Maple Terroir had a page ranking at position 7.3 for "best Canadian maple syrup." 204 impressions a month. 1.5% CTR. The page was good. The content was accurate, well-researched, and the brand voice was right. Every instinct said optimise it: tighten the meta title, add a few internal links, refresh the intro, push it from 7 to 3.
I built a second page instead. The original kept doing what it was already doing. The new page started ranking within weeks. Combined, they pulled more traffic than the optimised version of the original ever could have.
This decision, build versus optimise, is one of the most consequential calls I make on a client account. It is also the one most agencies get wrong, because the default move is to protect existing authority. That default is why so many sites have 1 page slowly improving when they could have 2 pages each winning their own SERP.
The Signal I'm Actually Reading
The question is never "is this page good?" The question is "does this page match the format the SERP is rewarding?"
For "best Canadian maple syrup," I opened the top 10 results and classified each one. 8 out of 10 were listicles. Brand comparison roundups. "Top 7 Maple Syrup Brands" type pages. The Maple Terroir page was an educational guide explaining grading, terroir, and production methods. Excellent content. Wrong format for the keyword.
That is not a problem you fix by optimising a meta title. The page was being shown to searchers who wanted to compare 7 brands and pick 1. They landed on a 2,000-word essay about sugar bush ecology. They bounced. Google read the bounce, the short dwell time, and the low CTR, and capped the page at position 7.
The Two Forces Pulling You Toward Optimisation (And Why They're Wrong)
The first force is sunk cost. The original page took research, drafting, photography, internal linking. Throwing more weight behind it feels like respecting that work. It isn't. The work is preserved. The page keeps existing. It keeps ranking for the queries it actually matches, which in Maple Terroir's case were educational queries around grading and terroir.
The second force is the fear of cannibalization. "If I build a second page, won't they compete?" Only if they share an intent. I run a SERP overlap check before I even brief the new page. If 3 or more URLs in the top 5 rank for both keywords, that is 1 page. Different SERPs, with different page types ranking, means 2 pages serving 2 different jobs. Maple Terroir's educational guide and the listicle had no overlap in the top 10. They were different products in Google's eyes.
The Decision Framework I Use
Before I touch a page that's stuck, I run 3 checks. They take 20 minutes and they decide whether the next 10 hours of work go into a rewrite or a new build.
| Check | Optimise the existing page | Build a new page |
|---|---|---|
| SERP format | Top 5 results match the page's current format | Top 5 results are a different format than the current page |
| Intent overlap with target keyword | Page already matches what the searcher wants | Page answers a different question than the keyword implies |
| Authority signals | Position 7-20, real impressions, some ranking history | Page is ranking for adjacent terms but not the target |
| What the existing page is winning | Nothing meaningful, or only the target keyword | Already ranking well for related queries that fit its format |
Maple Terroir failed the SERP format check and passed the "already winning something" check. Optimising would have cost the educational rankings without guaranteeing the listicle ranking. Building preserved both shots.
What "Building a New Page" Actually Looks Like
This is the part most teams skip. A new page is not a duplicate with a different angle. It is a structurally different artifact built from the SERP up.
For the listicle, I opened the 8 ranking listicles and classified the structure of each. Number of brands featured. Whether they linked to the brand's site or kept the click on-page. Whether they used a comparison table. Whether they ranked by price, quality, or category. The pattern was clear: 7 brands, on-page comparison table, brief tasting notes, transparent ranking criteria.
The new page matched that structure and added 1 thing the others didn't have: actual sourcing context for each brand, because Maple Terroir is a syrup company and the depth on producer relationships is something the affiliate listicles couldn't fake.
That last part is the test I apply to every section we write: could you paste this onto a different listicle without anyone noticing? If yes, the section is filler. The producer-context sections passed the test. The generic "how to choose maple syrup" intro got cut, because every other listicle had the same 200 words.
The Result Across the Engagement
Maple Terroir's site-wide organic traffic grew 1,500% over the engagement. Not all of that came from the listicle decision, but the principle, build a new page when the format is the constraint, applied across multiple keyword clusters on the site. We made the same call for product comparison queries, recipe queries, and gift-related queries. Each cluster got its own format, its own page, its own SERP fight.
The original guide is still ranking. It is still doing the job it was built to do. The new pages are doing jobs that page was never going to do, because Google had already decided what shape those queries deserve.
According to Conductor, 25% of Google searches now show an AI Overview, up from 13% 6 months prior. AI Overviews pull from the pages that match the question structure best, not the pages with the most authority. Format match matters more in 2026 than it did in 2024, not less.
The Takeaway
Stop trying to make 1 page win every related keyword. Open the SERP. Classify the top 5. If the format your page is in does not appear in those 5, you are not 3 positions away from the top. You are in the wrong race.
Optimisation moves a page that's already in the right race. A new page enters a race the existing one was never built for. Knowing which problem you have is the difference between work that compounds and work that disappears.
Liam Lytton is the founder of The 66th, an SEO and GEO agency in Vancouver that has driven results including 1,500% organic traffic growth, 5x revenue, and 4x lead volume for clients across North America.
Results vary based on domain authority, competition, content quality, and execution consistency. Past client results are not a guarantee of future performance.