On-page at scale

IntermediateDuration ~15 min video + 90 min hands-onTools A site crawler (Screaming Frog free tier), A spreadsheet, Search Console

Why this lesson

In Level 1 you optimized one page by hand. That works when you have ten pages. It falls apart at a hundred, and it’s impossible at ten thousand. Real sites are built from templates — a product template, a blog template, a category template — and a single template stamps out title tags, meta descriptions, and heading structures across thousands of URLs. Fix the template and you fix every page it produces. Get the template wrong and you’ve broken every page at once. This lesson is about working at that altitude: patterns instead of pages, plus the three maintenance jobs that keep a large site healthy — killing self-competition, cutting dead weight, and reviving content that’s decaying.

Watch

Watch for: Sam Oh's point that the top-ranking page ranks for ~1,000 keywords, not one — so the target keyword is only a slice of the traffic. Note the on-page elements he treats as repeatable patterns: target keyword in the title (or a close variant when the angle changes), a short descriptive URL slug, meta descriptions that Google rewrites ~63% of the time, internal links, and image alt text. Every one of these is something you set once in a template, not per page.

The explainer

The shift: pages become patterns. A big site isn’t a pile of hand-crafted pages — it’s a few templates repeated. So on-page work at scale means template optimization: editing the rule that generates an element rather than the element itself. Instead of writing 2,000 titles, you write one title pattern — {Product name} — {Category} | {Brand} — and every product page renders a unique, keyword-relevant title from it. Do the same for meta descriptions, H1s, and URL slugs. The on-page principles from Level 1 don’t change; you just apply them one level up, in the CMS or template file, so a single edit propagates everywhere. The catch is that a template mistake also propagates everywhere: one broken pattern can wreck a thousand pages, so you validate on a sample before you ship.

Find the pages, then find the pattern breaks. You can’t eyeball ten thousand pages, so you crawl. A crawler like Screaming Frog pulls every URL’s title, meta, H1, status code, and word count into one table. Now the outliers jump out: missing titles, duplicated titles across pages, empty metas, multiple H1s, thin pages. You sort the column, spot the pattern that’s failing, and fix it at the template — or export the offending URLs and hand them to a developer. This is on-page at scale in one sentence: crawl to see the whole site as data, fix the rule, re-crawl to confirm.

Kill the self-competition: cannibalization. When two or more of your own pages target the same keyword and intent, they fight each other — Google isn’t sure which to rank, they trade positions, and each dilutes the other’s signals. That’s keyword cannibalization, and on a content-heavy site it’s everywhere. Find it by searching site:yourdomain.com "target keyword" or by grouping your crawl/Search Console data by query and spotting keywords where several of your URLs rank. The fix is almost always consolidation: pick the strongest page, merge the useful content into it, and 301-redirect the weaker ones into the winner so their links and authority flow to one URL instead of being split.

Prune the dead weight. Not every page deserves to exist. Content pruning is auditing your pages and removing or consolidating the ones that get no traffic, earn no links, and serve no user — thin tag archives, expired listings, near-duplicate posts. Cutting genuine dead weight can raise overall performance: it concentrates your site’s authority and crawl attention on the pages that matter. Prune by redirecting (if the URL has any equity or backlinks) or by returning a clean 410 Gone for pages that should simply disappear.

Refresh what’s decaying. Rankings rot. A post that ranked #3 in 2024 slides down as competitors publish fresher, more complete pages and as its facts go stale. A content refresh — updating the information, re-matching the current search intent (re-read the SERP), adding sections newer winners cover, and re-publishing — usually recovers a decaying page faster and cheaper than writing a new one, because the page already has age, links, and history. On a large site, systematically refreshing your decaying-but-valuable pages is often the single highest-ROI content activity there is.

Do it

1. Crawl your practice site with Screaming Frog (free tier, ≤500 URLs). Export titles, metas, H1s, status codes, and word counts to a spreadsheet.
2. Sort and scan for pattern breaks: missing/duplicate titles, empty metas, multiple or missing H1s, thin pages. Identify which template each break comes from.
3. Write a corrected title/meta pattern for one template (e.g. {Page topic} | {Brand}) and note where in the CMS it would be edited.
4. Hunt cannibalization: pick two target keywords and run site:yourdomain.com "keyword". Where two pages compete, decide the winner and plan the 301.
5. Build a prune list: pages with zero traffic, zero links, zero purpose. Mark each as redirect or 410.
6. Pick one decaying page and draft a refresh plan: what facts to update, what the current SERP now rewards, what to add.
7. Log all of it in the Level 2 workbook’s on-page-at-scale audit tab.

Level 2 workbook — crawl-audit sheet, cannibalization finder, prune list & content-refresh plannerlevel-2-workbook.pdf116 KBOriginal course material — free to use

Terms introduced

Terms introduced in this lesson

• On-page at scale
• Template optimization
• Keyword cannibalization
• Content pruning
• Content refresh

Each term links to the glossary, where it is defined by the way you just used it.

Check

Check yourself

1. On a 2,000-page e-commerce store, what is the smart way to fix title tags?

   Fix the template that generates titles — write one pattern like "{Product} — {Category} | {Brand}" so every page inherits a good title at onceHand-write a unique title for every one of the 2,000 pagesIgnore titles on product pagesCopy the homepage title onto every page

2. You find three of your own blog posts all trying to rank for "best running shoes," and they keep swapping positions in the SERP. What is happening and what do you do?

   Add the keyword more times to each pageDelete all three pagesThis is keyword cannibalization: the pages split signals and compete with each other. Consolidate them into one strong page (301 the losers into the winner) or clearly differentiate their intentNothing — more pages is always better

3. A page that ranked #3 two years ago has slid to #14 and its traffic is falling. What is the first move?

   Immediately delete itRefresh it — update the facts, re-match current search intent, add what newer competitors cover, and re-publish; a content refresh often recovers rankings faster than a brand-new pageChange the domain nameBuild 50 backlinks to it

Check answers

You can move on when you can… crawl a site, spot on-page pattern breaks and fix them at the template, find and resolve keyword cannibalization by consolidating pages, build a defensible prune list, and plan a content refresh for a decaying page.

Mark this lesson complete

Go deeper

• Ahrefs — “On-Page SEO: The Beginner’s Guide”: the full element-by-element checklist behind the patterns you’re now applying at scale.
• Next: 2.3 · Technical SEO I — crawl & index control — control exactly what search engines are allowed to crawl and index.