How to check whether your link actually works: four conditions and the mistake almost everyone makes

You placed or bought a link. It is there, you can see it, the contractor reported it done. Whether it actually works is the question usually answered by opening the page source and searching for rel="nofollow". We checked 27 platforms and found that on one in seven this method returns the wrong answer — and it manages to be wrong in three different directions. Below: what "the link works" actually means, how to check it in a couple of minutes, and why the familiar method lets you down.

What "the link works" actually means

A link is only useful when four conditions hold in sequence. Not one of the four — all four, and a break at any step cancels out the rest.

We ran 27 platforms commonly used for link placement through these four conditions — one fresh publication each. Here is what survived:

Of the 27 platforms, a link to the target address exists at all on 16. Of those sixteen, 13 are open for passing weight. And of the thirteen, four sit on pages closed to indexing. Nine out of twenty-seven actually work.

Where the familiar check breaks

Now the main point. The standard advice — open the source and look for rel — rests on the assumption that a search engine sees the same markup you see in View Page Source. It does not. Google describes its own process plainly:

So the decision is made on the page after scripts have run. And scripts change the page — including in places you would not expect.

In our sample four platforms out of 27 diverged — that is one in seven. And the errors came in three different kinds, not one:

• A script strips nofollow. In the source of telegra.ph and graph.org the link carries rel="nofollow". After the scripts run, the attribute is gone — the editor redraws the article and loses it. Googlebot executes JavaScript and sees an open link. A source check would reject a working platform.
• A script adds nofollow. notepin does exactly the opposite: the server returns a clean link and rel="nofollow" appears only in the browser. A source check would approve a platform that passes nothing.
• A script creates the link itself. On Strikingly there is no link in the source at all — the tag appears only after rendering. A source check reports "no link" and is wrong.

Anyone can verify this right now: open such a page and compare View Page Source with the Elements tab in developer tools. The difference is visible with the naked eye.

Dofollow does not yet mean weight is passed

The second place where the reasoning stops too early. "The link is dofollow" and "the link passes weight" are different statements, and the first does not imply the second.

If the page is closed to indexing, the search engine does not take it into the index, and so it does not count the link signal from it — however open the link itself may be. In our sample four of the thirteen open links sat on such pages: an honest link, an invisible page, a result of zero.

Both signs have to be checked, and in this order: first whether the page is indexable, then whether the link is open.

The category nobody warns you about

The largest group in our sample is platforms where there is no link at all: eleven out of twenty-seven. The address is visible on the page as text, it looks like a link, but there is no <a> tag in the markup either before or after rendering.

Most often these are services for publishing notes and code snippets that do not turn addresses into links automatically. The contractor's report looks impeccable: the page exists, the address is on it, a screenshot is attached. The value is zero.

A checking protocol: six steps

The order in which any placement is worth checking. Each next step only makes sense once the previous one has passed.

What to do when there are many placements

A dozen links can be checked this way by hand. When there are hundreds, you need automation — and one detail matters here: an ordinary request to the page will not do, because it does not execute scripts. You need a headless browser that renders the page the way a search engine does.

1. Open the page in a headless browser and wait for it to load
2. Give the scripts time to run (a couple of seconds is usually enough)
3. Find the link to your own domain in the DOM
4. Read its rel
5. Separately read meta robots and the X-Robots-Tag header

That is exactly how our own checking works — and it is built not into a report but into the mechanism that picks platforms in the first place.

How this works in PromoPilot

Every platform in the system has a priority, and it is not set once and for ever — it is recalculated from that platform's actual performance. Three things feed into it.

Indexing. We regularly check whether published pages reach the index, and the hit rate moves the priority within a range of roughly −25 to +30. The scale is continuous, with no cliffs: a platform with one accidental indexing out of a hundred gains no advantage over a completely dead one. A platform that forbids indexing in its own markup gets the minimum and cannot climb at all.

Passing weight. The very property this article is about — and it is checked precisely in the rendered DOM. Confirmed weight passing earns a solid plus; confirmed failure to pass earns the largest penalty in the system. The logic is simple: a useless link is worse than a weak one.

Domain metrics. The platform's Domain Rating counts too, but carefully: it is scored from a threshold rather than from zero — otherwise the difference between a strong blog and a very strong one would eat the whole range, and a notes service would collect the same bonus as a major platform.

The sum of these adjustments is applied on top of the base priority, and the ordering of platforms shifts on its own as data accumulates. A platform whose indexing has dropped, or which has started closing its links, sinks without anyone deciding it should. A new platform with good numbers rises. That is why the set of active platforms is not static — the ones at the top are always those performing best on current data.

What not to do

In short

A link works when four conditions come together: the page is alive, it is indexable, a link tag exists, and it is not closed by an attribute. In our sample all four came together on nine platforms out of twenty-seven.

And the main thing: the familiar source-code check is wrong on one platform in seven, because the search engine decides on the page after the scripts have run while you are looking at it before. The difference shows up in developer tools in ten seconds — open the Elements tab instead of View Page Source.

Check your site

The crawl shows what keeps your pages out of the index — and a non-indexable page passes no weight through any link.

Frequently asked questions

How do you check whether a link passes weight?
Four checks in sequence: the page answers with a 200, it carries no noindex either in the meta tag or in the X-Robots-Tag header, a link tag pointing at your address exists in the rendered DOM, and that tag carries no nofollow, ugc or sponsored value. A break at any step cancels the result.

Why can you not check nofollow in the page source?
Because the search engine indexes the page after JavaScript has run, while the source shows it before. Scripts can strip the attribute, add it, or create the link itself. In our sample of 27 platforms the source and the real DOM diverged on four, that is on one in seven.

Where do you look at the rendered DOM?
In your browser's developer tools, on the Elements tab. That is the state of the page after scripts have run. View Page Source shows something else — the server response before scripts. For checking at scale you will need a headless browser; an ordinary request to the page is not enough.

The link is dofollow — is that enough?
No. If the page is closed to indexing it will not enter the index, and the link signal from it will not be counted, however open the link itself may be. Both properties need checking, starting with whether the page can be indexed.

My address is visible on the page but the checker says there is no link. Who is right?
Most likely the checker. Some platforms output the address as ordinary text and never create a link tag — to the eye it is indistinguishable from a link, but to a search engine it is just letters. Search the DOM: what you need is an actual link tag, not text containing the address.

How long after publication should you check indexing?
The first checks make sense after a couple of weeks, a final verdict after a month. Permission to be indexed and the fact of being indexed are different things: a page can be fully open and still never enter the index.

Sources

PromoPilot's own measurement, 11 September 2026: 30 link-placement platforms, one fresh publication each; the server response and the rendered DOM (headless Chromium) were compared. 27 completed the check; on three the page did not finish rendering.
Google Search Central, "JavaScript SEO basics": the three processing phases (crawling, rendering, indexing), JavaScript execution in a headless Chromium, and the use of rendered HTML for indexing.
Google Search Central, documentation on rel attributes: the nofollow, ugc and sponsored values.