SEO Galaxy is now GetMentioned. The platform stays the same. Want a demo? Book your slot here →
Content decides whether a page enters the running. Links decide which of the qualifying pages ranks first. How to tell which half you are missing, with the median figures from the last Semrush study and five common claims checked.
David Hahn · August 19, 2026 · 13 min read
The short version
The question sounds like a matter of principle. In practice it is about money.
The quarterly budget is fixed, and it covers one of the two.
Both are parts of the same search engine optimization. Content is built on your own web page, backlinks are built on other people's.
As a principle it has no answer. Both work at different points of one process.
Content answers what a page is about. It decides whether a search engine considers it for a query.
Without matching text there is nothing for a link to point at.
Links answer how others rate the page. Backlinks, also called inbound links, are references from other web pages and work as a trust signal at second hand.
They separate pages that all fit the query.
That is where the gap between position 3 and 14 opens up.
So a different question matters. Which of the two parts is missing on your site?
Content decides admission. Backlinks decide the order.
A page will not rank for a keyword it does not cover, even with a hundred links.
A page with excellent content stays behind weaker pages carrying more recommendations from elsewhere.
Three rules follow from that:
Everything beyond those three is settled by the competitors.
The most solid analysis comes from Semrush. It covered 16,298 US keywords and mobile results to position 20.
That is 300,000 organic positions, excluding Wikipedia and YouTube.
Collected in 2023, published in January 2024. No follow-up edition exists.
The interesting part is not the ranked list of factors. It is the benchmark table at the back.
That table puts top 10 pages and positions 11 to 20 side by side.
| Median metric | Position 1 to 10 | Position 11 to 20 |
|---|---|---|
| Text relevance to the query | 94 % | 85 % |
| Content quality score | 85 % | 78 % |
| Referring domains to the URL | 6 | 2 |
| Backlinks to the URL | 13 | 4 |
Source: Semrush, Ranking Factors Study 2024, collected 2023, last edition of the series.
Every page in that table sits in the top 20. All have cleared the content bar.
Inside that group the text metrics are 7 to 9 percentage points apart. The link metrics differ threefold.
That is the argument in numbers. Content puts a page in the running, and links sort the running.
Semrush also names text relevance the factor correlating most with high positions. Both findings fit together.
The diagnosis needs no special software. Search Console and three competitors are enough.
Content is the bottleneck when one of these applies:
The last point is most common. Many teams want a ranking for a topic with no page of its own.
A placement achieves nothing there. A link to a page without matching content amplifies a signal that is absent.
A page sits at position 30 and stays there. That gets read as a link problem.
Position 30 usually means Google knows the page and rates it a weak match.
The distance from position 30 to 10 is rarely link count. It is usually search intent.
The pattern is the reverse and easier to spot. The page ranks and stops short of the top.
Links are the bottleneck when one of these applies:
In our experience the second point is clearest. When a weaker page outranks you, the reason is rarely editorial.
The link gap compares the referring domains of your landing page with the three pages above you.
What counts is the number on the single URL, not on the whole domain.
Setting that comparison up is covered in our piece on backlink analysis.
With the bottleneck settled, the next question follows. How many links, and which ones.
The Semrush figures give an order of magnitude. The median across the top 10 is six referring domains per URL.
Six fitting links do more than sixty arbitrary ones. The reason sits in the arithmetic of every authority metric. A domain spreads its weight across every target it links to.
The influence of a single link also depends on the trustworthiness of the source. A domain that covers every topic and links in every direction carries little.
High quality backlinks share three traits:
How many links a landing page needs depends on the industry. Five carry a niche topic, fifty are few in a contested finance topic.
The risk sits in the distribution. A link pattern stands out when a hundred links come from one corner and all carry the same anchor text.
The timeline stands out too. A link profile that jumps from two to two hundred domains in four weeks looks bought.
Better: the build spreads across months, the sources come from different corners of the same field, and the anchor texts alternate between brand, address and descriptive wording.
Link building strategies differ by the reason somebody links at all.
| Method | Why the link happens | What it costs |
|---|---|---|
| Trade article in an industry outlet | editorial context | money and coordination |
| Your own analysis or infographics | the figure exists nowhere else | working time, the timing stays open |
| Mention without a link from digital PR | the brand is part of the story | money, effect mostly in AI answers |
Content marketing and link building work into each other here. A company that publishes its own figures gives other newsrooms a reason to link.
Sentences circulate around this keyword that cite each other. Five fail against the primary source.
The word "strongly" does not hold. Semrush measures 0.21 for its domain authority score and 0.19 for the page score, then calls that "strong".
On the Spearman scale a value near 0.2 is weak.
Then there is the direction problem, which Semrush raises itself. Pages ranking higher collect more links because they are visible.
Whether links produced the ranking or the reverse, none of these studies measures. More checked link building data sits in a separate piece.
Paid placements are a legitimate part of authority building when source, topic, landing page and context line up.
Google's spam policies define link spam by purpose. The target is links created primarily to manipulate search rankings.
The same document notes that a paid link stays within the policies as long as it carries rel="nofollow" or rel="sponsored".
Prohibition and exception share one page.
