Tagged “volatility”
-
Cliff, Slide, or Step: Reading the Shape of a Drop
The same 15-position loss has three different shapes in your history, and each one rules out different causes. Read the shape before you read the size.
-
Rank Movement and Your Deploy Log
Most ranking questions become answerable when you can see what shipped and when. Annotating your rank history with releases costs almost nothing.
-
What Recovery Looks Like After You Fix Something
Recovery is not the mirror image of the drop. It arrives staggered, partial, and later than the fix — and re-fixing on day three destroys the evidence.
-
How Often Should You Check Keyword Rankings?
Weekly for most sites, daily only when something specific is in flight. Checking more often adds noise, not information — here's the reasoning.
-
Why Do Google Rankings Change Every Day?
Daily ranking movement is mostly personalisation, geography, and SERP layout — not algorithm updates. Here's how to tell real drops from noise.
-
How to Diagnose a Ranking Drop Without Guessing
Work from the widest possible cause to the narrowest. Most drops are explained in the first two checks, and most wrong diagnoses skip straight to the last.
-
Sitewide Drops Versus Page-Level Drops
The two have almost no causes in common, and telling them apart takes one query of your own data. Here's how to separate them and what each implies.
-
Rank Alerts That Don't Cry Wolf
An alert on 'moved three places' fires constantly and gets muted within a week. Alert on band crossings, breadth, and persistence instead.
-
You Can't Read a Position Without a Baseline
A position number alone carries no information about whether it's unusual. The baseline is the keyword's own observed range, and it takes weeks to earn.
-
What SERP Volatility Trackers Actually Measure
A volatility index measures how much a fixed keyword sample reshuffled since yesterday. It's a useful 'not you' signal and nothing more specific than that.
-
Weekday and Seasonal Effects in Position Data
Comparing a Monday to a Saturday, or November to January, builds an offset that looks like a trend. Fix the weekday and use like-for-like periods.