How to Rank on Google in 2026

What changed in Google search this year?

AI Overviews changed the economics of ranking. Google now answers a large share of questions at the top of the results page, and the click that used to follow simply does not happen.

The practical consequence is uncomfortable but simple: you can hold position one and still lose traffic. If your Search Console shows steady rankings and falling clicks, you are not being penalised — you are being summarised.

Two things follow from that:

• Informational pages lose the most. "What is X" and "how does Y work" are exactly what an AI answer resolves without a click.
• Commercial pages hold up far better. People comparing providers, checking prices, or ready to buy still click through, because they need to see who they are dealing with.

That single split should reshape where you spend effort. Guides earn citations and brand trust. Service and product pages earn revenue. Stop treating them as the same job.

Does ranking still matter, or only AI citations?

Both, and they feed each other. Pages that already rank well are the pool AI systems pull from, so classic SEO remains the entry ticket rather than the finish line.

The mistake is treating "AI optimisation" as a separate product. In practice the work overlaps almost completely with good SEO: clear structure, genuine expertise, valid markup, trustworthy sourcing. What changes is emphasis — machines reward clarity that human rankings merely tolerate.

A useful mental model: Google decides whether you are eligible. The answer engine decides whether you are quotable.

How do I make a page quotable?

Answer engines lift short, self-contained passages. If your answer only makes sense after three paragraphs of context, it will not be selected.

Four changes do most of the work:

• Lead with the answer. Resolve the question in the first 40–60 words, then expand underneath for readers who want depth.
• Phrase headings as real questions. "How much does a website cost?" beats "Pricing". The heading and the paragraph below it become a clean question-and-answer pair.
• Write declaratively. "A landing page takes one to two weeks" is extractable. "Timelines can vary depending on many factors" is not.
• Keep paragraphs short. Two or three sentences. This helps mobile readers and machine extraction at the same time.

None of this is a trick. It is the same discipline that makes writing good — it just now has a second beneficiary.

Are backlinks dead in 2026?

No. Links remain one of Google's core ranking signals, and pages with strong link profiles tend to get cited in AI answers more often, not less. Be sceptical of anyone who tells you otherwise — it is usually a prelude to selling you something else.

What genuinely died is volume-based link building. Buying hundreds of links now risks the domain rather than helping it. A handful of relevant, editorially earned links from sites in your industry outperforms a warehouse of purchased ones.

The safest sources are also the least glamorous: genuine partnerships, industry directories that predate SEO, being quoted by a trade publication, and publishing something worth referencing.

What technical work still matters?

The unglamorous foundation carries more weight than ever, because AI crawlers are less forgiving than Googlebot.

• Ship your text in the HTML. Many AI crawlers do not execute JavaScript. If your service descriptions, prices or FAQs only appear after scripts run, those crawlers can see an empty page.
• Fix Core Web Vitals properly. Measure on real mobile hardware, not a desktop test that flatters everyone. See Google's Core Web Vitals guidance.
• Use valid structured data. Organisation, Article, FAQ and Breadcrumb markup is the machine-readable description of what your page is. Validate it with the Rich Results Test rather than assuming.
• Make authorship visible. A named author, a publish date and an update date are cheap to add and are exactly the trust signals both systems look for.

What should I stop paying for?

A cottage industry has grown up around AI search, and a lot of it sells files rather than outcomes.

Treat these with caution:

• "AI schema" packages. There is no special schema type that makes an AI quote you. Standard structured data, correctly implemented, is the whole story.
• Guaranteed AI Overview placement. Answer engines select a small handful of sources per response and change constantly. Nobody can guarantee inclusion.
• Bulk content packages. Publishing forty thin articles a month was a weak strategy in 2020. Now it actively signals low quality to systems trained to spot exactly that.

If a proposal cannot explain why a tactic works, assume it does not.

What does a realistic timeline look like?

Technical fixes show movement in about four to six weeks. They lift every page at once, so they compound faster than content.

Restructured pages can earn AI citations surprisingly quickly — sometimes within a month, because answer engines re-crawl aggressively. Competitive commercial keywords still take three to six months of consistent work, and longer in crowded markets.

Anyone promising page one in thirty days is either targeting searches nobody makes, or using tactics that will eventually cost you the domain.

What to do this week

Five actions, in order of return:

• Run your ten most important questions through Google, ChatGPT and Perplexity. Write down who gets cited. That is your real baseline — rankings alone will not show it.
• Add a direct answer to the first paragraph of your five most valuable pages.
• Rewrite your headings as questions people actually type.
• Check your text exists without JavaScript. View the page source and search for your own service description. If it is not there, fix that before anything else.
• Validate your structured data and add a visible author and update date.

None of it requires a bigger budget. It requires deciding that clarity is the strategy.

Frequently asked questions