Most "on-page SEO checklists" repeat the same five generic tips. Here are the fixes that I have actually seen move rankings and crawl coverage on real client sites.
1. One H1 per page, and it should mean something
A surprising number of sites either skip the H1 entirely or duplicate the same H1 across every page. Every page needs exactly one, and it should describe that specific page — not the whole site.
2. Unique title tags and meta descriptions on every page
Duplicate `<title>` tags across a whole site are one of the most common issues I find on audits. Every page — including thin utility pages — needs its own title and description written for what that page actually offers.
3. Image alt text that describes, not stuffs
Alt text should describe the image for someone who cannot see it, ideally including relevant context naturally. "Screenshot of the [Client] homepage" is useful; repeating the same keyword ten times across ten images is not — and can actively hurt you.
4. A real robots.txt and XML sitemap, kept in sync
I have found sites where the sitemap listed pages the robots.txt blocked, or where a stray `noindex` meta tag was left on from a staging environment. Both quietly stop pages from ever getting indexed. Check both files after every deploy.
5. Structured data (JSON-LD)
Adding Person, Organization, Product, FAQPage or BreadcrumbList schema does not guarantee rich results, but it gives search engines unambiguous facts about the page instead of making them guess from prose — and it costs nothing to add.
6. Internal links with descriptive anchor text
"Click here" tells search engines nothing about the destination page. Linking with the actual topic as anchor text ("see our Shopify development services") reinforces what that target page is about.
7. Page speed is an SEO fix, not just a UX one
Compressing images, deferring non-critical JavaScript and cutting unused CSS is standard performance work — but it is also directly rewarded in ranking signals tied to Core Web Vitals. I run a Lighthouse pass on every project before launch, not after something breaks.
None of these are exotic. They are just easy to skip under deadline pressure — which is exactly why checking them systematically before launch pays off.