Kyle Roof: Testing What Actually Ranks

Testing What Actually Ranks

Written for website owners who cannot get found. Not here: what to do before you have an offer. Traffic + Offer

Kyle Roof did something that offended a lot of SEO professionals. He took the most sacred belief in the industry, that ranking requires knowing what you are doing, and put it on a test bench. He built a page of Lorem Ipsum gibberish, optimised the on-page elements mathematically, and ranked it. His reading of that result was not that quality is irrelevant. It was that SEO is testable, measurable, and a good deal less mysterious than the industry likes to pretend.

Who They Are

Kyle co-founded High Voltage SEO and created PageOptimizer Pro, a tool that reverse-engineers the top-ranking pages for a query to work out which signals are being rewarded. He represents the empirical branch of SEO, the branch that holds the algorithm can be measured and tested rather than merely theorised about.

The Lorem Ipsum experiment became famous because it threatened the comfortable idea that content quality is what Google is measuring. The target query was "rhinoplasty Plano", nose surgery in a Texas city, and the method was arithmetic. He calculated the keyword counts and placements from the pages already ranking at the top, built a page of placeholder gibberish that hit the same numbers, and watched it climb. No meaning in it, no expertise behind it, not one readable sentence. Just the maths. His own argument from it was narrow, and it is worth repeating accurately: the algorithm, at least some of the time, will reward hitting specific mathematical targets over any grasp of what your content actually says.

What They Teach

The core methodology is the Scientific On-Page Method, and the process runs like this. Take a keyword or search query, pull the top ten ranking results, then analyse them systematically across dozens of on-page variables: word counts, heading structure, keyword density, paragraph length, semantic variations, outbound links, internal link context. Kyle has a name for the industry default of following opinions and second-hand advice. He calls it GuesSEO, and his whole method is a refusal to do it.

Then you find the patterns. Maybe the top three results all carry a heading with a semantic variation the rest are missing. It might be one particular word the competitors never reach for, or a structure of bullet lists where everyone else runs paragraphs. Those are not opinions. They are measurable patterns sitting in the data, and anyone who repeats your analysis can check them.

Then you apply the patterns to your own content. Your work is not just writing something good, it is writing something that matches the mathematical profile Google is rewarding for that particular query. This is not about gaming a system. It is about reading what the system is looking for and then supplying it.

PageOptimizer Pro automates the analysis. You put in a keyword, the tool pulls the top results, works across the variables, and returns a profile for you to target. The recommendations are concrete rather than directional. A Shopify merchant gets an instruction as specific as adding a keyword a certain number of times to a product description, and a B2B marketer gets structural adjustments for a landing page. The discipline underneath it is deconstructing your competitors mathematically rather than thematically: the words and the counts, not the vibe.

The other insight Kyle emphasises is single-variable testing. Do not change ten things and hope your ranking improves. Change one thing, measure it, and see whether the ranking moved. If it did, you have a signal. If it did not, that variable does not matter for this query. His own controlled setup is stricter than most practitioners would bother with, using identical pages and often invented keywords with no existing search results, so nothing but the test variable can explain a movement. He holds a US patent for the method. This is how you build knowledge instead of folklore.

His thinking also runs to inverse testing. Test what happens when you violate the pattern: remove the heading structure, hold everything else steady, and watch whether the ranking drops. In the controlled experiments it goes the other way too. Apply the winning change to the control pages, then see whether the original test page loses its lead, and if it does, you have confirmed the variable twice. Too many people see a correlation, that top-ranking pages have longer content, and assume a cause, that longer content ranks. Testing is what separates the two.

How It Maps to Opportunity and Authority

Kyle's work is heavily weighted toward Opportunity. He teaches you how to match what the algorithm is explicitly rewarding in a particular result, which is not about building trust or brand or credentials. It is on-page mathematical matching. Even where the testing brushes against trust, identifying signals that correlate with E-E-A-T assessment such as author mentions, particular phrasing, or the schema types on your ranking competitors, the play stays algorithmic. Implement what the data says correlates, and let the algorithm draw its own conclusion.

The Authority component is indirect. Ranking for a keyword means visibility, and matching the profile the algorithm rewards increases that visibility. But it only works on top of some baseline of credibility. An untrustworthy site does not get pushed to the top by perfect on-page maths, no matter how precisely you match the profile. So Authority is your foundation, and Kyle's work is the optimisation laid over it.

Think of it this way. Authority is what makes Google willing to consider you at all. Opportunity is what makes you the obvious choice for the query once you are being considered.

The Strategy Breakdown

Kyle's strategies lean hard on one lever. Here is the honest split, so you know what each one buys and what it does not.

