Three years ago a client rang me on a Monday morning, genuinely distressed. Her e-commerce site had dropped from roughly 14,000 organic sessions a week to under 4,000. No warning. No recent redesign she was aware of. She'd already been on the phone to two other agencies who both told her it was "probably the algorithm." That's not a diagnosis. That's a shrug dressed up in jargon.
I spent about ninety minutes with her in Google Search Console and Screaming Frog and we found the actual cause: a developer had set the entire site to noindex during a staging migration, published it, and nobody noticed for eleven days. Eleven days. The fix took four minutes.
The point isn't that every traffic drop has an obvious fix. It doesn't. But every traffic drop has a specific cause, and finding it is a process, not a guess. Here's the framework I use across Seahawk's client base.
---
Step One: Establish the Timeline First
Before you touch a single tool, you need a date. Not a range. A date.
Go into Google Search Console, pull the Performance report, and switch the date comparison to "Compare" mode. Look at the graph with the longest window available (sixteen months). You're looking for the exact week, ideally the exact day, that the slide started. Write it down.
Then open a second tab and check Google's Search Status Dashboard alongside the community-maintained Google Algorithm Update History on Search Engine Land. Cross-reference your drop date against confirmed algorithm activity.
If the drop lands within five days of a confirmed core update, broad core algorithm movement is your prime suspect. If the drop is clean, vertical, and lands on a random Tuesday with no algorithm activity, something you did (or your developer did) is far more likely.
I cannot stress this enough: establish the timeline before anything else. I've seen agency owners spend two weeks auditing content quality when the real problem was a redirect chain introduced on a specific deploy date. The timeline tells you where to look.
Questions to ask at this stage
- Did the traffic drop all at once or fade gradually over weeks?
- Did it affect the whole site or specific page clusters?
- Did you (or anyone) deploy, migrate, or update anything around that date?
A sudden single-day drop almost always points to a technical issue or manual action. A gradual six-week fade usually means algorithm sensitivity or content decay.
---
Step Two: Check for Technical Causes Immediately
Technical problems are the most embarrassing causes and also the easiest to fix. So run these checks first.
Crawl the site with Screaming Frog (the free tier handles up to 500 URLs, paid is worth every penny for agencies). Filter for:
- Pages returning non-200 status codes, especially 404s on URLs that previously had backlinks or ranked well
- Noindex tags that shouldn't be there
- Canonical tags pointing to the wrong URL
- Redirect chains longer than one hop
Then jump into Google Search Console's Coverage report. Look at the "Excluded" tab. If you suddenly see a spike in "Excluded by noindex" or "Alternate page with proper canonical tag," something changed in your CMS configuration.
Seahawk had a SaaS client last year where a WooCommerce update quietly added noindex to all paginated archive pages. Took about eight minutes to identify in Screaming Frog. The traffic recovery after fixing it was visible in GSC within three weeks.
Also check your robots.txt. Sounds basic. But I've audited sites where a Disallow: / was sitting in production robots.txt and the site owner had no idea. You can view it at yourdomain.com/robots.txt and test specific URLs inside GSC under Settings > robots.txt.
---
Step Three: Separate Brand from Non-Brand Traffic
This one step eliminates about 30% of the "mystery" cases I see.
In Google Search Console's Performance report, filter queries by brand name. See what's happening to branded versus non-branded clicks separately. If branded traffic is flat or growing but non-branded traffic collapsed, you've narrowed the problem significantly. It means your site authority is probably fine but specific rankings dropped.
If branded traffic also dropped, that points toward something more structural: a penalty, a major indexation issue, or a domain-level problem.
I always set up this split as a saved filter in GSC for every client account I manage. Takes thirty seconds and it immediately gives you a diagnostic axis that most people miss in the panic of a drop.
---
Step Four: Identify Which Pages Actually Lost Traffic
Go back into GSC Performance, add the Pages dimension, and sort by click decline (use the comparison mode you set up in step one). Export to a spreadsheet. You want to see which twenty to thirty URLs account for the largest absolute decline in clicks.
Now look for patterns:
- Is it one content category (blog posts, product pages, landing pages)?
- Is it pages targeting a specific keyword intent (informational vs transactional)?
- Is it pages with a specific URL structure or template?
If thirty blog posts all dropped simultaneously and they all share the same template, you probably have a template-level issue: a canonical tag change, a structured data error, or a significant Core Web Vitals regression on that page type.
If it's scattered across unrelated pages, a broad core update is more plausible.
Run your top declining URLs through Ahrefs or Semrush's position history. I use Ahrefs' Site Explorer for this almost exclusively. Look at whether rankings dropped, or whether rankings held but click-through rate collapsed (which would suggest a SERP feature like a featured snippet or AI Overview took the click instead).
