Guide · Apr 16, 2026 · 10 min read · by Cynthia Madison

How we evaluate a website before placing a link: the 9-point check

Most bad links don't look bad in a sales deck. The seller shows you a domain rating in the seventies, a screenshot of monthly traffic, and a clean homepage, and on paper it's a fine site. The problems live one layer down, in the shape of the traffic curve, the keywords the site actually ranks for, and the company it keeps. If you only look at the headline metrics, you'll buy links on sites that look healthy and are quietly rotting.

This is the checklist we run before any placement. It's nine checks, and none of them takes long once you know what you're looking at. The goal isn't a perfect score, real sites have flaws, it's to separate soft warnings you can live with from hard fails you can't. We'll cover each check, then how to add them up.

1. Organic traffic trend: look at the shape, not the number

A single traffic number tells you almost nothing. The shape of the trend over the last year or two tells you almost everything. Pull the organic traffic graph and ask what direction it's pointing.

Healthy sites tend to show a gradual climb, or a stable plateau with seasonal wobble. The patterns that should worry you are a sharp recent drop, traffic that fell off a cliff and never recovered, or a sudden vertical spike with no story behind it. A spike often means the site got a burst from something temporary, or worse, that the metric is being inflated. A cliff usually means the site got hit by an algorithm update or did something it shouldn't have. You want to place links on sites that are trending up or holding steady, not ones on the way down dragging your link with them.

2. Traffic-to-DR ratio: do the numbers agree with each other

Domain rating and organic traffic should roughly track each other. A site with a high authority score should be pulling meaningful organic traffic. When they diverge sharply, something's off.

The classic red flag is a high domain rating paired with almost no organic traffic. That combination usually means the authority was manufactured, the site bought a pile of links to inflate its score but never earned the rankings that real authority produces. A site with a rating in the seventies and a few hundred visitors a month is a site whose metric was gamed. The reverse, modest authority but strong, relevant traffic, is often a much better link than its score suggests. Trust the traffic over the rating when they disagree.

3. Do the ranking keywords make sense

This is the check that catches the most disguised junk, and almost nobody does it. Look at the actual keywords the site ranks for. They should match what the site is supposedly about.

A genuine cooking blog ranks for recipes, ingredients, and techniques. If you pull the keyword list and find it ranking for casino terms, loan offers, or pharmaceutical names in three languages, the site has been hijacked, sold off, or is running hidden content. Mismatched keywords are one of the clearest signs that a site isn't what it presents itself as. Spend two minutes on the keyword report; it exposes problems that a clean homepage hides.

4. Outbound link density

Open a handful of the site's articles and count how many external links point off-site, and where they go. A normal editorial article links out a few times to relevant sources. A page stuffed with outbound links, especially commercial ones to unrelated industries, is a link farm wearing an editorial costume.

What you're really measuring is whether this site sells links to everyone. If every other post has a paid-looking link to a casino, a loan site, and a supplement brand, your link is about to join a crowd, and the value of being in that crowd is close to zero. Sites that link out sparingly and relevantly pass far more value than ones that link to anyone with a budget.

5. Content quality on arrival

Read the content. Actually read two or three full articles, don't just scan the layout. You're looking for whether a competent human wrote this for readers, or whether it was churned out to fill space around links.

Signs of trouble: thin articles that say nothing, obvious filler padded to hit a word count, content that reads like it was translated badly, or a site where every post is the same length and shape because it's mass-produced. Signs of quality: genuine expertise, first-hand detail, articles of varying length because they were written to cover their subject rather than to a template. The principles in our piece on E-E-A-T signals you can actually build are the same ones to read for on someone else's site, since the qualities you'd build into your own content are exactly what marks a placement site as legitimate.

6. Editorial friction

How hard is it to get published here? This sounds backwards, but friction is a quality signal. A site that will publish anything, instantly, for a flat fee, with no questions about your content, is a site that publishes everyone's anything. That's not a publication; it's a vending machine.

Sites worth being on tend to have a process. They ask about your topic, they have content guidelines, they might push back on your draft or your anchor. That editorial gatekeeping is annoying when you're trying to move fast, but it's exactly what keeps the site clean and its links valuable. When a seller advertises zero friction and same-day publishing on any topic, treat the ease as the warning, not the selling point.

7. Author pages

Check whether articles have named authors, and whether those authors are real. Click a byline. Does it lead to an author page with a genuine bio, other articles, and an external footprint? Or does it lead nowhere, or to an "admin" account that wrote four hundred posts across nine unrelated industries?

