Post-mortem
Traffic drop after a core update: write the post-mortem
Everyone asks the same thing at 8 a.m.: is it Google, is it us, or is it the tracking? A post-mortem answers in that order and refuses to guess.
Start with the timeline, not the theory
A drop has a start date. Find it before you read a single forum thread about a traffic drop after a core update.
Google confirmed core updates in March 2024, August 2024, December 2024, March 2025 and June 2025, and each rolled out over about two weeks, so a drop on day three of a rollout may not yet be the final picture. Overlay your own change log on those dates: releases, migrations, template edits, robots.txt changes.
If nothing of yours moved and your Search Console graph bends within a rollout window, you have a candidate. If your own deploy landed the day before, you have a better one. Either way, the traffic drop after core update week deserves a written timeline. Write it down.
Causes ranked by how fast you can rule them out
| Suspect | How to check | Time to rule out |
|---|---|---|
| Tracking broke | Compare GA4 against Search Console clicks for the same pages; a gap that opened overnight is the tell | 10 minutes |
| Indexing or sitemap | Search Console page indexing report, plus a site: spot check on the dropped URLs | 20 minutes |
| Technical release | Diff robots.txt, canonicals, status codes and redirects against last week's crawl | 1 hour |
| Links lost or gained | Diff two refdomains snapshots; see lost links | 30 minutes |
| Cannibalization | Two of your URLs swapping places for the same query in the performance report | 1 hour |
| SERP change | Open the live results: AI Overview, forum threads, big-brand listicles where you used to sit | 30 minutes |
| Quality reassessment | What is left after every other line is clear | Days to weeks |
Link evidence: one comparison
Pull referring domains for the dropped site on two dates and print what left and what arrived. The index refreshes about monthly, so choose snapshots on either side of the update.
import json
old = {r['domain'] for r in json.load(open('2026-09-01.json'))['rows']}
new = {r['domain'] for r in json.load(open('2026-10-01.json'))['rows']}
print('lost :', len(old - new))
print('gained:', len(new - old))
print('sample lost:', sorted(old - new)[:20])
# Field names ('rows', 'domain') vary by action: print one response first.Example write-up (hypothetical numbers)
This is an invented example, not a real client. It shows the tone we would aim for.
Impact. Organic clicks fell 38% over nine days for a fictional recipe site with 140 pages. Detection. A weekly Search Console export crossed a 25% week-on-week alert. Cause. Referring domains moved from 410 to 402, which rules links out as the main driver. Two pages swapped rankings for the same query, which is cannibalization.
Action items. Merge the two pages with a 301, update internal links, recheck in four weeks. No disavow. No rewrite of the whole site.
Our position
Do not rewrite the site in week one
Core updates reassess quality over weeks. Change what you can prove broke, then wait for the rollout to finish before touching content.
- Fix measurement and technical faults immediately.
- Merge cannibalized pages once you have proof.
- Leave content strategy alone until the rollout ends.
What to write in the doc
Six headings are enough: impact, timeline, detection, cause, what we ruled out, action items. The ruled-out list is the part future you will thank you for.
Keep the evidence beside the claim. A line like "links ruled out, 402 vs 410 referring domains" survives a skeptical manager better than "links looked fine".
If the answer ends up being links, the alert rules for rivals and the weekly diff recipe stop it surprising you next time.
Questions we hear after every update
Can I recover from a core update drop?
Often partially, and rarely overnight. Recovery usually needs a later update to reassess the site, so expect months, not days. Nobody can promise it. What you can do is remove real faults and keep publishing pages worth citing.
Why did rankings hold while traffic fell?
The results page changed around you. AI Overviews, forum threads and shopping blocks can push the same position lower on screen. Compare click-through rate for the same queries before and after, not only average position.
Do lost backlinks cause a drop after an update?
Only sometimes, and Rankift cannot tell you which update did what. A link loss large enough to matter shows up as a clear fall in referring domains between two snapshots. Small churn of a few percent rarely explains a 30% traffic fall.
How long should I wait before acting?
Until Google marks the rollout complete on its status dashboard, unless you find a clear technical fault. Fix faults at once, but hold content changes. Editing during a rollout muddies the evidence you will need later.