Medical pages age in two different ways, and teams usually treat them as one. A page can lose clicks without a single fact on it becoming wrong. And a page can keep its rankings for months while the guideline it summarises has been revised, the drug it names has been withdrawn, or the price it states is two years old.
This article is for marketing and content leads at hospitals, clinic groups and health publishers who maintain hundreds of clinical pages and have to decide which ones deserve a refresh. It separates traffic decay from accuracy decay, gives a trigger list for when a medical update justifies work, sets out a triage for how deep a refresh should go, and describes how to measure whether a refresh recovered clicks, using a before-and-after design with a control group. The measurement method is the part most teams skip, and it is why many refresh programmes cannot say whether they worked.
No results are reported here. The method is for you to run on your own Search Console data. Where a page's clinical content changes, the reviewing clinician owns the decision about what the new text should say.
Two kinds of decay
Traffic decay is a measured fall in clicks or impressions for a page over time. Causes are many: competitors publishing better answers, seasonality, a core update, a change in how results are displayed, a lost link, or a technical fault.
Accuracy decay is a gap that opens between what the page says and what is currently true or recommended. Causes are external events: a guideline revision, a regulatory notice, a new approval, a price change, a change in how your own department works.
They overlap, but they need different responses. Accuracy decay is a patient-safety and trust issue and justifies a refresh regardless of traffic. Traffic decay is a performance question and justifies a refresh only if you can identify a cause the refresh addresses.
Be careful not to blame decay for what is really a change in the results page. A study by Ahrefs found that the click-through rate for the top-ranked result was 58% lower when an AI Overview was present. If a page's impressions hold steady while clicks fall, check whether an AI answer now appears for its queries before you rewrite a perfectly good page. Our article on AI Overviews and healthcare SEO looks at that shift in more detail.
The trigger list: when a medical update justifies a refresh
Write the triggers down and assign owners. Otherwise refreshes happen when someone remembers.
| Trigger | Where to watch | Typical page types affected |
|---|---|---|
| Guideline revision by a national or international body | ICMR, MoHFW and WHO publication pages; specialty society sites | Condition, screening, treatment pages |
| Drug approval, withdrawal or safety notice | CDSCO notices | Drug, treatment and side-effect pages |
| Change in a government scheme or insurance coverage | Scheme websites and insurer circulars | Cost, insurance, eligibility pages |
| Regulatory or advertising rule change | NMC and other regulator notices | Doctor bios, service pages, claims |
| New evidence that changes advice | Journals, society statements | Any page citing the old evidence |
| Price change | Billing and finance | Package and cost pages |
| Service change | Department heads | Service-line, location, booking pages |
| Clinician change | HR and department heads | Bios, reviewer blocks, doctor lists |
| Equipment or facility change | Operations | Treatment and technology pages |
The "where to watch" column should name a person and a frequency. A monthly check of a short list of regulator and guideline pages by one named person costs less than one avoidable inaccuracy. Link each watchlist entry to the pages that rely on it by using the sentence-to-source map described in our patient education content strategy work, so a trigger immediately produces a list of affected URLs.
Triage: how deep should the refresh be?
Not every trigger needs a rewrite. Classify the change before touching the page.
| Class | Example | Action | Review needed |
|---|---|---|---|
| A. Clinical change | Guideline revised; drug withdrawn | Rewrite affected sections; full review | Clinician plus source check |
| B. Factual change | Price, address, hours, doctor list | Targeted edit | Owner of the fact |
| C. Reference change | Link moved; source replaced with an equivalent | Update citation | Editor |
| D. Cosmetic | Reworded for style, no change in meaning | Do nothing | None |
Class D is where programmes waste money. Rewording a page for its own sake changes nothing for the patient and may disturb a page that was working. If the only reason to refresh is that the page is old, check whether its facts and sources still hold. If they do, record a review and leave the content.
The honest dates matter here. Google's guidance on byline dates says its systems estimate when a page was published or significantly updated, and recommends that visible and structured dates match. The sitemap documentation likewise says the lastmod value should reflect the last significant update, such as a change to main content, structured data or links. Do not change either for a class D edit. A reviewed date is a separate field: it can move when a clinician confirms that the page is still correct, even if no text changed.
Designing the before-and-after measurement
If you refresh twenty pages and then compare their clicks to the previous month, you will learn very little. Seasonality, algorithm changes, SERP changes and ordinary variance all affect the result. A defensible design does four things.
1. Define the unit and the window before you refresh. The unit is the page (canonical URL). Decide the pre-period and post-period lengths in advance, for example 28 days each, so you cannot choose a flattering window afterwards. Start the post-period when Google has recrawled the page. You can check the last crawl date in the URL Inspection tool and record it per page.
2. Add a control group. For each refreshed page, choose one or more similar pages that you did not refresh, matched on topic area, page type, age and prior click volume. The control tells you what happened to similar pages over the same dates, which strips out most seasonality and many search-wide changes.
3. Choose metrics that separate causes. Pull from the Performance report at page level and query level:
- Clicks and impressions. Clicks are the outcome; impressions show whether visibility changed.
- Click-through rate. If impressions rise and clicks do not, the snippet or the results layout, not the content, is the issue.
- Average position for non-brand queries. Treat it as directional. The average blends different queries and devices.
- The query set. Count the queries that earn impressions. A refresh that broadens the answered questions shows up here before clicks.
4. Exclude disturbed periods. Annotate the dates of Google's ranking and core updates and any site changes such as a migration, template change or robots.txt edit. If a core update lands inside either window, either move the window or flag the result as contaminated.




