What are the tips and tricks for SEO?

Most SEO tip lists have the same problem: they are flat. Thirty tactics presented as equals, with no indication of which take ten minutes and which take a year, or which stopped working in 2019 and are still being copied between blog posts.

So these are sorted by how long they take to pay off, and there is a section at the end for the ones that now actively cause damage.

The rule for judging any SEO tip

Before the list, the filter that makes the list unnecessary.

Search engines are trying to identify pages that genuinely satisfy the person searching. They are imperfect at it, and every gap between what they measure and what they intend to measure creates a trick. Tricks work until the gap closes.

So: a tactic that also serves a reader has an indefinite shelf life. A tactic that only makes sense as a manipulation of a measurement has a shelf life measured in months.

Faster pages, clearer titles, direct answers and better structure keep working because they are what the algorithm is trying to approximate. Keyword density, link exchanges and doorway pages stopped working because they were only ever proxies being gamed.

Apply that test to anything below, and to anything you read anywhere else.

Wins inside an hour

1. Rewrite titles on high-impression, low-CTR pages. Export Search Console by page, sort by impressions, look for anything with a click-through rate well below 2% at a decent position. Those pages rank and fail to convert the ranking. A rewritten title can move that inside a fortnight — the single fastest measurable change in SEO.

2. Put the answer in the first hundred words. Find your five most important pages and check whether the target question is answered before the first subheading. If not, move the answer up. This wins featured snippets and is the main determinant of whether an AI assistant can extract you.

3. Check your indexed count. Search Console, Pages report. If far fewer pages are indexed than you publish, stop reading tip lists — you have found your actual problem, and it outweighs everything else here.

4. Fix your internal anchor text. Search your site for “click here” and “read more” and replace them with descriptions of the destination. Anchor text is one of the clearest relevance hints you control, and these throw it away.

5. Add one contextual internal link to each priority page. From your strongest existing page, in body text, not a footer or sidebar module.

6. Compress your hero images. On most sites the largest element on the page is an uncompressed image, and it is the main input to Largest Contentful Paint. Modern formats routinely cut file size by 70% with no visible difference.

7. Delete or consolidate one thin page. Every site has them. They dilute quality signals and consume crawl budget.

Wins inside an afternoon

8. Run a cannibalisation check. For each important query, search your own site and see how many pages target it. Where two or more compete, merge them and redirect. Consolidation regularly beats a quarter of new publishing, because it concentrates signals that were being split three ways.

9. Build an internal link map. List pages by inbound internal links. The pages you most want to rank are usually near the bottom, because links accumulate by accident. Fix the top ten mismatches.

10. Read the SERP before writing anything. Search your target query and look at what format dominates — comparison tables, tutorials, listicles, tools. The result set is the clearest available statement of what the query wants. Bringing the wrong format to it is the most common reason a well-written page fails. Our SERP analysis guide covers the read in detail.

11. Add a comparison table to any evaluative page. Tables are the most extractable structure on a page — for featured snippets, and disproportionately for AI assistants, which lift structured comparisons far more readily than prose.

12. Fix the pages one click from your homepage. Crawl depth correlates with crawl frequency. Anything important sitting four clicks deep is being visited rarely.

13. Write FAQ sections that answer real questions. Not invented ones. Pull them from your support inbox and sales calls. These match long-tail queries you would never have guessed, and they are the most cited part of most pages.

Wins that take a quarter

14. Build topical depth rather than breadth. Six pages that thoroughly cover one subject outperform twenty scattered across six subjects. Depth is how a site becomes the obvious answer for a category rather than an occasional result within it.

15. Refresh rather than republish. Take pages that ranked well and have decayed, and update them substantively — new data, removed dead sections, expanded coverage where the query has shifted. Changing the date without changing the content does nothing and is trivially detected.

16. Earn links by publishing something citable. Original data, a genuinely useful free tool, or a strong position on a live debate. Link building fails when there is nothing on the page worth linking to; the outreach is downstream of that.

17. Improve the pages that already convert. Ranking a page that converts at 4% is worth roughly ten times ranking one that converts at 0.4%. Most SEO plans are sorted by volume when they should be sorted by commercial value.

18. Fix Core Web Vitals properly. Not for a score, but because slow pages lose visitors before they read anything. Chase the field data in Search Console rather than the lab score in a testing tool.

19. Build a genuine author presence. Named authors with real credentials, consistent bylines, and pages that establish who they are. This matters more in health, finance and legal categories, and it is increasingly the tiebreak that decides which of two similar pages gets cited by an assistant.

Tips specific to AI answers

Assistants search the web before answering, then compose from what they retrieve. That changes what gets rewarded.

• Write self-contained passages. An assistant lifts a section, not a page. A paragraph that depends on the three above it for meaning cannot be quoted, so it will not be.
• Be specific and attributable. “Typically six to twelve weeks” gets quoted; “it varies” does not. Concrete claims with numbers are far more citable than hedged prose.
• Make sure the page works without JavaScript. AI crawlers largely do not execute it, so content that only appears after a script runs may as well not exist to them.
• Keep claims current and dated. Assistants weight freshness heavily on anything time-sensitive, and an undated page is hard to trust.
• Get named on the pages assistants already trust. Comparison roundups, industry directories and active community threads are retrieved far more often than most brands expect. Our guide to getting content cited by AI covers which sources carry weight.

Tricks that now cause damage

These are still repeated in tip lists, and each now carries a cost rather than merely being useless.

Keyword stuffing and density targets. Semantic matching made term frequency irrelevant years ago. Writing to a density figure produces awkward text that readers bounce from, and bounce is a signal.

Exact-match anchor text at scale. A natural link profile has varied, often untidy anchors. Fifty links all reading the same commercial phrase is a pattern that exists to be detected, and it is.

Private blog networks and paid link schemes. Beyond the policy violation, the practical problem is that the sites hosting them have no genuine audience, so the links carry little weight even before they are discovered.

Mass AI-generated pages with no editing. Publishing volume without oversight produces pages that are individually plausible and collectively worthless, and the aggregate drags the whole domain. The mechanism is not a penalty for using AI; it is that thin content has always been a quality problem and generation made it cheap to produce at scale. Our note on AI content that holds up covers where the line sits.

Things people worry about that do not matter

An incomplete list of anxieties you can safely drop:

If you only have one day

A single working day, in order, on a site that already has some impressions. This is the highest-return eight hours available to most sites.

• First 30 minutes. Search Console, Pages report. Indexed versus published. If there is a large gap, abandon the rest of this list and spend the day on that instead — nothing else matters until it closes.
• Next 90 minutes. Export by page, sort by impressions, rewrite titles and descriptions for the ten with the worst click-through relative to their position.
• Next 2 hours. Take the three pages sitting at positions 8–20 that matter most commercially. Open the results above each one, find what they cover that you do not, and close the gap.
• Next hour. Cannibalisation sweep on your ten most important queries. Note every case where two of your pages compete; plan the merges.
• Next hour. Internal links. From your strongest pages, add contextual links to the three you just improved.
• Last 90 minutes. Take your single most important page and move the answer to the top. Then fetch it with JavaScript disabled and confirm the content is still there.

None of that requires a paid tool, a developer, or permission from anyone. Most sites have never done any of it, which is why the day is worth so much.

If you did only the seven one-hour items on a site that has existing impressions, you would outperform most of the sites that read thirty-item lists and implemented none of them. The constraint in SEO is rarely knowledge.

Frequently asked questions