Coming up with healthcare content ideas is the easy part. Symptoms, conditions, tests, treatments, specialties: a healthcare content marketing team can list hundreds, even thousands, of searches in an afternoon. The hard part is saying no. The real skill isn’t finding more keywords, it’s telling which searches genuinely need a new page and which ones an existing page should already answer.
Magnero’s content and SEO team built its first version of this process the familiar way: a keyword list, a search volume check, some topic grouping, a calendar. As the list grew, it became clear that ideas had never been the problem. The real question was simpler and harder at the same time: does every search really need its own page?
That question matters more now than it used to. People no longer take health questions only to a search box. When OpenAI launched ChatGPT Health in January 2026, it said more than 230 million people ask ChatGPT health and wellness questions every week. Search itself has changed too. Google’s AI Overviews and AI Mode can split a complex question into subtopics and search each one at once, a process Google calls query fan-out, and in Türkiye both features began rolling out in February 2026.
Google’s own guidance on generative AI search answers the core question directly: creating separate content for every variation of a query, when the main goal is to manipulate rankings, violates its spam policies, and more pages don’t make a site higher quality or more relevant on their own. Search volume tells a content team that demand exists. It doesn’t tell them which page should meet that demand, or in what form.
This is the decision system our team uses instead: eight questions, organized into three stages, applied before a search becomes a page.
Why keyword clusters hide different decisions
An SEO tool will often group these five searches into one topic cluster:
- What causes headaches?
- What is a migraine?
- How is migraine diagnosed?
- What are the treatment options for migraine?
- What kind of doctor treats migraines?
The tool sees five variations of one topic. The people behind those searches are making five different decisions. The first is trying to make sense of a symptom. The second wants to understand a condition. The third is curious about diagnosis, the fourth is weighing treatments, and the fifth is looking for the right care.
The trouble starts when every keyword gets its own URL before anyone notices that difference. Pages pile up answering the same question under different titles, one current, one outdated, one living on an entirely different URL pattern. Readers end up with three answers, editors end up maintaining three pages, and search engines end up with three pages competing for one query, a pattern SEOs call keyword cannibalization.
A hospital website might answer “how long is recovery after knee replacement?” in four separate places: the orthopedics page, an old blog post, an FAQ page, and a surgeon’s profile, each written by a different team at a different time, with estimates that don’t quite match. Search engines have to guess which page to show, and a patient can easily land on the least current one. None of these pages was a bad idea in isolation. Sprawl like this rarely comes from one big mistake. It builds up over years of sensible-looking “let’s add a page for this keyword too” decisions.
Stage 1: Should this page exist at all
The first three questions determine whether a page has a legitimate reason to exist before a new URL is created.
What decision is the searcher actually trying to make
The first question in review isn’t how much search volume a keyword has. It’s what decision the person behind that search is trying to make. Health searches map to five recognizable decision moments: noticing a symptom and wondering what it means, understanding a condition in full, preparing for a diagnosis, choosing between treatments, and finding the right place to get care. In the migraine example above, each of the five searches lands on a different one of these moments.
This lines up with the “why” question in Google’s own guidance on helpful content: content should exist primarily to help people, not to attract search traffic. So a keyword never gets turned straight into a URL. The need and the decision moment get defined first.
Do we already answer this
When a new keyword surfaces, the question isn’t “can we write about this.” It’s “do we already have a page that meets this need.” In healthcare specifically, too much content is as much of a problem as too little. Mustafa Çelik, who leads content and SEO strategy on our team, put it plainly:
Search volume shows that demand exists, but demand alone doesn’t call for a new page.
Existing pages get evaluated against the need first, and a new page only gets considered once that evaluation is complete and the purpose is genuinely clear. The outcome is one of five: create, improve, consolidate, reposition, or retire.
Two of these deserve a closer look. A page about to be retired often still holds something worth keeping, a strong paragraph, a good example, a recent data point. That material gets moved to the page that stays, and the old URL gets a permanent 301 redirect. Repositioning applies when a page has a genuinely distinct angle but is aimed at the wrong query: instead of deleting it, the target query changes and the title gets rewritten to match.
The test for retiring a page is its purpose, not its age. After CNET deleted thousands of old articles in 2023, Google said plainly that removing content just because it’s old doesn’t help on its own.
Can we stand behind the answer, clinically and legally
Just because people search for a health-related topic doesn’t automatically mean an organization should publish content about it. Wrong or incomplete health information affects real decisions, which is why Google’s Search Quality Rater Guidelines give the closest scrutiny to Your Money or Your Life, or YMYL, topics, and health sits squarely inside that category. The September 11, 2025 update refreshed the YMYL definitions and added examples covering newer features like AI Overviews.
