When traffic drops, first rule out a tracking problem (script missing, blocked or double-loaded), then find which source fell (Google search, Discover, social, AI, direct), then which pages or sections fell, and only then look at causes such as a Google update, Discover volatility, seasonality or a technical error like noindex or a broken canonical. Comparing with the same weekday last week avoids most false alarms.
A sudden dip in the traffic chart is stressful, and the first guess is usually wrong. Work through these steps in order: each one takes a couple of minutes and rules out a whole class of causes.
1. Is it real, or is it tracking? (2 minutes)
Most "drops" that are fixed in an hour are measurement problems.
- Open a page in a private window and check the real-time view: does your own visit appear?
- Is the script still on every template? A theme update, a new AMP template or a caching plugin can remove it from some pages. Check a page from each section.
- Did a consent banner or ad blocker setting change? Tools that wait for consent lose the readers who ignore the banner.
- Is the script loaded twice? That doubles pageviews; removing the duplicate later looks like a 50% drop.
- Compare with AdSense or your ad server. If ad impressions held steady, readers are still there and only the measurement changed.
PubPulse emails you when a site that normally has traffic records no pageviews for six hours, so a missing script rarely goes unnoticed for long.
2. Compare like with like (1 minute)
Compare with the same weekday last week, not with yesterday. Food and lifestyle sites peak at weekends, news sites on weekdays. Then look at the same week last year: December, summer holidays and exam seasons move traffic every year.
3. Which source fell? (3 minutes)
Split the drop by source. The cause is almost always visible in one line:
| Source fell | Look at |
|---|---|
| Google search | rankings of the pages that lost traffic, a recent core update, Search Console |
| Google Discover | Discover is volatile by nature; check the Discover report and recent article topics |
| Google News | whether your recent articles were indexed quickly, publisher center settings |
| AI assistants, AI Overviews | whether competitors are now cited for your topics |
| Social | a changed algorithm, a post that went viral last week and ended |
| Direct | often tracking or app in-browser changes; compare with your newsletter sends |
| Referral | a big site that linked to you last week |
See Google Discover traffic for why Discover needs its own line.
4. Which pages or sections fell? (3 minutes)
Sort pages by change, not by size. If the whole site fell evenly, think sitewide: an update, a technical problem, tracking. If one section, one author or a handful of articles fell, the cause is specific to them: a competitor's better article, an outdated post, a lost featured snippet. Custom dimensions for author and category make this step a single click.
5. Check for a Google update (2 minutes)
Look at Google's search status dashboard for core or spam updates in the days before the drop. If one rolled out, wait until it has finished before changing things: rankings often move back and forth during a rollout.
6. Technical checks (3 minutes)
- Search Console: manual actions, security issues, a jump in "Excluded" or "Not indexed" pages.
- A
noindextag orDisallowrule added by mistake, often after a staging copy went live. - Canonical tags pointing to the wrong URL, or to the home page.
- Server errors or very slow pages during the drop: check your host's status and error logs.
- Redirect chains after a URL change.
7. Write down what you found
Add an annotation on the chart with the date and the cause. Next time you look at a year of data, you will know why that week was low, and so will your team.
Set it up before the next drop
- Turn on traffic alerts, so a missing script or a spike reaches your inbox.
- Track author, category and publish date as dimensions.
- Keep a weekly summary email, so you notice slow declines that a daily look hides.
You can try all of this on sample data in the live demo.
Try PubPulse free for 30 days
Every feature, no pageview limit, no credit card. Paste one line of code and your first visits show up within seconds.
Start your free trialor click around the live demo first, no sign-up needed