--- title: "7 prompts to test if AI agents can find, use, and buy from your brand | Hyperize" description: "Seven copy-paste prompts that test whether AI agents can find your brand, read your pages, complete your transaction, and cite your evidence. A self-serve AI agent readiness audit: the same diagnosis Hyperize runs against DAX 40 brands, adapted so you can run it yourself in any AI assistant." canonical: https://www.hyperize.ai/en/insights/articles/can-ai-agents-buy-from-your-brand-seven-prompts lang: en last-updated: 2026-07-03 --- # Can AI agents buy from your brand? Ask them. [Home](/en/) [Insights](/en/insights) Can AI agents buy from your brand? Ask them. Leadership · Agent Experience The self-serve version of the tests we run on the DAX 40. Seven prompts, one hour, no tool. Marc Seefelder 11 min read 7prompts · copy, paste, read 4gates · find, read, transact, cite 1hour · any AI assistant 0tools · no signup, no vendor Seven prompts and one hour tell you what a year of SEO reporting cannot: whether AI agents can find your brand, read your pages, and finish a purchase. We run these tests against the DAX 40, wave after wave [\[S1\]](#source-s1). The pattern is stable: strong brands, blank pages, intercepted sales [\[S2\]](#source-s2)[\[S3\]](#source-s3). Below are the exact prompts, what a failing answer looks like, and why each failure costs revenue. No tool. No signup. Run them today. Contents - [01 · Your SEO audit cannot see this](#section-01) - [02 · Load the context first](#section-02) - [03 · Prompt 1 · The invisibility check](#section-03) - [04 · Prompt 2 · The memory check](#section-04) - [05 · Prompt 3 · The blank-shell check](#section-05) - [06 · Prompt 4 · The task check](#section-06) - [07 · Prompt 5 · The interception check](#section-07) - [08 · Prompt 6 · The evidence check](#section-08) - [09 · Prompt 7 · The vocabulary check](#section-09) - [10 · Seven answers, one map](#section-10) - [11 · What the prompts cannot tell you](#section-11) Section 01 ## Your SEO audit cannot see this. AI agents do not read rankings. They assemble answers, and your brand is either material for the answer or it is not. Your SEO stack measures positions in a list that a human scrolls. An agent audit measures something else entirely: can a machine acting for your customer retrieve you, read you, transact with you, and cite you. Different questions. Different failures. Different money. FIG 01 · Two different questions An SEO audit asks An agent audit asks Where do we rank? Do we appear in the answer at all? Which keywords do we own? Which buyer questions get us cited? Is the page optimized? Can a machine read the page without a browser? How much traffic do we get? Can an agent finish the purchase? Who outranks us? Who intercepts the transaction? Is the content fresh? Does the AI's memory of us match reality? We run this audit against DAX 40 brands with real agents, on fixed tasks, wave after wave [\[S1\]](#source-s1). This article is the version you can run yourself. Seven prompts. Copy, paste, read the answer. Save this. You will need it. Section 02 ## Load the context first. Most people skip this step. It is also why most people get generic answers. Open your AI assistant and paste this before anything else. Fill in every bracket. It takes three minutes and it sharpens every prompt that follows. Setup · paste once You are testing how AI agents handle the brand \[brand\]. Context: website \[URL\]. Category: \[category\]. Market: \[market\]. The transaction a customer completes with us: \[buy / get a quote / book / apply\]. Our three main competitors: \[A\], \[B\], \[C\]. Hold this context for every test that follows. After each test, report three things: what you found, what you could not find or do, and where you went instead. The last line matters. "Where you went instead" is where the expensive findings live. Section 03 ## Prompt 1 · The invisibility check. Start where the money starts: a buyer question with no brand in it. Run this in a fresh chat, logged out or in a temporary session, so personalization cannot flatter you. Zero history, zero mercy. The prompt · fresh chat, web search on Search the web and answer: which \[category\] providers in \[market\] would you recommend to a customer right now, and why? List every source you used. Three possible outcomes. Named and cited: your pages are material for the answer. Named but not cited: borrowed visibility, the engine knows you from other people's