A prospect sent me a blog post last month and asked why it wasn't ranking. It was well-written. The keyword was in the H1. The word count matched what an SEO tool told them to hit. They had built 4 internal links to it.
I ran the search. The top 10 results were all comparison listicles. Their page was a 1,800 word explainer. It was never going to rank, no matter how many links they pointed at it.
That is the single most common mistake I see when a client hands me a site to audit. The page is fine. The page is just answering a question the searcher isn't asking on that keyword. And nobody read the search results before writing.
So I built a rule for our team. Before a writer opens a doc, before a keyword makes it into a content plan, before I quote anyone on a proposal, someone at The 66th spends 45 minutes reading the top 10 ranking pages. Every one. Fully. Not skimming the titles in an SEO tool.
What 45 minutes of reading actually catches
The reason this works is that Google has already run the experiment for you. The pages on page 1 are there because real searchers tested them and stayed. If you open them and read them, you are reading the answer key.
Three things surface almost every time you do this properly.
1. The format. Are the top 10 all service pages? All listicles? All blog explainers? All video results? A mix? The format that dominates the SERP is the format Google has decided the searcher wants. If you build a different format, you are fighting gravity, no matter how good your content is. This is the single biggest reason a striking-distance page stays stuck at position 7.
2. The buyer stage. The pages tell you what stage of the funnel the searcher is in. Educational content on page 1 means the searcher is researching. Product pages on page 1 means the searcher is ready to buy. Comparison pages mean they are shortlisting. The keyword itself does not tell you this. The SERP does.
3. The depth ceiling. Reading the top 10 tells you what "good" looks like for this keyword. Sometimes the ceiling is embarrassingly low, meaning a well-built page can win with modest effort. Sometimes the ceiling is a 6,000-word ingredient breakdown with 12 diagrams and 3 case studies, meaning you should not touch the keyword unless you can beat that.
The AetherHaus keyword that looked wrong on paper
When we started with AetherHaus, an SEO tool would have told you "cold plunge Vancouver" was a modest keyword. The volume looked small. There were bigger, sexier terms in the sauna and wellness category. If we had planned by volume alone, we would have pointed the site at the biggest number and lost.
Instead, I opened the top 10 for "cold plunge Vancouver" and read every page. Two things jumped out.
First, the SERP was a mix of directory listings, one thin blog post, and 2 studio pages. Nobody had built a proper destination page for that exact search. The ceiling was low.
Second, everyone ranking was answering a slightly different question than the searcher was asking. The searcher wanted to book a cold plunge in Vancouver, tonight. The ranking pages were mostly "what is cold plunge" content or general wellness directories. That is a format-and-intent gap you can drive a truck through.
We built a landing page purpose-built for the booking intent. It ended up at position 1 for "cold plunge Vancouver" and traffic went from around 40 monthly visitors to 4,800. Revenue lifted 50% within 2 weeks of hitting the top spot. The read is what caught it. A tool would have told me to chase a bigger number and miss.
What we look at in each of the top 10
The 45 minutes is not free-form reading. I have our team follow the same pass for every URL so we end up with a comparable read across the 10.
| What we check | Why it matters |
|---|---|
| Page type (service, listicle, blog, video, directory) | Tells you the format Google has picked as the winner. Build the same format or don't compete. |
| Who the page is written for | The language register reveals the buyer. A page written for practice owners reads differently from one written for patients. |
| Stage of the funnel | Determines whether you send a service page, a comparison page, or a blog post to this keyword. |
| Depth and structure | Sets the ceiling. Sections, subheads, embedded assets, tables. If the winner has all of them, you need more. |
| Where the page is thin or wrong | The gap you can exploit. Missing local proof, dated info, no comparison, no pricing signal. |
| What kind of proof is on the page | Reviews, case studies, video, before/after, testimonials. Match or exceed. |
Every page gets 4 or 5 minutes. 10 pages is roughly 45 minutes. It is boring. It is also the difference between a plan that works and a content calendar that quietly bleeds budget for 6 months before anyone admits it isn't moving.
The one shortcut I actually endorse
You do not have to do this on every keyword forever. Once you have read the SERP for the 8 to 12 core money keywords in a category, you start to see the shape. The insurance category behaves one way. Local trades behave another. DTC food is different again.
What you cannot skip is doing it on the ones that matter. The pages that will represent the bulk of your revenue capture. Those get the full 45 minutes, and I still do them myself for our biggest clients rather than delegate.
Search Engine Journal has been reporting for a while now that intent mismatch is one of the largest single causes of underperforming content, more than word count or backlink volume (Search Engine Journal, 2024). That matches what I see across every audit we run. The pages that fail almost never fail because they are bad. They fail because they are aimed at a version of the keyword that does not exist.
Why AI search rewards this even more
The habit gets more valuable, not less, as AI search takes a bigger share of how people find things. AI models are pulling from the pages that already rank, plus the community and review sources that reinforce them. If your page is the wrong format for the query, you are not just missing the top of Google, you are missing the source pool that ChatGPT and Perplexity draw from when they answer.
We saw this at scale with a DTC client, Maple Terroir. Organic traffic grew 1,500% and AI citations grew 3,300%, from the same underlying discipline. Read the SERP. Match the format. Build depth the winners do not have. AI citations came as a byproduct because the foundation was right.
Gartner has projected that traditional search volume will drop 25% by 2026 as more queries move to AI answer engines (Gartner, 2024). The businesses that survive the shift are the ones whose pages were the right format for the right question in the first place.
The rule, in one line
You do not get to write for a keyword you have not read for. If you have not opened the top 10 and read them properly, you are guessing. Guessing loses to reading, every time. If you want to see how this discipline lives inside a full engagement, our SEO services page walks through where the read fits in the build.
The 45 minutes feels expensive. The 6 months of a page that never ranks are more expensive. Pick which one you'd rather pay.
Liam Lytton is the founder of The 66th, a Vancouver-based SEO and GEO agency. He works founder-direct with owner-operators, DTC brands, and venture-backed startups building organic search as a compounding asset.
