Link Exchange Guidelines: Domain, Placement, Anchor, and Quality Standards

The reason teams end up with backlink profiles they cannot explain is rarely that nobody had judgment. It is that the judgment belonged to one person and the volume outgrew them.

If a new hire could not apply your standards without asking you, they are preferences.

Watch how it happens. One person qualifies every partner for the first year and the profile stays clean, because they were applying twenty criteria they could not have listed if you asked. Then a second person joins, or that person goes on holiday, or the target goes from six links a month to twenty. The criteria do not transfer, because they were never stated, and the new approvals are looser in ways nobody notices for two quarters.

That is the whole argument for writing this down, and it is worth being clear that the writing is the hard part rather than the deciding. Most teams already agree on what a bad partner looks like. Very few can produce the sentence that would let somebody else reject the same one.

One practical note before the standards themselves. Write them as criteria a person can check, not as adjectives. Relevant site is not a standard. A site whose ranking keywords overlap with ours, or that our buyers would plausibly read, is one, because two people applying it reach the same answer.

This is the most commonly skipped standard, and skipping it is how teams end up with links on pages nobody has visited since publication, from sites they were right to approve.

A good site can hold a page you would never want to be on.

A site can be excellent and still hold a thin roundup written for link swaps, an archive nobody maintains, or a page that was noindexed during a migration and never restored. None of that is visible from a domain report, and all of it decides whether your link does anything.

So run a second, shorter pass on the specific URL. Is the page indexed. Does it rank for anything. Does the content look like somebody wrote it for readers. How many outbound links does it already carry, and where do they go. Would you have pitched this page cold, before any exchange was on the table.

That last question is the one to write into the standard, because it collapses the other four for anybody who has not learned them yet.

Two rules on placement. Links go inside the main content of a page, in a sentence that explains why the destination is relevant, and not into the template that renders on every URL of the site. If you want the reasoning, we sorted every position by what it is actually worth in the guide to backlink placement types.

Whoever writes the sentence chooses the anchor. Neither side dictates it to the other.

Three rules on anchors, and they are stricter than most teams expect. Whoever writes the sentence chooses the anchor, which means neither side dictates it to the other. The anchor describes the destination rather than the keyword somebody wants to rank for. And the same commercial phrase does not repeat across partners, because the repetition is the pattern regardless of how reasonable each individual anchor looked.

The practical version for a standards document: brand names, product names and descriptive phrases are all fine. A supplied exact match commercial anchor is an automatic decline, and it is worth writing that as a hard rule rather than a preference, because it is the single request that most reliably predicts what the rest of the arrangement is.

One rule covers most of this. If money, product, a free subscription, or any other consideration changed hands, the link carries a sponsored or nofollow attribute, and that is not negotiable regardless of how editorial the placement looks.

Two things worth adding to a standards document beyond that. First, verify the attribute after publication rather than assuming it, because a great many sites qualify external links by default through a plugin and nobody on their side considers it worth mentioning. Second, an attribute is a disclosure and not a repair: adding nofollow to an arrangement you would not want explained does not make it acceptable, it just removes the ranking claim.

The mechanics of the four link states, including why there is no such thing as rel=dofollow, are covered in dofollow versus nofollow.

Ask a team whether an exchanged link is live and you get one answer. Ask what they checked and you usually get one thing: they opened the page and saw it.

Every one of those failures renders as a perfectly normal link in a browser.

Live is not one fact. It is at least seven, and they fail independently. The URL can be a variant or a redirect. The anchor can have been edited. The page can carry a noindex, or a canonical pointing somewhere else, either of which means it passes nothing. The page can be indexable and simply not indexed. The link can be injected by JavaScript or routed through a tracking redirect. The rel attribute can be something nobody agreed. The placement can have moved into a footer.

Every one of those renders as a perfectly normal link in a browser. That is the whole problem, and it is why a verification standard has to list the checks rather than say confirm the link is live, which is the sentence most guidelines actually contain.

The honest note on effort: doing this by hand takes a few minutes per link and nobody sustains it past about thirty placements, which is roughly where scheduled rechecking stops being a convenience.

Worth separating, because teams write one and call it the other, then wonder why it does not settle disputes.

Guidelines explain themselves. A policy does not have to, and that is the point of it.

Guidelines describe how a good exchange should be done. They are educational, they explain reasoning, and they help somebody make a judgment call in a situation the document did not anticipate. This page is guidelines.

A policy defines what your organization allows, prohibits and requires. It is short, it is specific, it does not explain itself, and its job is to be quotable when somebody disagrees. Prohibited categories, mandatory attributes, who can approve, what needs sign off.

Most teams need both and start with the wrong one. Write the guidelines first, run them for a quarter, then extract the parts that turned out to be non-negotiable into a policy. A policy written before you have applied anything is a list of guesses about which rules will matter.