That responsibility no longer stops at an organization’s own page, either. In January 2026, following an investigation by The Guardian, Google removed AI Overviews for some queries about normal ranges for liver tests, because the summaries hadn’t accounted for differences such as age, sex, ethnicity, and nationality. For the same queries, the featured snippet at the top of the results had been pulling its reference ranges from a private hospital chain’s page in India. Google maintained that in many cases the information itself wasn’t inaccurate. The lesson still stands: a table without the right context, published on a single hospital page, can end up surfacing at the very top of the results for everyone.
That’s why certain topics cannot be treated as SEO exercises alone. These include treatment suitability, risks and alternatives, medication use, diagnostic criteria, success rates, and recovery expectations. These pages need traceable sources, clinical sign-off, and regular updates. The editorial workflow follows from that directly. Writers make the information understandable, a qualified physician checks its medical accuracy, and editors review clarity, sourcing, and any required disclaimers.
Local regulation shapes this stage as much as Google does. In Türkiye, a regulation on promotion and information in health services, in force since November 12, 2025, includes provisions that directly affect content planning. Only health professionals qualified in the relevant field may provide information about health services. Content may not steer patients toward a specific physician or facility, directly or indirectly. Prices, discounts, campaigns, and promotions may not be mentioned. The keywords used on search engines, and the information shown on results pages, must also comply. Websites must clearly show when information was last updated and how to reach the site’s editor. Rules differ for platforms aimed at international patients specifically, so health-tourism content needs its own separate review, and none of this substitutes for an actual legal check before acting on it.
Put simply: before asking whether a topic will bring traffic, the team asks whether it can take editorial and legal responsibility for publishing on it at all.
Stage 2: What kind of page it should become
Once a page has earned the right to exist, two questions shape what it actually becomes.
Which page type it belongs to
The same topic doesn’t always call for the same kind of content. A symptom page and a condition page do different jobs, and so do a treatment page and a department page. Someone searching for knee pain may not know the cause yet, so their page needs to explain possible causes and when to see a doctor. Someone researching a diagnosed condition is asking different questions entirely: what it is, how it’s diagnosed, which treatment suits them, and what the risks and alternatives are.
In practice, that means working with a small set of recurring page types, each with its own job: symptom pages (explain possible causes and when to seek care), condition pages (explain the condition as a whole and bridge to diagnosis and treatment), diagnostic test pages (explain why and how a test is done and how to read results), treatment pages (compare options, suitability, risks, and alternatives), medication pages (explain use, side effects, and warnings), and department or service pages (show which conditions a department handles and how to reach it).
Telling these apart isn’t always obvious, so a few boundary rules handle the gray areas. For content classification purposes, a symptom may be assigned to a condition-style page when it has a distinct management or treatment pathway, as can be the case with hearing loss. Symptoms such as chest pain, where treatment depends on identifying the underlying cause, are generally better suited to symptom pages. If a procedure’s main purpose is diagnosis, it gets a diagnostic page; if its main purpose is treatment, it gets a treatment page. A recognized, formal treatment method gets a treatment page, while practical day-to-day advice becomes lifestyle content instead.
Major health libraries draw the same lines. Mayo Clinic covers migraine across separate pages for symptoms and causes, diagnosis and treatment, and doctors and departments, and our boundary rules were checked directly against the library structures of Mayo Clinic and Cleveland Clinic.
Can the page’s purpose fit in one sentence
This is the simplest test in the whole process: can the page’s purpose be written down in a single sentence? “This page covers migraine” isn’t a purpose, it’s a topic. “This page helps someone looking into a migraine diagnosis understand what will happen during their evaluation” is a purpose. Once that purpose is clear, it shapes the headings, the sources, the internal links, and even the call to action.
If a page’s purpose can’t fit in one sentence, the page isn’t ready to be written yet. Google’s own quality rater guidelines judge page quality partly by how well a page achieves its purpose, which is exactly why the purpose statement, not the keyword, sits at the center of every content brief on our team. A typical brief for a migraine diagnosis page would define the purpose statement itself, the intended reader and their decision moment, and the specific questions the page needs to answer. It would also define what’s explicitly out of scope (treatment options, linked out to the treatment page instead), the evidence level required, and who the clinical reviewer is. Finally, it would set the internal links in and out, the call to action, the success metric, and the review cycle and owner.
