Questions and answers

Usually because the sources those engines read in your category do not name you. We put ten to thirty-five buyer questions to five engines, three measured runs each, and record who is named instead of you and which pages the engines cited to name them. What comes back is a list of specific questions and specific sources, not a score.

One category, ten buyer questions, five engines, three measured runs each. A written findings document naming the vendors the engines named instead of you and how often, the questions you lose, the sources the engines read to name them, your position on the absence ladder, and three prioritised fixes stated as which page to update or which publisher to get in front of. Findings within 48 hours of the category being confirmed.

Because we do not ask once. Every question is put to every engine three times, and where the three runs disagree we run that pair up to twice more. The disagreement is reported next to the rate rather than averaged away, so you can tell a real position from a coin landing one way.

Citations are not the same as being on the shortlist. Every question we ask is one of four types: shortlist, role, use case, evaluation, and findings report where you are named per type. The most common shape we find is a company cited on use-case questions and absent when a buyer asks who to buy from. The findings name which questions and which sources to fix first.

No. Google's own guide to generative AI search lists all three under what you do not need to do, and we agree. The entity work we do is the standard structured data the engines already read, nothing AI-specific. What moves an answer is being named by the sources the engines read.

No. The Category Audit and the Diagnostic are written diagnoses: every finding carries its source, its run count and its fix. The program is done-for-you work with a full re-score each quarter. A tracking view is included with the program, but it is not the thing you are buying.

We check before anything else. Hosting rules, security settings and bot filters block AI crawlers silently, and most visibility tools report a healthy site anyway. It is a pass or fail check and it is the first line of every Category Audit.

Every engagement opens with a baseline on a fixed question set and re-scores against the same set, so movement is checkable rather than asserted. Every figure carries its numerator, its denominator and its run count, and the method is published on this site. Nothing is reported as work performed; it is reported as what was measured before and what was measured after.

Placement is often part of the answer, but which placement is a measurement question. We report which sources the engines actually read in your category before anyone spends on getting in front of them, so effort goes to the handful of surfaces that appear in your answers. Measurement first, placement second, and only where the engines are reading.

Ask the same five questions of anyone you are considering. Do they publish their method, with a version and a date? Do they run each question more than once, and report the disagreement? Does every finding name the source it came from? Do they sell llms.txt, AI-specific schema or content chunking, which Google says you do not need? Do they promise a rank or guaranteed leads? The answers separate a measurement practice from a rewritten SEO deck.

We measure which vendors AI engines name when buyers ask for software in a category. The program builds the evidence AI systems can use, then reports mentions, recommendations and citations separately.

We test five engines: ChatGPT, Claude, Perplexity, Google AI Overviews and Google AI Mode. The 2026 State of GEO, Volume I, held one engine constant; its research figures describe that study.

The Category Audit is $490 once: one category, ten buyer questions, five engines and three measured runs. You receive the vendors named instead with counts, the questions lost, the sources engines read, your position on the absence ladder and three prioritised fixes. Findings arrive within 48 hours of category confirmation.

The AI Visibility Diagnostic is $990 once. Method v1.1 uses 35 questions, five engines, three scheduled runs per question-and-engine pair and up to two adaptive runs, giving 525 scheduled observed answers. The findings show your baseline, competitors and sources, with a prioritised plan.

The Category Audit and the Diagnostic are sold by checkout, with no call. The Diagnostic requires no request form. Talk to us about the program through the contact page.

The full $490 Category Audit fee credits against the Diagnostic within 30 days. The full $990 Diagnostic fee credits against the program's first monthly invoice. Buyers receive the promotion code by email.

The Free 10-question check runs ten buyer questions through one engine in the browser in under a minute, with one result per email address per day. It is a first look, separate from the paid Category Audit and Diagnostic.

The program is $13,500 per 90 days, billed $4,500 monthly. The initial term is 90 days, then month to month. The full $990 Diagnostic fee credits against the program's first monthly invoice.

Each month: four answer pages, a schema and entity pass, third-party placement work, and reporting on mentions, citations and sources. The full 35-question set is re-scored quarterly under the published method.

If Broadcastwell is not the right fit, you get your money back. A standalone Category Audit or AI Visibility Diagnostic is refundable in full, no questions asked, if requested within 30 days of delivery of the findings. The first month of the program is refundable in full, no questions asked, at any time before month two is invoiced. You do not need to give a reason, and you keep every deliverable produced up to that point. The only exception is third-party costs paid on your behalf, and we tell you about any of those before they are incurred.

At the day-90 re-score, the named count must be strictly greater than baseline on the same agreed questions, engines and method version. A tie fails. If the gate is not met, Broadcastwell does not invoice month four and you may end the engagement with no further charge. Under the current terms, v1.1 covers all 35 agreed questions across five engines. The panel is frozen, each question-and-engine cell counts once, and cells lost to engine errors are excluded from both sides and reported. Except under the refund clause, months one to three are not refundable. No inclusion, ranking, revenue or commercial result is guaranteed.

No. We document observed answers and the work performed. We do not promise inclusion, ranking, traffic, leads, revenue or a particular score.

Broadcastwell LLC is registered in Indiana . We are 12 people, working on different roles: management consultants, AI engineers, AI architects and specialists across Bloomington, Chicago, San Francisco and New York.

Yes. The research, method and datasets are published openly. The research volumes have DOIs and have not been peer reviewed. Read the work and its stated limitations under Research.

Approved case studies cover Vantaa Tax, the named B2B software case, followed by Neurvalis AI Automation and Valren Atelier. Client baselines, recommendations and outcomes remain private unless expressly approved for publication.

Agency wholesale is $1,000 per client per month for the first client and $490 per client per month for each additional client, white-label, with the agency retaining the client relationship. The opener is one free Diagnostic on one agency client.