pages and other people control what it says. Absent: for this question, you do not exist. Why this matters: retrieval is the gate. Everything downstream of this answer, the comparison, the shortlist, the purchase, happens inside material you are either in or not. And absence is rarely a crawling problem: in our own ledger, an engine denied that a live, indexed dataset existed while quoting numbers from it the same week [\[S4\]](#source-s4). The gap was entity association, not coverage. Section 04 ## Prompt 2 · The memory check. Every AI model carries a frozen picture of your brand. Training data ages like sediment [\[S7\]](#source-s7). The prompt · fresh chat, search OFF What do you know about \[brand\]? Answer from memory only, do not search the web. Then state how current you believe your picture is, and what you are unsure about. Then open a second fresh chat, ask the same question with web search on, and compare the two answers. Why this matters: the gap between the two answers is your dependence on retrieval. If memory is stale and prompt 1 showed weak retrieval, agents are working from the fossil layer. > If retrieval fails, the agent sells yesterday's version of your company. Section 05 ## Prompt 3 · The blank-shell check. GPTBot, ClaudeBot and PerplexityBot do not render JavaScript; Gemini, riding Googlebot, is the measured exception [\[S5\]](#source-s5). If your page builds itself in the browser, most agents see the scaffolding, not the content. The prompt · your most important product page Fetch \[product page URL\] and list everything a customer could learn from it: products, prices, specs, availability, and the next step. Then list what you could not read or open. Why this matters: this failure hides in plain sight, because the page looks perfect in your browser. When we tested German car brands, a coding agent found 227 interactive elements on the BMW configurator and could read none of them [\[S2\]](#source-s2). The page title said "Konfigurator". The page content, to a machine, was an empty string. > A page an agent cannot read does not exist. Section 06 ## Prompt 4 · The task check. Reading is stage one. The real test is the job. The prompt · the transaction itself Act for a customer who wants to \[buy / get a quote for / book\] \[product\]. Start at \[URL\]. Get as close as you can to the point where only payment or a signature remains. Log every step: what you clicked, what you could not read, where you stopped, and what you would tell the customer to do instead. Read the log like an accident report. Where exactly did the agent stop: a cookie wall, a configurator, a form it could not parse, a login it could not pass? Why this matters: on the map we draw for every commerce wave, the top-right zone, found and fully usable, keeps coming back empty [\[S1\]](#source-s1)[\[S2\]](#source-s2). And the last line of the log is the quiet killer: "what you would tell the customer to do instead" is the sentence your customer will actually receive. If it names someone else's website, that is your handoff, gone. Section 07 ## Prompt 5 · The interception check. The most expensive failure never shows up in your analytics: the agent completes the job, on someone else's domain. The prompt · fresh chat, no brand mentioned I want to \[buy / insure / book\] \[product category\]. Where exactly should I do that? Name the one place you would complete it, then the second-best option. If the answer is your domain, you own the close. If it is a comparison portal, a marketplace, or a dealer, the agent has decided your own surface is not the place to finish the job. We call this pattern Third-Party Interception, and our insurance wave shows both directions of it: comparison portals capture the German car-insurance query before Allianz is even named, and where the close truly sits with an intermediary, the AI names the brand but not the route to the deal [\[S3\]](#source-s3). > You are not losing the answer to a competitor. You are losing the transaction to the middleman. Section 08 ## Prompt 6 · The evidence check. When an AI recommends you, it needs material to argue with. The question is whose material. The prompt · sources mandatory Why should a customer choose \[brand\] over \[competitor\]? Cite a source for every claim you make. Read the citations, not the prose. Your pages: you control the argument. Trade press and forums: others control it. No sources at all: the engine is improvising, and it will improvise