The short version
- Every known backlink carries its own next check time, and a sweep runs every six hours to check whatever is due.
- Check frequency backs off as a link proves durable: roughly daily while new, weekly once established, monthly once stable.
- Each check records whether the link is present and whether it is dofollow or nofollow — a link changed to nofollow is a real event, not an unchanged one.
- Won and lost links both trigger an email alert, so a removal surfaces in days rather than at the next audit.
- Every check is stored as history, so a link that flickers is distinguishable from one that is genuinely gone.
Earning a link is the part everyone measures. Keeping it is the part nobody watches.
Pages get edited during a content refresh, and your entry does not survive the cut. Sites migrate and the article becomes a redirect. A section gets rewritten and someone converts every outbound link to nofollow as a policy. A publisher folds and the domain lapses. In none of those cases does anyone tell you. The link simply stops existing, and you find out at the next audit — if there is one.
Every link has a next check time
Monitoring is built around a single field on each known backlink: when it should next be checked. A schedule runs every six hours, collects the links that are due, and fans out an individual check for each.
This shape matters more than it might appear. Sweeping everything on a fixed cadence means either checking stable links far too often or checking new ones far too rarely. Storing the next check per link makes the cadence a property of the link, which is what lets it vary sensibly.
The cadence backs off
A link that was published yesterday and a link that has survived a year are not equally likely to vanish next week, and checking them equally often would be a waste at one end and negligence at the other.
| Stage | Interval | Why |
|---|---|---|
| New | About every 24 hours | Fresh links are the volatile ones. Editors revise a post in the days after publishing, and a placement removed in review should be caught immediately. |
| Established | About weekly | The link has survived the editing window. It is probably real, and no longer needs daily attention. |
| Stable | About monthly | A link that has held for a long time rarely disappears next week. Monthly catches migrations and redesigns without wasting requests. |
Freshly won links are volatile and worth watching closely. A link that has survived a year rarely vanishes next week — the check schedule should reflect that, not average it.
What a check actually looks at
A check fetches the page and looks for a link to your site. It resolves two separate questions, and conflating them would hide a common and consequential change:
- 01
Is the link still there?
The page is fetched and inspected for a link pointing to your domain. A page that no longer resolves, or that resolves without the link, is a loss.
- 02
Is it still dofollow?
A link converted to nofollow is still a mention and no longer the same asset. It is recorded as a distinct status rather than being counted as unchanged — or, worse, as lost.
Backlink check cadence
Swept every 6 hours · each link carries its own next check
New
~24 hours
The editing window. Placements cut in review are caught the next day.
Established
~weekly
It survived the revisions. Probably real.
Stable
~monthly
Catches migrations and redesigns without re-fetching daily.
Live
Present and dofollow
Nofollow
Still a mention, no longer the same asset
Lost
Gone — you get an email
Every check is kept. A single failed fetch surrounded by successful ones reads as an anomaly, not as a lost link.
History, not just current state
Every check is written to a history, not just applied to the link's current status. That is what makes the record useful rather than merely current.
The web is unreliable in ways that look exactly like link loss: a site is down for twenty minutes, a CDN serves an error, a page ships broken markup for an afternoon. Without history, one bad fetch reads as “link lost.” With it, a single anomalous check surrounded by successful ones is visibly an anomaly, and a link that went missing three weeks ago and never came back is visibly gone.
Checks are also retried on transient failure, so a momentary network problem does not become a false alarm in the first place.
Alerts when it matters
Two events are worth interrupting you for, and both send an email:
- A link went live. The placement you pitched is confirmed. It is the moment the work paid off, and it is also the moment to thank the person who made it happen.
- A link was lost. The window in which you can do something about it is measured in days. An email that has just been quietly cut can often be restored with a short, friendly note; the same note six months later is a cold pitch to someone who has forgotten you.
Why this closes the loop
Discovery, qualification, contacts, and outreach all point one direction: getting a link placed. Monitoring is what turns that into something you can actually report on.
Without it, a link-building programme measures effort — emails sent, replies received, placements agreed. With it, the measure is what survived, which is the only number that corresponds to the value being claimed. It also produces the honest version of the story: not “we earned forty links this year,” but “we earned forty and thirty-four are still live, and here is when the other six went.”