There is no healthy percentage. A single page carrying most of your organic search arrivals is only a problem when nothing else on the site could absorb the loss, and that is a question about your library, not about the number.
Why does a percentage threshold not exist?
Because the right shape depends on what the site is for. A brand with one famous product should see its homepage and that product page dominate. A publisher with six hundred articles should not see one of them take half the traffic.
The distribution of search demand itself is lopsided. A handful of queries in any subject attract most of the searching, and the page that answers the broadest of those queries will naturally outrank the rest of your library in volume. Concentration is the normal output of a working site, not a symptom.
What changes the reading is replaceability. If your top page lost its position tomorrow, would the queries it serves land on another page of yours, or leave the site entirely? That is the only version of the question with an actionable answer.
What actually goes wrong when one page carries the site?
The failure is rarely gradual. A page that holds a dominant position tends to lose it in one step, because the things that move it are discrete events rather than slow drift.
A core update reweights what a query rewards. A new search feature appears above the results and takes the clicks without taking the ranking. A competitor publishes something better aimed at the same intent. The template the page sits on ships a bug and the page drops out of the index for a week. In each case a site with distributed traffic loses a slice, and a site with concentrated traffic loses its quarter.
There is a second-order cost that shows up later. When one page supplies most of the arrivals, your whole funnel is tuned to that page's audience, and your site-wide conversion rate is really that page's conversion rate wearing a disguise. Decisions made from the blended number get applied to pages that behave nothing like it.
Concentration is not a number to fix, it is an exposure to price. The useful question is not "what share does my top page hold" but "what happens to revenue if it halves," and the answer to the second one tells you whether the first one matters.
How do you measure concentration without fooling yourself?
Pull clicks by page from the performance report in Google Search Console over a full year, not a month, and sort descending. A year flattens seasonality and campaign spikes that make a normal distribution look dangerous or a dangerous one look fine.
Two traps are worth knowing before you read the export. Search Console reports clicks by page accurately but withholds rare queries for privacy reasons, so the queries listed under a page will not sum to the page's own total. Use the page totals for concentration work and the query lists only for understanding what the page serves. The second trap is analytics landing-page data, which blends branded navigation and campaign arrivals into the same rows as search; a homepage looks far more dominant there than it does in search specifically.
Then run the subtraction. Remove the top page's clicks and look at what remains. Remove the top five and look again. A site where the top five pages are the whole business has a different problem from one where the top page is twice the second but the tail is deep.
Which kinds of concentration are fine and which are fragile?
| What you see | Read it as | Action |
|---|---|---|
| Top page is the homepage, mostly branded queries | Demand for the brand, not a ranking dependency | Leave it; track branded and unbranded separately |
| Top page answers the single broadest query in your subject | Earned concentration with real downside | Build the sibling pages that could catch the overflow |
| Top page is a free tool, calculator or template | A durable asset, but one with no substitute on your site | Accept the exposure, monitor it weekly |
| Top page ranks for queries unrelated to what you sell | Volume without a next step | Stop counting it as a dependency; it is not load-bearing |
| Top page is dominant and the tail is almost empty | A one-page site with extra files | Treat library depth as the priority, not the page |
| Top page's share grew while total traffic fell | The rest of the library is decaying | Audit the tail before touching the winner |
How do you spread the load without damaging the page that works?
Build outward from the queries the winner already attracts, not on top of them. The goal is pages that catch adjacent intent, so a position change on the dominant page costs you one slice of a cluster instead of the cluster.
Read the query list under your top page and find the questions it answers in a sentence but does not really serve. Those are the sibling pages. Each one should take a question the winner mentions in passing and answer it properly, then link back up to the winner for the broad version. That arrangement keeps the broad page as the strongest candidate for the broad query while giving the narrow queries somewhere better to land.
What not to do is rewrite the dominant page to chase more queries, or split it into several pages to look more diversified. Both trade a known position for an unknown one. Diversification that destroys the asset is not risk management.
Publishing near-duplicates of the winning page to reduce dependence on it. Two pages aimed at one intent do not halve the risk; they compete with each other and often cost you the position you were protecting.
What is the decision in front of you?
Run the subtraction on a year of Search Console data and put a revenue number on losing the top page. If that number is survivable, concentration is a reporting curiosity and you should spend the quarter on whatever actually grows. If it is not survivable, the work is not on the winner at all: it is on the three sibling pages that do not exist yet, and the clock on them started before you read the report.