SEO Strategy 2026: The Complete Guide for Growing Businesses

Google says it sees more than 5 trillion searches a year (Google, March 2025), and many of them never send anyone to a website. SparkToro's 2024 zero-click study found that 58.5% of US Google searches ended without a click. When the Pew Research Center analysed the March 2025 searches of 900 US adults, 18% produced an AI summary, and people clicked a traditional result on 8% of visits to a results page that had one, against 15% on pages without.

So an SEO strategy now has two jobs: earn the click where a click is still on offer, and be the business that gets named when the answer is given on the page. This guide is for founders and marketing leads in Australia and New Zealand deciding where to put the effort. It covers the foundations, how Google's AI features and assistants such as ChatGPT fit in, links, the overlap with paid search, and how to measure all of it.

Why most SEO strategies underperform

The underlying mistake is treating SEO as a campaign rather than an asset. A paid campaign stops when the budget stops. Pages that rank, and that other sites and AI answers cite, keep working after the work that built them is done. Programmes that don't deliver usually fall into one of three traps:

1. Treating technical SEO as a one-off audit. Pages slow down as features and plugins are added, template changes break structured data, and a CMS migration can quietly block or de-index whole sections. Without regular checks, the damage shows up months later as lost rankings.
2. Publishing without a coverage plan. A single well-optimised article rarely beats a competitor that answers every question a buyer asks, each on its own page. Google says its ranking systems work mainly at the page level, but that site-wide signals also contribute to how pages are understood.
3. Ignoring AI answers. Google's AI Overviews and AI Mode now answer many questions on the results page, and assistants such as ChatGPT and Perplexity answer research questions directly. If your pages aren't among the sources they draw on, buyers form a view of your category before they ever reach you.

What an SEO strategy has to cover now

An SEO strategy is a plan for being found and chosen in three places, each with its own rules:

• Google's classic results: crawlable, fast pages, content that matches what the searcher wants, and links from sites that vouch for you.
• Google's AI Overviews and AI Mode: the same ranking and quality systems, used to choose the pages an AI answer is built from. Google's documentation says there are no additional requirements to appear in them: no special markup, no special structured data and no AI text files.
• Other AI assistants such as ChatGPT, Perplexity, Microsoft Copilot, Claude and Gemini: these find pages through their own crawlers or a search engine's index, so they can only cite what they can reach, and their answers draw on what other sites say about you as well as on your own pages.

The disciplines overlap far more than the new acronyms suggest. Google's position, in its guide to optimising for its generative AI features (updated July 2026), is that from its side, optimising for generative AI search is still SEO. What has changed is measurement. Fewer searches end in a click, visibility in AI answers doesn't always move with rankings, and the reports that show it are new. That's covered in the measurement section below, and in detail in how to measure GEO success.

Three layers: infrastructure, authority, conversion

It helps to think of a search programme as three layers. Each depends on the one beneath it, and a gap in any of them caps the return on the others.

Infrastructure is whether search engines and AI crawlers can reach, render and understand your pages: technical health, a clear structure with a page for each service and each place you serve, internal links, and accurate structured data. It's the layer most teams know best. It's necessary but not sufficient: a technically perfect site with nothing distinctive on it has nothing worth citing.

Authority is whether your site is the most specific and trustworthy source on the questions your buyers ask. That means content only you could write, drawn from first-hand work; proof such as case studies with dated results and named people; business details that match everywhere they appear; and links and mentions from sites that chose to give them. Most of what gets called GEO happens in this layer.

Conversion is whether a buyer who arrives, from a search result, an AI answer or an ad, can act, and whether you can tell where they came from: service pages that define the offer precisely, contact options that are easy to find, calls and form submissions tracked as key events in Google Analytics, and a source recorded against every enquiry in your CRM. Visibility you can't convert or attribute is hard to defend when the budget is reviewed.

Technical SEO: the foundation

Technical SEO decides whether the rest of the work can be seen at all. The essentials are below; for a fuller checklist, see our guide to technical SEO foundations.

Core Web Vitals

Core Web Vitals are Google's measures of real-world loading, responsiveness and visual stability: Largest Contentful Paint (LCP), Interaction to Next Paint (INP, which replaced First Input Delay in 2024) and Cumulative Layout Shift (CLS). They're assessed at the 75th percentile of page loads, separately for mobile and desktop. The thresholds:

• LCP: 2.5 seconds or less is good; over 4 seconds is poor.
• INP: 200 milliseconds or less is good; over 500 milliseconds is poor.
• CLS: 0.1 or less is good; over 0.25 is poor.

Google says Core Web Vitals are used by its ranking systems, but also that good scores don't guarantee top rankings, and that chasing a perfect score purely for SEO may not be the best use of your time (Google's page experience guidance). Treat them as a threshold to clear, not a score to maximise.

