Which Pages to Build Links To: Diagnose Before You Pitch

Build links to a page only after you know why it isn't ranking. "Not ranking" isn't one problem. A page that's six weeks old, a page answering a question hardly anyone searches for, and a page your own site barely links to all look the same in a traffic report: flat line, no clicks. They need opposite responses, and an external link is the right answer for only some of them. Before any page goes on an outreach list, ask four questions: can you read its results yet, is there real demand for what it answers, where does it sit right now, and how many of your own pages link to it, counted properly. The answers decide whether the page needs links from other sites, links from your own, a rewrite, or to be left alone.

Why "build links to the pages that aren't ranking" sends effort to the wrong place

The standard advice goes something like this: find the pages that matter to the business, see which ones are underperforming, and point your link building at them. It sounds disciplined. It treats importance plus underperformance as a diagnosis.

It isn't one. Importance tells you what you'd like to rank. Underperformance tells you something is wrong. Neither tells you what, and link acquisition is the most expensive fix in the toolbox to apply to the wrong problem. A placement on a respected site can take weeks of pitching and writing. Spend that on a page whose real trouble is that hardly anyone searches for its topic, and you've paid full price for nothing that can show up in the numbers.

We run a triage on our own site every few weeks, and the rule at the top of it is that no page gets any action until it's been sorted into a bucket. Not an update, not new links, nothing. That rule exists because the traffic report can't tell these cases apart and the fixes pull in different directions.

Question one: can you read the result yet?

Our cutoff is 90 days. Anything younger is too young to judge, and we don't touch it, because changing a page resets the clock you're using to evaluate it. At our domain rating, which sits around 30, we plan on three to six months for a page to find its level.

That matters for link building in a specific way. If you point external links at a page in its first few weeks and it climbs, you can't tell whether the links did it or the page was simply settling in. If it doesn't climb, you can't tell whether the links failed or it's just early. Either way the campaign teaches you nothing, and the next time you decide where to spend outreach effort, you're guessing again.

The exception is your own internal links. Adding links from related pages on your site is the one fix we treat as safe on a page of any age. It's cheap, it's reversible, and it doesn't cost you a placement you could have spent elsewhere.

Question two: is anyone actually looking for this?

This is the bucket that eats link budgets. If the phrase a page targets has close to zero search volume and the page has close to zero impressions, links won't create the demand. Age won't either. We reclassify those pages as support content, meaning they stay live as supporting material that other pages link to and that AI search tools can draw on, and we stop counting them as failures.

Two cautions, one of them learned by getting it wrong.

First, measure it. In one pass we labelled a comparison page no-demand without measuring and moved on. The next pass pulled the actual number: 1,900 searches a month, plus another 1,000 on a reversed wording. The rule was fine. We'd skipped the measurement. The rule now says every page headed for the no-demand bucket gets its volume measured in that run, even pages an earlier run already filed there.

Second, volume isn't the whole test. A phrase can carry searches and still be the wrong target for your page, because the results page shows the searcher wants something else, like a software product when you've written an explainer. Pull the results page before deciding. If the top ten are all a kind of page you didn't build, links to your page are pushing against what the searcher came for.

Question three: where does the page sit right now?

Look at average position and impressions over a 90-day window. Two patterns matter.

The first is a page sitting somewhere between position 5 and 30 with at least 10 impressions. In our triage that's the striking band, and those pages get flagged as update candidates, because they're already showing up for something real and losing on the margin. My own view is that this is also where an external link has its best odds. The page has proven relevance. It's close. A small push has somewhere to go.

The second is a page over six months old where demand exists but it sits below position 30, or barely gets impressions. We call that stalled, and the diagnosis runs in a fixed order:

• Re-pull the results page. Is the top ten rewarding a different type of page than the one we built?
• Are the title and meta description correct?
• Does the page have at least two links from our own editorial content?
• Is there a variant of the phrase with real demand that's a better target?

Links are third on that list, and they're internal links. A stalled page that loses on page type won't be rescued by a placement, because the thing it's losing on isn't authority.

Question four: how many of your own pages link to it?

This is the question most link programs skip, and it's the cheapest one to fix. Our threshold is two. A page with fewer than two inbound links from genuinely related content on our own site is an orphan, whatever its age, and it gets flagged for new links from related pages on our own site.

The hard part is counting honestly. Strip out navigation, headers, footers and index pages first, because a link that sits on every page tells a search engine nothing about this one. Then look for anything else that repeats sitewide.

We learned that one directly. Our blog template has a section labelled as related posts, and it turns out it isn't topical at all. It shows the three newest posts on every article. When we counted it as editorial linking, the three newest posts appeared to have 48 to 52 inbound links each. With that section stripped out, the real counts were 1, 5 and 2. One of them was an orphan hiding behind a module that looked like internal linking and wasn't, and another sat right on the line.

Two more lessons from actually doing the fix:

• A page with zero internal links needs two, not one. In our first round, two pages that had none got a single new link each. The change description said 11 pages had been brought up to two. A recount of the live site found two of them still at one. Count the gap per page before you start adding.
• A link only counts if it renders. That same recount turned up something worse. Links written into certain content fields, FAQ answers among them, were showing on the live page as literal brackets and text instead of working links, because those fields skipped the step that turns them into anchors. Verify by crawling after the change goes live, not by reading what you intended to do.

So which pages should get external links?

Run the four questions and the outreach list gets short, which is the point.

• Under 90 days old: internal links only. Wait for a readable result before spending placements.
• No measurable demand, or the results page wants something else: no links. Reclassify it as support content and let it do that job.
• Stalled with a page-type or title problem: fix that first. Links come after.
• Fewer than two internal editorial links: fix that first too. It's an edit on your own site, not a pitch.
• Old enough to read, real demand, sitting in the 5 to 30 band, already linked from your own site: this is the page that's earned a place on the outreach list.

That last group is usually a handful of pages, not dozens. That's fine. A small list of pages where a link has a clear job beats a long list where most placements land on pages with a problem links can't touch.

It also changes how you talk to publishers. When you know why a page is on the list, you know what kind of mention would help it, and you can pitch the article or answer that fits, rather than asking for a link to whatever page matters most to you this quarter.

Common questions

Should I stop building links to my homepage and service pages?

No. Those pages usually carry your brand searches and the bulk of any mentions you earn naturally. The point is narrower: don't pick them for a campaign because they're important. Pick them because the four questions say a link is the fix. A service page with the wrong page type for its results page needs rebuilding before it needs authority.

How long should I wait before judging whether a link helped?

Longer than most reports assume. We don't judge a page younger than 90 days at all, and on a site with a domain rating around 30 we plan on three to six months for a page to settle. Change one thing at a time where you can, so that when the page moves you know which change moved it.

Aren't internal links a weaker signal than backlinks?

They're a different lever, and on an orphan page they're often the first one that matters. A page your own site doesn't link to is hard for search engines to find and easy for them to discount. Fixing that costs very little, and it means any external link you add later is working on a page that's already connected to the rest of your site.