The claim holds for link packages, link networks, off-topic domains and over-optimised anchors. For an editorial placement in a fitting publication it does not.
Half right. Google's spam update documentation says its systems remove the effect of such links and any earlier ranking benefit is gone.
The second half sits next to it. Violations are caught by automated systems and by human review.
A manual action can push a site down in the results. Bad links are not free of consequences.
No study with a published method exists. The sentence compares two cost types that cannot be weighed without context.
An expert article costs working hours, works slowly and works across a topic cluster.
A placement costs money, works over months and works on one URL.
Which calculation comes out ahead depends on which part you are missing.
True, and not news. Google has never claimed there is a single ranking factor.
The sentence reads like a discovery because it refutes a position nobody holds.
The useful question is the role links play in ranking. They act where several fitting pages compete for one position.
| Claim | Verdict |
|---|---|
| Backlinks correlate strongly with traffic | fails, published values run 0.19 to 0.30 |
| Buying links violates Google's guidelines | holds only for links meant to manipulate rankings |
| Google devalues spam links automatically | half holds, manual actions still exist |
| Excellent content gives a better ROI | fails, no study measures this |
| Backlinks are not the only ranking factor | holds, and never was otherwise |
AI answers add a third quantity. Mentions without a link.
Ahrefs studied 75,000 brands and checked which factors line up with a brand named in AI Overviews.
The filter was Domain Rating above 40 and keywords with at least 800 monthly searches.
The highest value goes to web mentions at 0.664. Referring domains reach 0.295 and backlinks 0.218.
Mentions therefore sit around three times higher than links.
The author qualifies this herself. Every factor shows moderate to very weak correlation, and correlation is no cause.
The split between content and links falls short once AI answers are part of the goal.
A language model needs evidence from outside sources. A plain mention works as evidence too.
How the two combine is covered in our piece on mentions and link building.
Diagnosis is the easier half. Sourcing comes next.
This is text on your own domain. The result is checkable on four points:
We write and deliver those texts, including for our clients' own pages. Publishing stays with the client.
This is about pages you do not own. A good placement shows five marks:
Those figures sit side by side at GetMentioned before you book a placement.
If you want to hand the build over, we run it as a link building agency in full service. That covers link building and mentions end to end, with no full service SEO on your website.
In the conversations we have, the text is rarely the problem. Far more often the page has no recommendation from anywhere, and that changes faster than a relaunch.
There is no fixed ratio. There is a mapping from starting position to focus.
| Starting position | Bottleneck | Where the budget goes |
|---|---|---|
| Page ranks for no keyword in the topic | Content | text, structure, coverage |
| Page held position 6 to 20 for months | Links | placements on that one URL |
| Competitors with thinner text rank above | Links | closing the link gap |
| Brand never appears in AI answers | Mentions | digital PR and brand mentions |
| New revenue page with no referring links | Both | text first, placements after |
Our experience says teams overlook the second row the longest.
A page at position 9 gets rewritten for years while the text was fine from day one.
How that becomes a repeatable plan is covered in our piece on backlink strategy.
Three things stick. Content decides admission to the running.
Links decide the order inside it. In AI answers, mentions without a link arrive as a third quantity.
For your own website the question takes ten minutes. If the page does not rank, content is missing. If it ranks and stalls, signals from other websites are missing.
SEO success then rests on the order, not on a ratio. Buy the wrong half first and you pay twice.
Each decides a different stage. Content puts a page in the running. Backlinks decide which fitting page ends up on top.
On a new page with content, because a link to an unfitting page points nowhere. On a page at position 6 to 20 with links.
In low competition niches, yes. The median for top 10 pages is 6 referring domains per URL (Semrush, collected 2023). Half get by on very few links.
There is no general number. What counts is the referring domains of the three pages ranking above you.
Google's policies target links meant to manipulate rankings. An editorially placed, topically fitting article falls outside that. Link packages, link networks and off-topic domains are the problem.
Yes, they sit behind mentions. Across 75,000 brands, web mentions correlated at 0.664 with being named in AI Overviews and backlinks at 0.218 (Ahrefs, 2025).
By bottleneck rather than by ratio. If your pages never surface for their topics, budget goes into content. If they surface and stall, into placements.
About the author
Managing Director, GetMentioned
David has been building link acquisition and digital PR processes since 2016, first as an agency under SEO Galaxy, today as a platform with GetMentioned. He has scaled his own projects from zero to seven-figure monthly traffic and delivered thousands of campaigns for clients. Here he writes about what works in practice, and about what only costs budget.
Compare publishers, SEO data and prices in one place and book the placements that fit.
How search engines and AI systems weigh links, which methods exist, what link building costs and how paid placements should be judged. The overview, with pointers into the detail.
August 12, 2026 · 14 min read
Verified link building statistics with source and year of collection. How strongly backlinks work, how many top-ranking pages hold, what placements cost and how common buying links is. Three widely quoted numbers failed the source check.
August 16, 2026 · 10 min read
How to evaluate a link profile: which data you need, which metrics say something, how the competitor comparison works and what comes out as a work list.
August 13, 2026 · 11 min read