Stage 3: Priority, measurement, and ownership
Once a page’s purpose and format have been defined, three final questions determine its priority, performance metrics, and long-term ownership.
What makes it a priority beyond search volume
High search volume and high priority aren’t the same thing. A topic can be searched heavily and still sit far from an organization’s real expertise, already be covered by an existing page, or lack the clinical support a safe answer would require. The reverse is just as true: low-volume searches about treatment suitability, preparation, or alternatives often land on a patient’s single most critical decision.
That’s why other signals get weighed alongside raw search data: Search Console queries, on-site search, paid search terms, the questions call center and scheduling teams hear most often, and the questions physicians answer most often in the exam room. A simple scoring model helps weigh them together, scoring each topic on demand, decision-moment criticality, fit with the organization’s actual expertise, the size of the existing coverage gap, and how manageable the clinical and legal risk is.
Regulation affects priority directly, too. Price and campaign-related queries can carry real search volume, yet in Türkiye they can’t be answered with price information for domestic audiences at all, no matter how high they rank on a priority list. AI search shifts the balance further: per Google’s guidance on generative AI search, AI features can break a question into subquestions and match pages by meaning even when the exact query never appears on the page. So instead of a thin page built for every subquestion, one strong page that does its job well, supported by a handful of linked pages, tends to be worth far more.
What success actually looks like for this page
If a page’s reason for existing is clear, its definition of success should be too; not every page should be judged by form fills or raw organic traffic. A symptom page succeeds if it ranks under the right URL and readers move on to the related condition or department page. A condition page succeeds if readers find what they need and continue toward diagnosis or treatment. A treatment page should help readers understand the risks and alternatives while reducing inquiries from patients for whom the treatment may not be appropriate. A department or service page succeeds if inquiries reach the right department. And across every page type, the ongoing question is whether it’s competing with another URL on the site and whether it’s still up to date.
Visibility in AI answers is now part of that picture as well: Google recommends its generative AI performance report in Search Console specifically for tracking how content performs inside its AI features.
Judging an informational page only by how many forms it generates slowly changes what the page is actually for. CTAs added purely under conversion pressure can turn a page meant for someone still gathering information into a sales page, and in Türkiye there’s a legal limit here too: content may not steer patients toward a specific physician or facility, directly or indirectly. So the call to action gets matched to the page’s actual purpose. On an informational page, the next step is another informational step. On a department page, the department’s contact details come first.
Who is responsible for it six months from now
A content decision doesn’t end at publication, it lasts as long as the page stays live. Mustafa Çelik sums up the whole system in a single question:
“Who will be responsible for making sure a page we publish today is still accurate six months from now?”
That question is what turns a publishing calendar into an actual ownership system. Before publication, every page needs four things defined: an owner, a clinical reviewer, a review cycle, and a change log, with review frequency following risk, so topics like medication use and treatment suitability get checked more often than lower-stakes pages. In Türkiye this isn’t optional either; the same regulation requires websites to clearly show when information was last updated and how to contact the site’s editor.
Authorship has to be real, too. In October 2026, Google added an explicit warning against deceptive authorship information to its helpful content guidance. It specifically names made-up author names, AI-generated headshots, and false credentials used to make content look expert-written as a form of deception that makes a page untrustworthy. Good examples put this transparency directly on the page. Mayo Clinic’s migraine page clearly shows the medical editorial team behind it, the page’s date, and its references.
Health information even has its own independent quality standard outside of Google entirely. In the UK, the PIF TICK assesses producers of trusted health information against ten criteria, including a documented process, meeting a genuine need, up-to-date evidence, user feedback, and measuring impact. One organization that helped update those criteria shared a familiar lesson: trying to measure impact sent them back to the purpose of each resource, which is the same discipline this whole framework is built around.
Running a healthcare content marketing system at scale
Asking eight questions about a single page is straightforward. Applying them consistently across thousands of healthcare pages requires turning those questions into a structured workflow rather than relying on a mental checklist.
When Magnero’s team takes on a large hospital content archive, these eight questions get turned into a classification and brief system with a few core rules. Every topic gets classified by page type before anything gets written, with ambiguous cases routed to a human reviewer and topics that don’t belong in a health library routed to the general blog instead. The decision gets made once per parent page, and child pages inherit it, which turns thousands of individual URLs into a much smaller number of real decisions. Each page type carries its own required set of sections, such as emergency warning signs and “when to see a doctor” guidance on every condition page; comparing our own structure against competitors’ is actually how the team caught that condition pages were missing a “living with the condition” section, which then got added across the board.
