The Extreme-Outlier Guard is a safeguard in RankGear’s Deficit math that stops one implausibly extreme competitor value from becoming the target a Maximum strategy tells you to chase.
| Term | Extreme-Outlier Guard |
|---|---|
| Category | Statistics and Evidence |
| Also known as | Outlier Guard, Maximum Guard |
| Where it appears | Deficit math |
What it means in RankGear
When a Deficit strategy sets its target from the highest value in the competing set — a Maximum strategy — a single unusual page can distort the goal. One competitor that stuffs a factor far past everything else around it would, left unchecked, pull your target up to a figure no ordinary ranking page comes close to. The Extreme-Outlier Guard is the check RankGear applies inside Deficit math to catch that case: it recognizes when the top value stands apart from the rest of the distribution and keeps that lone extreme from being handed to you as the number to hit.
How to interpret it
Treat the guard as a reason to trust a Maximum target more, not less. Its job is quiet — when it fires, the target you see reflects the realistic high end of the competing pages rather than a freak reading. When you are reading a Maximum-based deficit, still look at the spread of the underlying values yourself: a target that sits well above the pack, even after the guard, is a sign the factor is highly variable and worth checking before you commit to it. Whether a value was genuinely measured, missing, or a true zero also shapes how far to trust any single reading the guard is weighing.
Example
Suppose nine of the top ten pages for a keyword use a factor somewhere between 8 and 15, and the tenth uses it 240 times. A raw Maximum strategy would point your target at 240 — a figure that reflects one anomalous page, not what competes. The Extreme-Outlier Guard flags that 240 as detached from the rest of the distribution and holds the target at the realistic upper band the other nine pages describe, so the goal you write toward matches pages that actually rank rather than a single outlier.
Important considerations
- The guard reduces the influence of extreme values on a Maximum target; it does not erase the underlying variability. A wide spread among the remaining pages still calls for judgment.
- Any target drawn from competitor data is a comparative indicator of what ranking pages look like, not a score Google assigns. Matching it is an association, not a proven cause of ranking.
- Small SERP samples are unstable, and with only a handful of results even the guarded target can shift between runs without anything meaningful changing.
Related terms
Part of the RankGear glossary · how RankGear measures · the 870 factors.