differently tomorrow. Why this matters: in our ledger, the same brand page won and lost on the same day. With a citable evidence page live, engines cited it above Wikipedia. In plain language, without the page's own vocabulary, they fell back to trade press [\[S4\]](#source-s4). If your best argument lives in a PDF or a brochure page a machine cannot parse, the AI argues your case with someone else's numbers. Section 09 ## Prompt 7 · The vocabulary check. The strongest citation position is a term you own. This prompt tests whether you actually own yours. The prompt · fresh chat, do NOT mention your brand Explain \[your coined concept or product term\]. Who is the authority on it, and which sources would you cite? Three outcomes again. You are cited: the term is yours, build on it. A competitor is cited: the term is not yours, stop pouring content into it and rename or prefix. Nobody is cited: open ground, the cheapest land grab in AI visibility. Why this matters: owned vocabulary is the sharpest effect in our citation ledger. Queries carrying vocabulary a brand coined get cited near-always; plain-language versions of the same question lose to incumbents and trade press [\[S4\]](#source-s4). Section 10 ## Seven answers, one map. The seven prompts sort into four gates. An agent that fails one gate never reaches the next. FIG 02 · Four gates, seven probes Two of these gates are the published axes of our Agent Success Score, and the weighting is the uncomfortable part: AI Usability carries 70 percent of the score [\[S6\]](#source-s6). Most brands invest in being talked about. The score pays out on being usable. FIG 03 · The scorecard | Check | It tests | A failing answer sounds like | | --- | --- | --- | | 1 · Invisibility | Are you material for the buyer answer? | "Here are five providers", and you are not one of them. | | 2 · Memory | How stale is the model's picture of you? | A confident description of your 2023 company. | | 3 · Blank shell | Can a machine read the page? | "The page appears to contain mostly scripts." | | 4 · Task | Can an agent finish the job? | "I could not proceed past the configurator." | | 5 · Interception | Who owns the close? | "I would complete this on \[comparison portal\]." | | 6 · Evidence | Whose material argues for you? | Every claim cited to trade press, none to you. | | 7 · Vocabulary | Do you own your own terms? | Your concept, explained with a competitor as the source. | Section 11 ## What the prompts cannot tell you. Honesty section. One prompt is one engine, on one day, with one phrasing. Engines drift week to week, answers vary run to run. A single pass is a smoke test, not a measurement. What the instrumented version adds: fixed tasks so results compare across brands, repeated runs so one lucky answer cannot flatter you, scoring per lane, and re-probes on a schedule, because citations decay [\[S4\]](#source-s4). That is what we run against the DAX 40 [\[S1\]](#source-s1). If the smoke test stings, the free Snapshot is the same diagnosis run as an instrument: one lane, one task, measured and documented. It is the top of our funnel and we are not hiding that. The prompts above are yours either way. What this article does not contain The prompts diagnose. They do not prescribe. No build recipes, no scoring derivation, no task protocols, no measurement harness. You can find the gap with this page; closing it is the work. That line is deliberate. Ninety percent of readers will save this and never run a single prompt. The agents will keep visiting their websites anyway, reading what they can, buying where it works. Now stop reading and run prompt 1. What comes next Re-run the seven prompts monthly. Engines move, answers drift, and a pass in July is not a pass in October. Our own re-probe discipline exists for exactly that reason: citations decay, so proof that is not re-measured stops being proof. \[ FAQ \] ## Three questions, answered straight. ### Can I test my brand against AI agents without a tool? Yes. Seven prompts in a fresh AI chat cover the four gates that decide agent outcomes: can agents find you, read you, transact with you, and cite you. You need an AI assistant with web access, your own URLs, and about an hour. The limits: one run is a smoke test, not a measurement; engines vary by day and phrasing. [The prompt stack · from section 03](#section-03) ### What is the difference between an SEO audit and an agent audit? An SEO audit measures positions in a ranked list