Structured data templates do double duty in this system. They aren’t used only for markup, they also decide upfront which information belongs in a table, a list, or plain text. Every new page gets checked against the site’s existing blog archive for cannibalization before publishing. That check reads actual titles, not just assigned target keywords, since many older posts never had a target keyword properly set in the first place. Every conflict the check turns up gets one of four actions: a 301 redirect, retargeting the page’s query, marking it as a content donor, or closing it out as a false alarm, and none of those four gets applied automatically; each needs a human sign-off. Critical errors stop publication outright; warnings go to an editorial list for later review instead.
The title-based check matters more than it sounds. If a planned section is called “early signs of breast cancer” and the blog archive already has a post titled “4 early signs of breast cancer to watch for” with no target keyword ever filled in, a check that only reads assigned keywords will miss that conflict completely. A title-based check won’t.
Behind every one of these rules sits one underlying principle: a rule has to be concrete enough for a non-expert to apply without needing to interpret it. “Determine the page type” isn’t a usable rule. “If the answer to this specific question is yes, choose this type” is. The sequencing is deliberate too: existing content gets classified, rewritten, and redirected first, and only then does the team move on to genuinely new topics, so new pages are never born already competing with old ones.
What this framework has in common with connected content architecture more broadly
This eight-question system for healthcare didn’t emerge in isolation. It reflects a broader shift already underway in how content strategy gets practiced across sectors, one Mustafa Çelik lays out in a recent LinkedIn article on building connected, maintainable information architecture:
“Topical authority, as he puts it, is a capability a team builds and maintains, not a score a plugin calculates or a folder structure on a server.”
That framing matches almost exactly what sits underneath this healthcare system. A hub page should orient a reader toward a real next step. Internal links should connect genuine decisions, not just repeat a shared keyword. And a page’s structured data should describe what it actually, visibly contains, since even accurate markup never guarantees a rich result on its own, a point Çelik,
“Makes directly in discussing the retirement of FAQ rich results from Google Search as of May 7, 2026.”
The same article makes a point directly relevant to the authorship and review-cycle questions above: E-E-A-T, shorthand for experience, expertise, authoritativeness, and trustworthiness, isn’t a single ranking factor or a score, it’s Google’s own description of signals it looks for, weighted most heavily on exactly the YMYL subjects where mistakes could affect health, safety, or financial stability. An author bio, a citation, and a review date only mean something when they describe work that actually happened. “Expert reviewed” has to be a true claim about a real process, not a label added for search.
FAQ
How many pages should a hospital or clinic website have for one condition?
There’s no fixed number. It depends on how many genuinely different decisions the condition involves, understanding it, diagnosing it, treating it, living with it, each of which may deserve its own page, or may not if the search volume and distinctness don’t justify splitting it.
What’s the difference between a symptom page and a condition page?
A symptom page addresses something whose treatment depends entirely on an underlying cause (chest pain, for example). A condition page addresses something with its own defined treatment path independent of a separate root cause (hearing loss, for example).
Does every piece of healthcare content need a clinical reviewer?
Content touching diagnosis, treatment suitability, medication, or risk needs clinical sign-off. General awareness or administrative content (like appointment logistics) may not, but the decision should be made deliberately, not by default.
How often should healthcare content be reviewed for accuracy?
Review frequency should follow risk, not a flat schedule. Medication and treatment-suitability content typically needs more frequent review than lower-stakes informational pages.
Can AI safely help write healthcare content?
It can help with drafting and structure, but every material claim still needs human verification. Google’s own guidance now explicitly extends that fact-checking requirement to titles, meta descriptions, structured data, and alt text, not just body copy.
The real question behind every page
At the start of this process, our team had thousands of keywords in front of it, which looked at first like thousands of separate content opportunities. As the work went on, it became clear that thousands of keywords never meant thousands of pages. The real challenge was understanding the intent behind each search, rather than focusing solely on search volume.
Today, every new healthcare content idea runs through the same eight questions. What decision is the searcher trying to make, and do we already answer this? Can we stand behind the answer clinically and legally, and which page type does it belong to? Can we state the page’s purpose in one sentence? And finally: what makes it a priority beyond search volume, what does success look like for this page, and who is responsible for it six months from now?
These questions turned a publishing calendar into a list of decisions, each with a reason behind it. The goal was never to publish more health content more often. It’s being able to explain, for every single page, why it exists at all. That’s where choosing the right healthcare topics actually begins: not with finding the keyword, but with proving the page needs to exist in the first place.