---
Step Five: Look for Lost or Toxic Backlinks
Not every traffic drop is content or technical. Sometimes it's links.
Pull your backlink profile in Ahrefs and sort by "Lost" links within the relevant time window. If you lost several high-authority referring domains in the same period the traffic dropped, that's a strong signal, particularly for pages that were heavily link-dependent for their rankings.
On the flip side, if you recently did any kind of link building campaign, check whether you accidentally attracted spammy links from link farms. Pull the referring domain list, filter by low Domain Rating (under 10), and look at the anchor text. Exact-match anchors from low-quality domains are a pattern worth flagging to the Google Disavow tool, though honestly Google is better at ignoring these than it used to be.
Back in 2021 I worked with a travel blog that had used a cheap link building service before coming to us. Their traffic had been declining for four months. The backlink profile had 600+ referring domains from what looked like a private blog network. Once we disavowed them and cleaned up on-page thin content, the recovery took about five months. Not quick. But it worked.
---
Step Six: Audit for Keyword Cannibalisation
Cannibalisation is slow and insidious. Two pages on your site compete for the same query. Google can't decide which one to rank. Both end up ranking poorly.
I catch this by exporting GSC data into a spreadsheet and looking for queries where multiple URLs are appearing in the "Pages" column for the same keyword cluster. If yourdomain.com/blog/best-crm-tools and yourdomain.com/services/crm-tools are both ranking for "best CRM for small business," they're splitting signals.
The fix is usually a canonical tag (point the weaker page to the stronger) or a proper 301 redirect if one page is genuinely redundant. Sometimes it's a content consolidation job.
This won't cause a sudden crash, but it explains slow multi-month erosion well.
---
Step Seven: Check Core Web Vitals and Page Experience
Google has been using Core Web Vitals as a ranking signal since 2021. It's not the dominant signal, but for competitive queries it tips the balance.
Go into GSC's "Experience" section. Look at the CWV report. Are there pages marked "Poor" for Largest Contentful Paint or Cumulative Layout Shift? Cross-reference those URLs against your list of pages that lost traffic.
Run specific pages through PageSpeed Insights for lab data and field data. Look at LCP particularly. Anything above 4 seconds is a problem. Anything above 2.5 seconds is worth addressing.
I don't want to overstate this. I've seen sites with genuinely terrible CWV scores ranking fine for low-competition queries. But at the medium-to-high competition level, it matters. And when everything else checks out, slow pages are often the remaining culprit.
---
The Diagnosis Decision Tree (Quick Reference)
Here's how I triage the above in order:
- Check the exact drop date and cross-reference against algorithm updates
- Crawl with Screaming Frog for technical issues (noindex, redirects, 404s)
- Split brand vs non-brand traffic in GSC
- Identify the top 20-30 declining pages and look for patterns
- Check whether rankings dropped or just CTR (Ahrefs position history)
- Review lost backlinks in the same time window
- Check for cannibalisation across GSC page/query data
- Run declining pages through PageSpeed Insights for CWV
Most of the time, by step four or five, the cause is obvious. The remaining steps are for when it isn't.
---
FAQ
How long does it take for traffic to recover after fixing the cause?
It depends heavily on the cause. A noindex bug fixed today can see recovery in one to three weeks once Googlebot re-crawls and re-indexes affected pages. Algorithm-related drops tied to broad core updates can take months, sometimes not recovering until the next update rolls out. Backlink-related issues are the slowest, often three to six months minimum.
Should I be worried if the drop is small, say 10-15%?
Honestly, yes and no. A 10-15% drop on a site doing 500 sessions a month is noise. A 10-15% drop on a site doing 200,000 sessions a month is worth a full diagnosis. Proportional significance matters. Also look at whether revenue-driving pages dropped or just informational ones.
GSC is showing impressions are fine but clicks dropped. What does that mean?
Your pages are still appearing in search results but fewer people are clicking them. This is usually a CTR problem, not a ranking problem. Two common causes: a competitor took a featured snippet for your query, or Google's AI Overviews started answering the query directly in the SERP. Check the "Average Position" column alongside CTR. If position held steady and CTR fell, it's a SERP feature issue, not a ranking issue.
Do I need paid tools to do this properly?
You can get surprisingly far with GSC and the free tier of Screaming Frog. For backlink analysis you really do need Ahrefs or Semrush. There's no free alternative that gives you reliable referring domain data. If you're managing client sites professionally, the Ahrefs subscription pays for itself on the first investigation.
---
Traffic drops feel catastrophic in the moment. I get it. But they're almost always explainable, and explainable means fixable. The agencies that look calm and competent when a client panics are just the ones who have a process. Now you do too.