How do I rank higher in Google search results?

Ranking advice is usually delivered as if all positions were the same problem. They are not. The work that moves a page from 18 to 9 has almost nothing in common with the work that moves it from 4 to 1, and applying the wrong one is how teams spend a quarter and move nothing.

Find your band first

Open Search Console, filter to the query you care about, and read the average position. That number determines everything that follows.

Averages hide things, so check the trend as well as the level. A page averaging 12 that was 25 three months ago is working and needs patience. A page averaging 12 that was 6 is a different situation with a different cause.

Unranked to page 2: a relevance problem

If you are not ranking at all, or you are sitting past position 20, the page is not yet recognised as a credible answer to the query. Almost always this is intent or coverage rather than quality.

Check the format first. Search the term and look at what ranks. If the first page is entirely comparison articles and you have a product page, format is the whole problem. Google has published its opinion about what the query means; matching it is not capitulation, it is reading the brief.

Then check coverage. List the H2s of the top five results. The subtopics that appear in most of them are what searchers expect this page to contain. Missing three of five is usually enough to keep you off page one regardless of how well the parts you did write are written.

Then check specificity. If the term is dominated by sites far stronger than yours, the winnable version is more specific. “Project management software” may be unreachable while “project management software for architecture firms” is wide open, converts better, and builds the topical authority that eventually makes the broader term reachable.

Page 2 to top 5: a depth problem

This is the highest-return band in SEO, and the most neglected. A page ranking 8 to 20 has already cleared relevance — Google agrees it is about the right thing. What it lacks is enough substance to be preferred.

Four moves work here, in rough order of return.

Close the subtopic gaps. Take the top three results, list every section they cover, and add the ones you are missing that genuinely belong. Not padding — the specific questions a reader would have next.

Answer earlier. Move the direct answer to the first two sentences of each section. This helps human skimmers, improves your chances at featured snippets, and makes the page far easier for an AI engine to extract and cite.

Refresh the substance. Update figures, remove advice that has aged badly, add what has changed since publication. Substantive updates move pages; date changes alone do not.

Strengthen internal links. Link to the page from your strongest related pages using descriptive anchor text. This is free, entirely within your control, and consistently underused.

Top 5 to number 1: an authority problem

Once you are in the top five, further on-page editing produces diminishing returns quickly. Everyone at this level has a competent page. The remaining differences are authority and preference.

Links from relevant sources. Not volume — relevance and credibility. A single link from the definitive site in your niche outweighs fifty from directories.

Brand strength. Searches for your brand name, mentions across the web, and recognition all feed the sense that you are a real entity worth ranking. This is slow and compounds.

Genuine differentiation. Original data, a tool, a dataset, primary research — something a competitor cannot produce by writing harder. This is the only durable advantage at the top of a competitive term, and it doubles as the strongest asset for being cited by AI engines.

Satisfaction. At this level, whether people who arrive stay and stop searching matters. If they return to Google and click the next result, that is visible.

What works in every band

• Match the query’s phrasing in the title, in the words a searcher would use rather than your internal vocabulary.
• One topic per page, covered completely. Not one keyword — one question a reader is actually asking.
• Clean, descriptive headings that map to real sub-questions.
• Specific, verifiable claims. Numbers, dates, named sources. These help human trust, featured snippets and AI citation simultaneously.
• Fast, stable, accessible pages. Not a ranking lever on its own, but a reliable tiebreaker and a conversion one.

What wastes time in every band

• Keyword density targets. Not a thing. Write naturally and cover the topic.
• Publishing volume for its own sake. Thin pages compete with your good ones and dilute site-level signals.
• Chasing every technical score to green. Fix what blocks crawling and rendering; ignore the rest until it demonstrably matters.
• Changing dates without changing content. Transparent and ineffective.
• Bought links. The risk is asymmetric and the effect is temporary.

If you are not sure which band you are in, that is the first thing to establish — everything else follows from it. Our guide to which pages to improve covers how to triage a whole site this way rather than one page at a time.

How long each band takes

Expectations cause more abandoned SEO programmes than results do. Rough timelines, assuming the work is actually done rather than merely planned:

The last row is why title work is the best first move in almost every situation: it is the only lever with a payback measured in days rather than months, and it operates on traffic you have already earned.

Prioritising across a whole site

One page at a time does not scale past about twenty pages. For a whole site, sort the work by expected return rather than by which page someone happens to be annoyed about.

1. Export every query with more than 100 monthly impressions. Below that threshold, even a large percentage improvement produces nothing worth the effort.
2. Bucket by position band — 1–5, 6–20, 21+ — and by CTR against the typical rate for that position.
3. Work the 6–20 bucket first, highest impressions down. This is where a small ranking improvement produces the largest absolute traffic change, because the click-through curve is steepest between positions ten and four.
4. Then the 1–5 bucket with below-par CTR. High visibility, poor conversion, fixed by writing rather than by earning.
5. Then the 21+ bucket, but only where the query is commercially valuable. Most of what sits below 20 belongs there and moving it is not worth a quarter.

That ordering routinely produces more traffic in a quarter than a content calendar of new pages, at a fraction of the cost, because it operates entirely on demand that has already been proven to exist.

The one report worth building

Build a single sheet with one row per query and five columns: query, current position, position ninety days ago, impressions, and click-through rate against the typical rate for that position. Sort by impressions. Everything you should do next is visible in the first thirty rows, and it takes about an hour to assemble from a Search Console export. Most teams have never built it, which is why most teams work on whichever page a stakeholder mentioned most recently.

One caveat worth stating: position is now only part of the outcome. Even at number one, an AI Overview can sit above you and absorb the click, and an assistant can name three competitors without your page entering the conversation. Ranking work remains necessary — retrieval depends on it — but it has stopped being sufficient on its own. Our AI Search 2026 report covers what changed and by how much.

Sources

• Ahrefs — click-through rate impact of AI Overviews; 34.5% decline for position one (April 2025), approximately 58% (February 2026).
• Pew Research Center — approximately 68,000 real search queries; clicks fell from 15% to 8% of visits when an AI summary was present.
• Google Search Central — documentation on indexing, canonicalisation and Search Console position reporting.

Frequently asked questions