read by humans. An agent audit measures whether a machine acting for a customer can retrieve your brand, read your pages without a browser, complete your transaction, and cite your evidence. A page can rank on Google and still be a blank shell to an AI agent, because major AI crawlers do not render JavaScript. [Two different questions · Section 01](#section-01) ### What do these seven prompts not measure? Variance and repeatability. A single prompt is one engine on one day with one phrasing. An instrumented audit fixes the task, repeats the runs, scores per lane, and re-probes on a schedule, because citations decay. The prompts find the gap; they do not size it. [Honest limits · Section 11](#section-11) Sources ## Evidence and provenance. S1 internal DAX 40 Agent Success Index, measured by real agents, wave by wave Hyperize · live · per-brand scores on two axes, re-measured; the instrumented version of the diagnosis in this article [https://www.hyperize.ai/en/dax40-index](https://www.hyperize.ai/en/dax40-index) The measurement program these prompts are distilled from, and the pattern the map in section 10 describes. S2 internal Can AI agents use German car configurators? (Wave: Automotive) Hyperize · 2026-05 [https://www.hyperize.ai/en/insights/articles/four-german-cars-tested-by-ai-agents](https://www.hyperize.ai/en/insights/articles/four-german-cars-tested-by-ai-agents) The blank-shell finding behind prompt 3: a coding agent found 227 interactive elements on the BMW configurator and could read none of them. S3 internal Three German insurers, tested by AI agents (Wave: Insurance) Hyperize · 2026-06 [https://www.hyperize.ai/en/insights/articles/three-german-insurers-tested-by-ai-agents](https://www.hyperize.ai/en/insights/articles/three-german-insurers-tested-by-ai-agents) The interception pattern behind prompt 5: brands get named while intermediaries own the close. S4 internal The Proof Flywheel, public citation ledger Hyperize · 2026-06 · query classes declared per entry; wins and losses published; fixed re-probe dates [https://www.hyperize.ai/en/insights/articles/the-proof-flywheel](https://www.hyperize.ai/en/insights/articles/the-proof-flywheel) The citation evidence behind prompts 1, 6 and 7: a vocabulary-led win and a plain-language loss on the same page, and the whitespace loss whose diagnosis is entity association, not crawling. S5 internal AI agents built our websites (making-of) Hyperize · 2026-06 [https://www.hyperize.ai/en/insights/articles/ai-agents-built-our-websites](https://www.hyperize.ai/en/insights/articles/ai-agents-built-our-websites) The verified crawler finding behind prompt 3: none of the major AI crawlers currently render JavaScript. S6 internal Agent Success Score, methodology Hyperize · live · two axes, published weights [https://www.hyperize.ai/en/methodology/agent-success-score](https://www.hyperize.ai/en/methodology/agent-success-score) The published axes and weighting behind the map in section 10: AI Usability carries 70 percent of the score. S7 internal Is your brand ready for AI agents? (the Fossil Layer Test) Hyperize · 2026-03 [https://www.hyperize.ai/en/insights/articles/brand-ready-for-ai-agents](https://www.hyperize.ai/en/insights/articles/brand-ready-for-ai-agents) The frozen-memory model behind prompt 2: training data ages like sediment; retrieval is the living layer. Related ## Run the prompts. Check the score. Get the instrument. [ Answer ### The five-minute version. Can you test your website's AI readiness without a tool? The short, direct answer to the question this article unpacks. Read the answer](/en/answers/test-website-ai-readiness)[ Methodology ### The score behind the prompts. Agent Success Score: two axes, published weights, per-lane measurement. What the instrumented version of this diagnosis measures. Read the methodology](/en/methodology/agent-success-score)[ Free snapshot ### The instrumented version. One lane, one task, run as a measurement instead of a smoke test. Free, documented, and yours to disagree with. Get a snapshot](/en/#snapshot) Page type · Article (Leadership) Published2026-07-03 Updated2026-07-03 Next review2027-01-03 [Machine-readable record](https://www.hyperize.ai/en/insights/articles/can-ai-agents-buy-from-your-brand-seven-prompts.json) [Back to Insights](/en/insights) --- Canonical: https://www.hyperize.ai/en/insights/articles/can-ai-agents-buy-from-your-brand-seven-prompts Machine-readable index: https://www.hyperize.ai/llms.txt