.com/sign-up)** ## References 1. Google Search Central, *AI Features and Your Website (AI optimization guide)* (2026). 2. Percepture, *How AI Search Optimization Tools Increase Organic Traffic (OPTK)* (2026). 3. Google Search Central, *A new resource for optimizing for AI* (2026). ## Next steps 1. **[Move the audit into a full delivery loop](https://primeaivisibility.com/articles/agencies/how-agencies-boost-client-ai-visibility)** once the pilot is approved, so quick wins become a repeatable program. 2. **[Set up the reporting layer](https://primeaivisibility.com/articles/agencies/agency-client-ai-visibility-reporting)** so the baseline you capture here turns into a defensible monthly review. 3. When you are ready, **[create a Prime AI Visibility workspace](https://app.primeaivisibility.com/sign-up)** and bring the 20-slot prompt set you built to run the audit against live engine answers. ## Frequently asked questions **What is an AI visibility audit template for agencies actually for?** It is a pre-sale and onboarding instrument that replaces assumptions with observations. It shows a prospect exactly how ChatGPT, Perplexity, Gemini, and Claude describe them today, ranks the findings by effort and impact, and proposes a bounded pilot — so the conversation is about evidence, not a pitch deck of guesses. **Is the 20-prompt sample a benchmark I can compare clients against?** No. It is a starter structure organized by intent, meant to be filled with each prospect's real category, competitors, and locations. The value is intent coverage — spotting whether the prospect is absent across a whole stage of the buyer journey — not a universal score you can rank clients on. **Can I promise results from what the audit finds?** No. Show the prospect what the engines return today and describe the work that would follow, but never promise a ranking or a citation. No engine publishes its selection logic, and answers vary by platform, model, search state, location, prompt wording, time, and run, so every finding is a bounded snapshot. **How do I decide between a software-only, agency-led, or hybrid engagement?** Use the audit's owner column. If the client has an internal team that can execute, software-only fits. If they want the work done for them, agency-led fits. If they have partial capacity but need specialist remediation, hybrid fits — the most common shape, and where a managed partner executes against the audit's findings. **How is this audit different from the monthly reporting process?** The audit is a one-time diagnostic that scopes an engagement; the reporting process is the recurring rhythm that tracks movement once work is underway. The audit answers "what should we do?"; reporting answers "did it move, and can we prove it?" Keep them distinct so a prospect understands what a pilot delivers versus an ongoing retainer. > **Turn a template into a scoped pilot** > > Use the effort/impact scorecard to propose a pilot the prospect can say yes to — a fixed prompt set, a baseline, and a short list of owned quick wins. > > **[Start the audit](https://app.primeaivisibility.com/sign-up)**