Traffic drop after core update: a link-data post-mortem
Someone's visits fall by half in a week, and the first guess is always links. Sometimes it is. This walkthrough, built on an invented site, shows how to find out in an afternoon.
LinkBuildingStat research team ·
The invented incident
Core update starts rolling out
The site is a hypothetical online shop. Google announces a core update; nothing is changed on the site.
Search Console shows clicks falling
Clicks from search go from about 1,000 a day to about 600. Impressions fall by less. That mismatch hints at positions slipping, not at pages vanishing.
The panic
Someone blames a bulk of cheap links bought a year ago and proposes a disavow file. Nobody has counted anything yet.
The afternoon of counting
The team runs three link lookups and one comparison. This page is about what those can settle.
Hypotheses, ranked by what link data can test
First, the profile may have shrunk. Run overview for today and compare it with the figure in your last monthly note. If referring domains fell from 380 to 250, that deserves a look; if they moved from 380 to 376, links are almost certainly not the story.
Second, rivals may have grown. Our compare call puts you beside them for 1 credit. A rival that doubled its referring domains in a quarter might explain a lost position, though not a halved click rate.
Third, and most often, the cause lies elsewhere: content judged weaker, search intent shifting, a feature on the results page taking clicks. Link data cannot test that, and it is better to say so than to invent a link story.
The afternoon, in order
Open Search Console and note which queries and pages lost clicks
Run overview and compare it with last month's note
Run compare against the rivals who now outrank you
Read anchors for a sudden pile of one phrase
Write what link data ruled out, and what remains open
Where link data does matter
Some drops do follow link problems: a manual action in Search Console, a large block of links you built yourself, a sudden loss of a few strong referring domains.
In those cases the numbers show up quickly in refdomains and anchors. The second step is deciding whether a disavow is warranted, and that decision should come from counts rather than a hunch.
Search Console remains the primary tool for what changed in search. Our index is the second opinion on the link side.
A note on timing
Core updates roll out over a couple of weeks. Reading a chart on day 4 is reading half a result.
The August 2025 spam update and the June 2025 core update fell in the same year, which is why one dip can have two candidate causes. Separate them by date before you blame anything.
What the team writes at the end
A post-mortem ends with a document, not a mood. Ours would have five lines: what happened, what we checked, what link data excluded, what remains open, and the date to look again.
Do not forget the last line. A core update's effects often settle over weeks, and checking on day 7 and again on day 35 beats staring at a chart daily.
Keep the snapshots from before the incident. Without a baseline, 'the profile shrank' is a guess.
Verdict on the invented case
Links explain little, and that is a result
In this made-up case the profile moved by 1 percent while the drop was 40 percent. That rules out a link collapse and sends attention to pages and queries. A negative result still saves a week.
Real updates, such as the March 2024 core update, hit many sites for non-link reasons
Recovery usually needs a later update, so expect weeks, not days