SEO SAVAGES field studies
THE SAVAGE LAB
Original-data SEO studies. We treat metrics and tactics like measurement instruments: we pull the real data, test what they actually measure, and publish what we find. Method stated, limitations included, no cherry-picking.
if you buy SEO
We measure before we claim
SEO reports are full of numbers that sound precise: keyword difficulty, search volume, expected click-through by position. Your budget and your board forecast rest on them. The Lab checks whether those numbers hold up before we let one drive a client decision.
The method comes from Blazej Mrozinski, a PhD psychometrician with 16+ years in stats and data science. Psychometrics is the discipline of testing whether a measure measures what it claims to. Each study below states its sample, its method and where it stops being reliable.
The same method runs our client work. See it applied in the case studies, or read how the fractional SEO service works if you want the two founders on your own numbers.
- Is Search Volume Real Demand? We Checked 30 Keywords
Across 30 keywords, only ~half of searches (median 51%) end in a pure-organic click; search volume overstates organic opportunity ~2x overall, and far more on commercial …
read the study → · n = 30 keywords / 9 niches · by Bobby K - The Ten Blue Links Are Gone: We Dissected 30 SERPs
We dissected 30 SERPs: 0 were ten blue links; 97% had People Also Ask and video, 77% an AI Overview, 50% ads; ranking #1 no longer means top of the page.
read the study → · n = 30 keywords / 9 niches · by Dr Blaze - How Common Are AI Overviews? We Checked 30 SERPs
23 of 30 SERPs (77%) we checked showed an AI Overview, every one cites sources, and search intent decides which SERPs get one.
read the study → · n = 30 keywords / 9 niches · by Bobby K - Is the CTR-by-Position Curve Real? We Tested 1,000 Rankings
Top-3 roughly tracks the curve, but on ~1,000 real rankings positions 11-20 out-click 4-10 (15.6% vs 8.7%); the canonical CTR curve is a population artifact.
read the study → · n = ~1,000 ranking queries · per-position n = 2-80 · by Dr Blaze - Does Programmatic SEO Work? We Analyzed 12 Million Pages
Across 12.3M programmatic pages on nine DR 89-94 sites, 96.7% get ≤100 organic visits/month and 29% get exactly zero. pSEO is a power law. Here's what actually separates …
read the study → · by Bobby K - Is Keyword Difficulty Accurate? We Audited 30 SERPs
A psychometric validity test of Keyword Difficulty: across 30 SERPs and 241 ranking pages, KD predicts page-level links well (ρ=0.91) but barely tracks domain authority (…
read the study → · n = 30 keywords · 241 ranking pages · by Dr Blaze
measurement-literate SEO
We measure
the construct.
A ranking report can look healthy while pipeline stays flat. We decide first which numbers track revenue for your business, then do the work. Tell us your numbers and we’ll tell you which ones mean something.
ask the lab
Which of your numbers can you trust?
Send us the metric your team reports on, or the forecast in your deck. We’ll tell you what it measures and what it misses.
received
Message sent ✓
We’ll get back to you fast, usually the same day.