Single-variable and inverse testing

Controlled experiments isolating one ranking factor at a time: identical pages, one change, movement measured against controls, then the inverse test to confirm the result was not luck. Opportunity impact: high. This is how you discover which specific on-page signals the algorithm is using in your niche, whether that is a keyword in the H1, a term in the title tag, or a particular content structure. Once you know which variables move rankings, on-page targeting stops being generic advice and becomes surgical. Authority impact: moderate, and strictly algorithmic. Testing surfaces the factors the algorithm correlates with quality, E-E-A-T-adjacent signals included, and satisfying them influences how the algorithm assesses your fitness for the query. It does not build the user-side trust that comes from being credible. Technical SEOs respect the rigour of the isolation while noting that Google's live algorithm is more entangled than any single test can capture.

The Scientific On-Page Method and PageOptimizer Pro

The operational framework built on those test results, packaged as a tool. POP analyses the top-ranking pages for a keyword across dozens of quantifiable factors, calculates the profile they share, and hands over specific targets: how many times to use a term, where to place it, what structure to match. Opportunity impact: very high. I think this is the most direct opportunity capture in the Expert Series, because it supplies the algorithm with the relevance signals it appears to reward for that query, with numbers instead of hunches. It pairs naturally with keyword research that has already identified queries you can realistically win. Authority impact: moderate, algorithmic only. A high POP score means your page matches the mathematical profile of pages the algorithm already treats as fit for the query, including the schema types and trust-signal placements your competitors carry. That satisfies the algorithm's criteria. It does not build brand, links, or reputation, which is why a B2B marketer can tune a landing page with POP and still need an entirely separate strategy for topical authority.

Test-derived trust signals

Deliberately adding the elements testing shows correlate with rankings and E-E-A-T assessment: author mentions, specific phrasing, structured data matched to what ranking competitors use. Opportunity impact: low. These signals are about convincing the algorithm you deserve a ranking, not about matching your site to more queries. Authority impact: moderate, with a hard ceiling. You are optimising the algorithm's reading of your trustworthiness on the strength of correlation, and that can move rankings. What it cannot do is create the user-side authority, the repeat visits and brand searches and links earned because people trust you, that AJ Kohn builds toward from the opposite pole of this framework. Kyle optimises for what the algorithm measures. Kohn optimises for the user the algorithm is trying to model. Knowing which pole you are working from is part of using either one well.

The Lorem Ipsum experiment

Not a strategy but a demonstration, and worth rating precisely because of what it exposed. Opportunity impact: none directly. Nobody should publish gibberish. The experiment was a proof that the mathematical signals the method targets are real, and strong enough to carry a page with no meaning in it. Authority impact: none directly, but the lesson is the point. That query laid bare the gap between algorithmic Authority, meaning the mathematical criteria, and real authority, the kind a human confers by trusting what they read. A page of placeholder text can hold the first for a while. It can never hold the second, and everything built with these methods still has to survive a person clicking through.

When to Learn From Them

Your diagnostic will point you toward Kyle if your problem is Opportunity specifically. You have created content, it might even be good content, and it is not ranking for the keywords that matter to your business.

He is also relevant if you understand SEO conceptually but have no way to test whether that understanding matches reality. A great deal of the field follows advice from blogs without ever checking whether it moves anything in their own niche.

If content is loved by readers and performs on social media while search visibility refuses to follow, this approach will diagnose it. Perhaps the keyword targets were too hard, or the on-page profile does not match what is being rewarded. Testing says which of the two it is, and that is more than an opinion will ever do.

And when time is short and every piece of content has to count, the methodology at least aims the effort at patterns that hold in a specific niche rather than at generic advice.

Where to Start

PageOptimizer Pro is the practical entry point, since it does the analysis automatically. Audit your current top pages with it, see what profile they match, and see what is missing relative to the competitors above you. That gives you a concrete place to begin rather than a vague sense that something is off.

Wherever he walks through the Scientific On-Page Method in a worked example, that is the thing to watch. He pulls a keyword, shows the analysis, shows the pattern, then shows the changes he would make, and watching the reasoning happen is worth more than a summary of the method.

The Lorem Ipsum experiment is worth reading for the philosophy rather than the tactic. His argument is that Google, at least some of the time, is reading for patterns rather than for meaning in the way a person does. Sitting with that shift in thinking is more valuable than any single optimisation tip.

The recurring theme across his public work is testing against guessing. Listen for the question underneath it: what have you actually tested, as against what you assume because you read it somewhere. That question is the whole difference between scientific SEO and folklore.

One critical point he emphasises. The approach assumes the content is already publishing-quality and relevant to the audience it was written for. The Scientific On-Page Method optimises for search visibility, and it does not replace the work of creating something worth ranking in the first place. Where the content is weak, no amount of on-page optimisation creates a durable ranking. Where the content is good and still not ranking, this methodology will tell you why and what to change.


Part of the Expert Series. Back to the framework or the diagnostic. Part of the Marketing Universe. Explore Traffic Plus Offer : The Trust Algorithm : 4-Quadrant AI. Read the book: Marketing Curious: Working the Noise.

One link on this page pays me a commission if you buy. The full list, with what I earn on each, is at timmybrown.co.nz/links.