Procedural guidance
Seeing under the skin in real time.
Anaesthetists, intensivists, nurses.
Case study
Even the number one brand is invisible when it matters most.
This page is a complete, independent AI visibility audit of a category leading medtech company, published as a worked example of the method. Butterfly Network makes handheld point of care ultrasound and is one of the best known names in it. Everything below comes from public data and public AI answers.
Independent analysis using public data and public AI answers. Not commissioned by or affiliated with Butterfly Network.
17
queries across four archetypes and five stages
34
AI answers, ChatGPT and Perplexity
0 of 14
problem and solution stage answers naming the brand
01 Why this company
Chosen against a strict test, not because the name was useful.
Three conditions had to hold. The clinician has to be the primary buyer, so that the person asking AI is also the person deciding. The product has to do genuinely different clinical jobs, so that real buyer archetypes exist rather than invented ones. And the company has to be large enough that a gap would actually mean something.
Several better known companies were rejected because they failed one of these, most often because the buyer turned out to be procurement or pharma rather than the clinician. Butterfly Network passed all three.
02 How the buyer was mapped
A clinical audience is never one buyer.
Four archetypes, built from the clinical job rather than the specialty. Each one carries its own exams, its own specialties and its own buying triggers.
Emergency triage
A fast answer in a crashing patient.
ED physicians, paramedics.
Focused diagnosis
A targeted read of one organ.
Cardiologists, ICU, GPs.
Quick measurement
One quantified answer.
Nurses, obstetrics, urology.
Overlaps flagged as a feature
A cardiac scan is focused diagnosis in clinic and emergency triage on a crashing patient. Context dependent archetypes were recorded as exactly that, not forced into one box to make the spreadsheet tidy.
Scope stated
Core adult hospital and point of care settings. Paediatric, veterinary, military and ophthalmic uses were deliberately excluded, and the exclusion was written down rather than left implied.
03 How the capture was controlled
Fixed first, then captured, so the test could not drift.
-
17 English queries fixed in advance: 4 problem, 3 solution, 4 category, 3 vendor and 3 objection.
-
ChatGPT and Perplexity, logged out, incognito, a fresh chat for every query. 34 answers in total.
-
Layered scoring adapted to this market. Clinicians already know ultrasound exists, so each answer was scored on three layers: was ultrasound mentioned, was a handheld option mentioned, was Butterfly named. The brand can lose at any layer, and only layered scoring shows which one.
-
Every Perplexity citation classified by source type, then split by whether Butterfly was present in that answer or absent from it.
04 What the method found
A cliff by stage, not by archetype.
Ultrasound was recommended in 34 of 34 answers, so the demand is not in question. A handheld option was mentioned in 0 of 6 solution stage answers. Butterfly was named in 0 of 14 problem and solution stage answers, and then in 13 of 14 device comparison answers, ahead of every competitor.
When buyers describe the problem
0 of 14
Butterfly is not named
When buyers compare devices
13 of 14
Butterfly is ranked first
Describes the problem
Explores solutions
Compares devices
Picks a brand
Category level invisibility, not competitive loss
No vendor of any kind appeared at the problem or solution stage. Butterfly wins the buying conversation. The buying conversation simply starts too late.
The mechanism, read from the sources
Answers without Butterfly drew mainly on academic and clinical education pages that teach the technique and never name a device. Answers with Butterfly drew on buying guides, forums and vendor sites. The brand lives in the buying world and is absent from the technique world.
Recommendations, not delivered results.
05 What was recommended
Output the client can check, and a baseline to move from.
The audit delivers a capture sheet, a scorecard workbook covering presence by stage and archetype, competitor appearance and source classes by presence, and a report carrying conditional recommendations.
-
Technique level content that answers the problem stage queries directly, because that is the layer where no vendor currently appears.
-
Presence in the clinical education sources the AI already trusts and cites, rather than more pages on the vendor site.
-
A quarterly KPI: handheld mentioned and brand named rates at the problem and solution stage, tracked from a 0% baseline.
Recommendations, not delivered results. Captured from the Netherlands. The finding rests on source type rather than regional brand preference, but capture location is a stated limitation.
Work with me
Let's make your product impossible to miss.
Send me your product. I will show you where you are invisible in AI, and what it takes to fix it.
eva@evalorger.comEva Lorger, healthcare and medtech content and AI visibility.