Tips
SEO practices worth keeping
Practical checks for useful content, clear links and accessible pages, with sources for the technical guidance.
Correction, 23 September 2026: the canonical advice now distinguishes preferred pages from duplicates. A duplicate should name its preferred URL, which can be a different page.
These are the practices we use to keep pages useful and technically accessible. Check their effect against your own search data.
Keep one page per question
The clearest signal you can send about a page is that it is about one thing. Everything downstream gets easier: the title writes itself, the internal links have an obvious anchor, and the page does not compete with a sibling.
When a page starts answering a second question well, that is a new page.
Keep the answer near the top
A reader who has to scroll to find out whether the page is relevant usually does not scroll. The same structure that serves them is what gets a passage quoted in a featured snippet or an AI answer.
Keep the technical floor clean
Give each page a clear title and description. A preferred page can name itself as canonical. Duplicates should name the preferred URL, following Google’s canonical guidance. Avoid redirect chains on your most requested URLs. Put useful content in the HTML. Sitemap listing only pages that return 200 and are indexable.
None of this ranks anything on its own. All of it removes reasons for a page to be discounted.
Keep the internal links deliberate
Decide which pages are important. Then link to them from the body copy of the pages people actually read, using anchor text that says what the destination is.
This is the only part of link equity distribution that is entirely yours.
Keep dates honest
Publish dates and update dates, visible, accurate. Updating the date without updating the content is the version of this that gets noticed, and it costs trust with readers who checked.
Keep the site usable on a phone in bad conditions
Contrast that works in sunlight. Touch targets you can hit while walking. Text that does not reflow after you have started reading it. Pages that load on a mobile network.
These are accessibility requirements. They also happen to be a large part of what page experience measures.
Keep sources attached to claims
A page that cites what it is standing on can be checked. A page that cannot be checked is competing on confidence, and confidence is cheap to produce.
What has not survived
Keyword density targets. Exact-match anchor text at scale. Article spinning. Private blog networks. Meta keywords. Directory submission campaigns. Buying links from a marketplace.
Google lists several of these practices in its spam policies. Effects depend on the practice and the evidence. There is no universal penalty described by this list.
The common thread is that they optimise the measurement instead of the thing being measured. Google separately confirms that it ignores meta keywords.
The practice underneath all of them
Be the most useful result for a specific question asked by a specific person.
That sounds like a platitude until you apply it to a page you have already written. At that point it usually produces a list of concrete edits.
If you want that applied to a site with a real history behind it, that is what our SEO work does.