Traffic Down, Rankings Stable? Here's What's Really Happening

You open your analytics and your stomach drops. The traffic chart looks like it fell off a cliff. Your first thought, the one that sends a cold spike of panic through you, is that you've been penalised by Google. You're already mapping out which pages to cull, which links to disavow, and preparing for a month of brutal, site-wrecking trench warfare to win back favour.

Stop. Before you tear your site apart, you need to diagnose what's actually wrong. Right now, two completely different forces are causing this exact symptom, and confusing them will send you down a rabbit hole of wasted work.

The first step is to look at the right data. Forget your main analytics for a moment and open Google Search Console. The answer is usually right there, hiding in the relationship between two simple metrics: impressions and average position.

Here’s a real example from one of my own projects. Month-on-month, traffic fell sharply. But when I looked at the Search Console data, my average position across thousands of queries held rock-steady at 8.7. My impressions, however, had cratered. They slid from around 500 a day to a trough of 110-170, before a partial recovery to ~370. Through all that volatility, the site’s actual ranking barely moved.

This tells me the problem wasn't a penalty. If you get penalised, your position drops. If your position holds but your impressions fall, it means fewer people are searching for the terms you rank for. This is a demand-side drop. It's not your fault. The market's interest has shifted, and no amount of on-page SEO tinkering is going to change that. The searches simply aren't happening.

But there's a second, more insidious problem often layered on top. In the same data, my site-wide click-through rate (CTR) was a tiny 0.11% on over 9,500 impressions. This is the pixel problem. Even when people are searching, Google's own features—like AI Overviews and rich snippets—are answering their questions directly on the search results page. Your perfectly-ranked link is just a footnote. The impression happens, but the click never comes.

So, how do you know which problem you have?

1. Impressions Down, Position Stable: This is a classic demand-side drop. The world changed, not your website. The audience that used to search for your core terms has moved on or is searching in a different way.
2. Impressions Stable, Clicks Down (Low CTR): This is the pixel and AI-search problem. You are visible, but you are no longer the destination. Google is satisfying the user's intent on the results page itself.

Each diagnosis demands a completely different kind of work.

If you have a demand-side problem, your job is not to "fix" your existing pages. It's to find the new demand. You need to do the hard work of market intelligence. Where did your audience go? What are they searching for now? The solution is often building new content or features that target these adjacent, growing areas of interest. You need to skate to where the puck is going, not obsess over the ice where it used to be.

If you have a pixel and AI-search problem, the work is more technical and strategic. Your fight is now for the click itself. You need to analyse the SERP for your most valuable queries. Can you win a rich snippet? Can you structure your data to be more appealing than the AI Overview? Sometimes, the answer is to target queries that are less easily summarised by AI—queries that demand comparison, deep customisation, or a trusted brand. You might need to change your content's format entirely, focusing on video, tools, or downloadable assets that can't be easily scraped and regurgitated.

This is a diagnosis problem before it is a fixing problem. Guessing that you've been "penalised" when your rankings are stable is the most common and costly mistake founders are making right now. It leads to weeks of panicked, unnecessary changes that do nothing to address the root cause. The data tells you the real story in an afternoon.

This is my core philosophy: AI at the edges, engineering at the core. We use smart tools like Claude to read the trend data and separate the signals, but we back it up with real engineering—pulling from the APIs, structuring the data, and implementing fixes that are grounded in a correct diagnosis.

If you're looking at a scary traffic chart and want to know which problem you actually have, this is the exact diagnosis I provide in my Claude Ops Audit. It's a 7-day deep dive into your Search Console data to give you a clear-eyed, actionable plan. I only have 3 slots open per month. You can find out more at /claude-ops-audit.