A founder emailed me last month with a 3,200 word service page his previous agency had built. It ranked position 47 for the keyword it was targeting. He wanted to know what to fix.
I opened the search results for his keyword. Every page ranking on page 1 was a listicle. His page was a service page. Nothing on that page was going to move it, because it was the wrong shape for the search.
That is what the two searches test catches. I run it before I write a single word.
The test itself
Open a fresh browser. Type the keyword. Read what is ranking. Then type the closest variant a real buyer would use and read that too. Two searches, maybe 15 minutes of clicking, and you know 3 things you cannot know from a keyword tool.
You know the format that Google has already decided wins for this search. You know the depth and angle the top pages are covering. And you know whether the searcher is actually the buyer you want, or somebody 4 steps upstream who will bounce off your pricing.
Keyword tools tell you volume and difficulty. They do not tell you any of that.
The 3 checks I run every time
Before I place any keyword on any page, I run 3 checks. Every SEO architecture mistake I have seen traces back to skipping 1 of them.
Check 1: Does the keyword match the product and the location in reality? If a plumber in Burnaby is targeting "emergency plumber Vancouver," the keyword and the service area do not line up. Google can see the disconnect, and so can the searcher who lands and bounces.
Check 2: Would the searcher be satisfied landing on this page? Somebody typing "how much do dental implants cost" is not ready to book. If I send them to a booking page, they leave. The keyword and the page's job have to line up.
Check 3: Does the SERP confirm the intent? This is the two searches test. Google has spent years testing what searchers actually click. The results are the answer. Ignore them and you are guessing.
What happened with AetherHaus
When we started with AetherHaus, they had 40 monthly organic visitors. Beautiful brand, sharp editorial voice, a full-navigation site. The content depth was not there and the pages were built to the wrong shape for what people were searching.
The keyword "cold plunge Vancouver" was the target for their money page. I ran the two searches. The top of the results showed a mix of local-pack listings, a couple of service pages from studios, and 1 or 2 reviews/roundups. That confirmed a service page was the right format, but it also told me the page needed local proof, specific amenities, and clear booking, not a wall of copy about the health benefits of cold exposure.
We rebuilt to that shape. Over 9 months, organic traffic went from 40 to 4,800 monthly visitors. AetherHaus hit position 1 for "cold plunge Vancouver," 89 AI citations accumulated across ChatGPT, Grok, AI Overviews, Perplexity, and Copilot, and revenue lifted 50% within 2 weeks of ranking 1. You can read the full AetherHaus case study for the sequence.
None of that is a story about better copywriting. It is a story about matching the page to what the SERP had already validated.
The mistake most agencies make instead
They read the keyword tool. They see volume. They write a page. They wait 3 months. Nothing moves. They tell the client SEO takes time.
What actually happened is the page went out the wrong shape. It never had a chance. And now the client is 3 months deeper into a retainer with no signal that the strategy is right.
Format match beats page authority. If the results are all 1 format and your page is a different format, optimising the old page is fighting gravity. Better to build new and let each page serve its own intent. That single check (what format is winning?) is the difference between position 3 and stuck-at-7-forever.
What the 2 searches actually cover
Here is how I structure the 15 minutes.
| Search | What I am looking for | What it changes |
|---|---|---|
| Primary keyword, exact | Format winning (service, listicle, blog, video, local pack). Page depth. Buyer stage. | Decides the page type and length before I write. |
| Closest buyer variant | How real buyers phrase it. Whether the intent shifts. Which competitors appear on both. | Confirms whether 1 page can win both, or if I need 2. |
| Both together | AI Overview presence. Local pack presence. Video carousel. People Also Ask. | Tells me which trust signals the page needs to earn a citation. |
The whole check takes less time than a keyword-tool export. It saves months.
Why volume is the least important column
Volume is the number I care about least in keyword research. A keyword with 50 searches a month that is easy to win and brings in actual buyers is often more valuable than 1 with 5,000 searches that is impossible to rank for.
AI-referred traffic converts roughly 4x organic (Previsible, 2025), and that traffic goes to pages that answer specifically. High-volume pages that answer generically get scrolled past. Low-volume pages that answer precisely get cited, clicked, and converted.
The SERP tells you which is which. The keyword tool cannot.
How to run this yourself today
Pick the keyword you most want to rank for. Open an incognito window (so your search history does not skew the results). Type the keyword. Do not click anything for 60 seconds. Just read.
Ask 4 questions. What format dominates? How long are the top pages? What is the assumed buyer stage? Are AI Overviews or a local pack pushing the organic results down?
Then run the second search with the closest variant a real buyer types. See what changes. If the format shifts, that is 2 pages, not 1. If the format holds, 1 page can serve both.
That is the test. It costs 15 minutes. It answers the question no keyword tool can.
The bigger point
Most SEO advice tells you to research keywords, plan a content calendar, and start writing. That order skips the only step that matters: reading what has already won.
The pages on page 1 are there because they were tested against real searchers and won. You do not invent structure from scratch. You read what won and build something better with your real edge. Writing what feels right is guessing, and you are usually wrong, because your instinct is what you would want, not what the searcher wants.
Run the 2 searches. Every time. Before the outline, before the brief, before the writer opens a document. That 15 minutes decides whether the next 3 months of work compounds or evaporates.
Liam Lytton is the founder of The 66th, a Vancouver-based SEO and GEO agency working with founder-led businesses across North America. He writes about the operational decisions that separate SEO work that compounds from work that stalls.
