Analysing a backlink profile

Domain overview’s Backlinks tab keeps the history of who links to your site and to the rivals you track. Looking costs nothing and asking for fresh figures costs something, and the split between those two is deliberate rather than incidental.

The button fetches; the tab reads

For your own site and the rivals you track, the tab has its own press, Analyze backlinks. It makes the calls and stores what came back. Opening the tab afterwards fetches nothing at all — every figure on it is read from what that press already paid for. One press asks for a summary of the domain you chose, a week-by-week count of referring domains gained and lost, and the backlinks index’s own top 100 referring domains of your own site and of up to 5 rivals you track.

A change needs two readings, and the first press is one

Every delta on this tab is one stored snapshot against an earlier one, so the first analysis has nothing to compare against and says so. This is also why the product stores its own history rather than reading a cache: a cache holds the newest answer, which can say what the profile is now and can never say what it was last month.

The follow share is domains over domains

The share of referring domains that pass authority divides domains by domains. The provider also reports a count of individual links, and mixing the two populations would give a plausible-looking percentage that is about nothing. A domain with no referrers at all has no share rather than a share of nought.

Authority is on a scale, and two scales are never subtracted

A domain authority figure only means something on the scale it was asked for. Where two stored readings were taken on different scales the screen says the scale changed rather than drawing the difference, because the same site can read very differently on two rulers and the subtraction would look like a collapse.

The three lists

The stored summary and its change sit above the list row as figures, on every list, along with the chart of referring domains gained and lost by week. Each list below them reads the referrers stored for the domain you are looking at. The link gap, the domains among a rival’s referrers and not among yours, is Backlink gap: with a project it lists the stored prospects, and you can instead name domains and compare them from the answers Domain overview has already fetched. Both sets are the index’s top 100, so a domain linking to you from outside your own can appear there.

• Referring domains: every referrer stored for this domain, with its authority, how many links it points at you, when the index first saw it and what it scores for spam.
• New: the same rows narrowed to the ones the index first saw inside a recent window, counted back from the day the analysis was taken. A referrer the index reported no first sighting for is not in here — unknown is not today.
• Spam-flagged: the referrers the index distrusts, counted over the referrers this product stored.

A referrer the provider reported no first sighting for is never counted as newly gained, and one it reported no distrust figure for is in no band rather than in the clean one. Both are the same rule as a dash elsewhere: an absent measurement is not a reassuring one.