Runbook
Sudden drop in organic traffic: the link-side runbook
Every thread about a traffic drop ends in guesses. This runbook only covers the link side, in an order that lets you stop early when the answer is clear.
Links explain some drops, not most. Before step 1, check that tracking works, that Search Console shows the same dip as GA4, and that nobody changed robots.txt or noindex tags. Links come after those.
The runbook, in order
T+0: Date the drop
Find the first day of decline in Search Console. Compare weeks, not days, because weekends and holidays bend the curve. Write the date in the ticket.
T+5: Check the Google calendar
Did a named update start within three days of that date? Examples: the March 2024 core and spam updates, or the March 2026 spam update. If yes, links may not be the cause, and rivals will have moved too.
T+10: Pull the snapshot pair
Call refdomains for your domain and diff it against your last saved file. No saved file? Take one now, and treat this run as the baseline for future incidents.
T+15: Read the anchors
Run anchors. A sudden pile of one commercial phrase is a different problem from lost domains. See the spam triage page.
T+20: Check the rivals
Run compare on three competitors. If they are flat and you lost domains, links are a suspect. If everybody moved, look at the update.
T+30: Decide
Write one of three outcomes: links ruled out, links suspected with a named list, or data too thin. Never write 'links caused it' without a lost-domain list.
Reading the evidence
| What you see | Likely meaning | Next move |
|---|---|---|
| Referring domains down 20% or more, rivals flat | Real link loss is plausible | Rank the lost list by the pages they pointed to, then contact owners |
| Domains stable, one anchor surged | Spam or paid links, possibly from a rival | Open the triage |
| Everything stable, rivals moved too | Algorithm update or SERP change | Wait for the rollout to finish before editing |
| Everything stable, you alone moved | Technical or content cause | Go back to indexing and templates |
Myth versus reality
Myth: a drop right after a core update means you lost links. Reality: core updates reweigh content and many other signals at once, and Google says improvements may only show at a later update. A rollout takes up to two weeks, so judge it after it ends. Wait.
Myth: more links will bring the traffic back. Reality: that works only when the loss was real and the lost links mattered. Buying volume to fill a hole is how a small problem turns into a manual action. Don't.
We cannot see your Search Console. This page never says 'links did it'. It says whether the link evidence supports that story, and often the honest finding is 'inconclusive', which is a useful result to log.
Myth: a rival's link spike explains my drop. Reality: it can, but it rarely does alone. Check the date alignment first, then the anchors. Skip the conspiracy. Let the data ask for it.
Ticket template
- Drop date, tool that shows it, and size in clicks and percent.
- Update calendar check: yes or no, with the update name and month.
- Snapshot dates compared, and the number of domains lost and gained.
- Anchor share of the top three phrases before and after.
- Competitor comparison result.
- Conclusion: ruled out, suspected, or inconclusive.
What to log afterwards
Write a postmortem. Three lines will do, even for a false alarm. What the signal was, what you checked, what you concluded. Teams that skip this repeat the same investigation every quarter, and each repeat costs an afternoon.
Real incident? Add the missing alert. A drop you only noticed in a monthly report means your thresholds were too loose or your cadence too slow. Fix the monitor, not only the site. Then re-run the table from the severity page.