Real authors with real profiles are a sign of a site that takes its content seriously. Anonymous or obviously fake bylines, especially the single "admin" account publishing everything, point to a content mill. This check pairs naturally with the content read in check five; together they tell you whether real people are behind the site or whether it's an assembly line.

8. Site history via the archive

Run the domain through the Wayback Machine and look at what it used to be. Domains get bought and repurposed constantly, and a site's past tells you a lot about whether its authority is genuine or inherited under false pretenses.

The pattern to watch for: a domain that was a legitimate site about one thing for years, went dark, then reappeared as a completely different site loaded with outbound links. That's a expired-domain play, someone bought the dead domain to harvest its leftover authority and is now selling links off the corpse. A site with a consistent history that matches what it is today is far safer than one with a suspicious gap and an identity change.

9. The link neighborhood

Finally, look at who else this site links to and who links to it. You're judging the neighborhood, because your link will live in it. A site surrounded by other quality, relevant publications sits in a good neighborhood. A site whose backlink profile is a swamp of spam, and whose outbound links all point to gambling, adult, and payday-loan sites, is in a bad one, and proximity matters.

This check overlaps with outbound link density but goes wider: it's the whole web of associations around the site. If everything connected to it looks low-quality, the site is low-quality too, regardless of how its individual metrics read.

Where the data comes from, and what it can't tell you

A quick word on tooling, because the checks are only as good as the inputs. Most of the trend, traffic, keyword, and backlink data comes from third-party SEO platforms that estimate these figures, they don't have the site's real analytics. That matters in two directions. The estimates are directionally useful but not precise, so treat a traffic number as "roughly this order of magnitude" rather than gospel. And because they're estimates, the patterns matter more than the digits: a clear downward trend is meaningful even if the exact monthly figure is off.

The archive check uses the Wayback Machine, which is free and surprisingly revealing. The neighborhood and outbound checks you can partly automate with backlink tools, but the most reliable version is still opening pages and looking. And several of the most important checks, content quality, author authenticity, editorial friction, aren't in any tool at all. They require you to actually visit the site, read it, and in the case of friction, start a real conversation with the editor. The sites that fail hardest tend to fail on exactly the checks no tool performs, which is precisely why sellers can show clean metrics on dirty sites. The data narrows the field; your own eyes make the call.

The two-minute human read that catches what tools miss

Before you commit, do one more thing that no metric captures: spend two minutes on the site the way a normal visitor would. Land on the homepage. Click into a couple of articles. Try to find out who runs it. Notice how you feel.

This sounds soft, but it's remarkably reliable, because content mills and link farms have a texture you can sense even before you can articulate it. The layout is cluttered with ads in awkward places. The articles are technically present but lifeless. There's no clear sense of who the site is for or who's behind it. Compare that to a real publication, where the editorial point of view is obvious within a few clicks and the site clearly exists to serve a particular audience. If your instinct says "this site exists to host links, not to be read," your instinct is usually right, and it'll often flag a problem faster than the formal checks. Use the nine points to confirm what the two-minute read suspects.

How to score it: soft warnings versus hard fails

You won't find a flawless site, so don't grade pass-fail on every line. Sort the findings into two buckets.

Hard fails are dealbreakers on their own. Any single one of these and you walk: a high domain rating with near-zero organic traffic, ranking keywords that don't match the site's stated topic, a content mill with fake or "admin"-only authors, an expired-domain history with an identity swap, or a clearly toxic link neighborhood. These signal that the site's value is fake or its association is harmful, and no other strength makes up for it.

Soft warnings are things you weigh together. A modest recent traffic dip, slightly thin content, a higher-than-ideal outbound link count, low editorial friction. One soft warning is usually fine. Three or four stacking up on the same site adds up to a no, even without a single hard fail. The judgment is in the accumulation.

1. Hard fail on any one of: traffic-to-DR mismatch, nonsensical keywords, content-mill authorship, expired-domain swap, toxic neighborhood.
2. Walk if soft warnings stack three or more deep, even with no hard fail.
3. Proceed when the trend is healthy, the numbers agree, the keywords fit, and the neighborhood is clean, accepting one or two minor soft warnings.

Run these nine checks and you'll stop buying links that look fine and quietly hurt you. The whole pass takes fifteen minutes once it's habit, and it's the cheapest insurance in link building. When you've built the discipline of vetting on arrival, the natural next step is making sure the links you do place actually earn their keep, which is where measuring link-building ROI picks up the thread.