Doctors say FUE. Patients don't.

Three and a half million search impressions across seven hair restoration practices show a consistent gap between clinical vocabulary and the words patients actually type — and a much larger gap between the words that get seen and the words that get clicked.

There is a working assumption in elective medicine that the words on the consent form are the words in the search bar. Practices name service pages after the systems they license, describe conditions with the diagnosis rather than the complaint, and organize their content the way a surgeon organizes a differential.

That assumption is testable. I had sixteen months of Google Search Console data across seven hair restoration practices in six states: 143,100 distinct queries and 3,566,763 impressions. It seemed a reasonable place to check whether clinical vocabulary and consumer vocabulary are the same vocabulary.

They are not. The size of the gap was roughly what I expected. What the gap turned out to predict was not.

What this study measured

This study measured the vocabulary gap between clinical and consumer language in hair restoration search, using first-party Google Search Console data from seven practices over sixteen months. The dataset covers 143,100 distinct queries, 3,566,763 impressions and 12,745 clicks between 23 April 2025 and 28 August 2026.

Every practice in the sample ranks across a wide share of the category's informational surface. That breadth is the reason the dataset exists at all, and it is worth stating plainly: these are well-indexed sites appearing for a very large number of relevant searches. The question this research asks is not whether that visibility was earned. It is what different kinds of visibility turn out to be worth.

What do patients call the procedure?

Patients overwhelmingly call it a "hair transplant." The phrase generated 768,637 impressions across 12,010 distinct queries, against 290,190 impressions and 2,716 queries for "FUE" — a ratio of 2.6 to 1 on impressions and 4.4 to 1 on distinct phrasings. "Hair restoration," the term most practices use in their own business names, generated 326,691 impressions.

The "hair plugs" question

"Hair plugs" is still searched but almost never clicked. The term produced 14,324 impressions across 219 distinct queries and 5 clicks — a 0.03% click-through rate, among the lowest of any term measured. I went into this analysis expecting to argue that the industry was wrong to bury the phrase. The data does not support that argument. The word survives in the public vocabulary, but the people still using it are not the people booking consultations.

What do patients call hair loss itself?

Consumers say "hair loss" and "balding." Clinicians write "alopecia." Across the study period "hair loss" generated 664,077 impressions against 55,398 for all forms of "alopecia," a ratio of roughly 12 to 1. "Balding" generated 193,200 impressions against 37,693 for "androgenetic alopecia," a ratio of roughly 5 to 1.

The full diagnostic term does appear, and at 37,693 impressions it is not rare. At a 0.05% click-through rate, however, that traffic behaves far more like professional and academic research than like prospective patients.

Do patients search for branded surgical systems?

Almost no one searches for branded surgical systems, and the few who do are largely not patients. The three licensed harvesting and robotic platforms in this dataset generated 44,482 impressions and 8 clicks combined across sixteen months, a 0.02% click-through rate. A substantial share of those queries asked about the cost or specifications of the machine itself — the search behavior of a competing clinic doing procurement research, not a consumer choosing a surgeon.

This is not an argument against licensing the platforms. Device partnerships carry real clinical and credentialing weight, and a recognized system name can do meaningful work inside a consultation, on a credentials page, or in a referral conversation. The finding is narrower and more specific: a device name is a trust asset for a patient who is already engaged, not an acquisition channel for one who is not. Those are different jobs, and the search data suggests only the first one is real.

Which searches actually produce clicks?

Queries containing a practice or surgeon name converted at 34 times the rate of every other query in the dataset. Branded searches accounted for 27,432 impressions — just 0.8% of the total — but produced 2,667 clicks, or 20.9% of all clicks recorded. Non-branded searches accounted for the remaining 3,539,331 impressions and converted at 0.28%.

When the full corpus is re-sorted by click-through rate rather than volume, procedure vocabulary disappears from the top of the list entirely. The thirteen highest-converting queries in the dataset were, without exception, names: the practice, the surgeon, the surgeon's name followed by "MD," and in several cases the practice name misspelled. Individual rates ranged from 23% to 60%. The first non-branded query appears at position fourteen.

The vocabulary ladder runs one direction. People who know what they want search for who does it. People who don't search for what it is.

Does search volume predict value?

Search volume did not predict conversion in this dataset, but the kind of intent a query expresses did. Aggregated across all seven properties, "hair transplant near me" generated 15,152 impressions at a 0.82% click-through rate, while "hair loss specialist near me" generated 381 impressions at 8.40% — roughly ten times the rate on one fortieth of the volume.

I first read this as evidence that longer, more specific queries convert better. Tested properly across the whole corpus, that is not what happens. Grouping all 143,100 queries by word count produces a shallow arc rather than a slope: 0.17% at one word, rising to 0.44% at four, then falling back to 0.19% by nine. Length carries a little information at the very short end and almost none after that.

Length is not the variable. What a query is trying to do is the variable. Queries containing provider-seeking language — "near me," a specialist role, a clinic, a doctor — converted at 0.70% across 223,094 impressions, against 0.29% for question-shaped informational searches across 1,001,605 impressions. Volume-sorted keyword research, which is how most content plans are built, ranks the provider-seeking terms far below the informational ones.

What this means for how a practice publishes

The finding is not that informational content fails. It is that informational content and conversion content are doing two different jobs, and the search data makes the division unusually legible.

1. Procedure and condition content is an awareness asset. It reaches an enormous audience early in consideration, and in this dataset it reached roughly three and a half million impressions worth. It should be measured on reach and assisted influence rather than on direct consultation volume, because direct consultation volume is not what it produces.
2. Name recognition is the conversion asset. Branded search converted at 34 times everything else. That means the highest-leverage question for most practices is not which keyword to rank for — it is how a stranger comes to know the practice's name in the first place, which is largely a job for reputation, referral, video, local presence and paid reach rather than for the blog.
3. Write in the patient's vocabulary, not the chart's. Page titles, headings and the first sentence of any consumer-facing page should use "hair transplant," "hair loss" and "balding." Clinical terms belong in the body, where they establish precision without costing discoverability.
4. Do not judge a narrow keyword by its volume. The terms with a few hundred impressions and double-digit conversion are worth more per impression than anything at the top of a volume-sorted list. They are also far cheaper to rank for.
5. Give device partnerships the right job. Platform names earn trust in the consultation and on credentials pages. Asking them to also acquire patients from search is asking for something the search behavior does not support.

What this research does not show

• It does not show that nobody searches "FUE." 2,716 distinct queries did, generating 290,190 impressions. The finding is about ratio, not absence.
• It does not establish causation on branded queries. People searching a practice by name click because they have already decided, and something else earned that decision. This dataset cannot identify what. That is the most interesting open question here and it needs a different study.
• Seven practices are not a market, and they are not seven equal voices. This is one agency's portfolio across six states — a large sample of a narrow slice, not a representative sample of United States hair restoration search. The largest property supplies 66.3% of the impressions analyzed, so the aggregate numbers lean heavily on it.
• Google does not disclose every query. Search Console withholds rare searches to protect user privacy. Measured directly across these seven properties, the query dimension accounted for between 40% and 77% of impressions, and between 33% and 49% of clicks on the five properties with meaningful click volume. Because suppression falls hardest on rare terms, the real long tail of patient phrasing is larger and more varied than anything shown above.

Frequently asked questions

Methodology

Sources

• Google Search Console Help, "Performance report" — documents that queries made by only a few users are omitted from the Performance report to protect user privacy.
• Google Search Console Search Analytics API reference — documents the 16-month data retention window and the 25,000-row response limit that governs this dataset.
• Primary data: author's analysis of Google Search Console exports, archived 31 August 2026. Aggregated figures available on request.