The fixes are usually targeted rather than a rebuild: compress and properly size images, stop banners and cookie notices from pushing content around as the page loads (CLS), and defer JavaScript that isn't needed for the first view (INP). Search Console's Core Web Vitals report shows field data from real Chrome users, which is the version that counts. For the detail, see our technical guide to Core Web Vitals and page speed.

Crawlability and indexation

Google can't rank what it can't crawl and index. Problems here are more common than most businesses realise, especially after a CMS migration or years of changes by different developers. The checks that matter:

• XML sitemap: submitted in Search Console, and listing only pages you want indexed, with no redirects, 404s or noindexed URLs.
• Robots.txt: not blocking sections you want found. It controls crawling, not indexing: Google says a blocked URL can still be indexed, without a description, if other sites link to it. To keep a page out of Google, use noindex and leave the page crawlable.
• Canonical tags: every indexable page declares its canonical URL, so duplicates don't split signals.
• Internal links: pages with no internal links pointing to them are hard for Google to discover and get no internal link signals. A sitemap entry alone is a weak substitute.
• Page indexing report: Search Console's report shows which pages are indexed and why others aren't. Check it after every significant change.
• Log files: on larger sites, server logs show which pages Googlebot actually crawls, and how often.

Site structure: a page for each service and each place you serve

A common structural fault is one general page doing the work of many. If a buyer searches for a specific service in a specific place and your site has only a general services page, there's nothing precise for Google, or an AI assistant, to match. Three rebuilds from our own client work show the pattern:

• BCS Broking had fifteen pages in total, with one general insurance page and one general surety page. The rebuilt site gives each of 19 insurance verticals and 11 surety bond types its own page (case study). What happened next is covered in the AI section below.
• MLA Traffic's Squarespace template didn't break out its services or the locations it covers, so a search for “traffic management plans Melbourne” had nothing specific to match, and every Google ad pointed at the homepage. The rebuild gave every service and every area its own page, and paid traffic now lands on the page that matches the search. Since the site launched in June 2026 it has brought in 72 enquiries (case study).
• Mr Mow Garden Services in Gore, Southland, had a three-page self-built site that was invisible for searches like “lawn mowing Gore”. The rebuild has fourteen pages, with service and area pages of their own. It went live in late March 2026, and between April and August its monthly Google search impressions tripled (case study).

Pages for places only work when you genuinely serve those places and each page says something specific about the work there. Generating hundreds of near-identical “[service] in [city]” pages is a different thing, and Google treats it as spam: see the section on programmatic SEO below.

Structured data

Structured data (schema markup) describes a page to search engines in a standard format: who the business is, who wrote an article, when it was published. It's worth having, for narrower reasons than many guides suggest. Google's guide to its AI features says structured data isn't required for AI Overviews or AI Mode and there's no special markup to add, while recommending you keep using it for rich results.

The types worth having for most service businesses:

• Organization on the home page: name, logo, URL, contact details and sameAs links to your official profiles.
• LocalBusiness if you serve customers at an address or in a defined area.
• Article or BlogPosting on content pages, with the author, datePublished and dateModified.
• BreadcrumbList to show where a page sits in the site.
• Product if you sell products online.

Two types that many guides still recommend no longer earn rich results. In August 2023 Google limited FAQ rich results to well-known, authoritative government and health websites, and from that September it stopped showing HowTo rich results altogether (Google Search Central). In 2025 it also phased out several less-used types, including ClaimReview, and since 7 May 2026 FAQ rich results no longer appear in Google Search at all (Google Search Central documentation updates). Google says markup it no longer uses does no harm, but it won't change how your result looks.

Mark up only what's visible on the page, and validate it: Google's Rich Results Test checks the types Google supports, and the Schema Markup Validator checks everything else. A single syntax error invalidates the whole JSON-LD block.

Topical authority: cover what buyers actually ask

Topical authority is shorthand for how thoroughly and credibly a site covers a subject. Google doesn't publish a topical authority score, but a site that answers a subject's questions well gives both readers and search engines more reasons to trust each page on it.

The practical model is a set of connected pages:

• Hub pages that cover a broad topic or service and link to the detail.
• Supporting pages that each answer one specific question in depth: as long as the question needs, and no longer. Google says it has no preferred word count.
• Internal links that connect them: supporting pages link up to the hub, and across to related pages where a reader would want to go next.

An example: an accounting software business shouldn't stop at a page on “accounting software for small business”. It should answer the questions around the purchase, such as payroll compliance, BAS lodgement, cash-flow forecasting, invoice automation and end-of-year tax, each on its own page and linked from a central page on the product.

Two cautions. First, depth has to come from somewhere. Google's guidance favours content with a unique point of view, based on first-hand experience, over summaries of what's already online. Second, don't create separate pages for every variation of a query: Google says doing that primarily to manipulate rankings or AI answers breaches its scaled content abuse policy.

Start with a content gap analysis. List the questions buyers ask at each stage, from the first problem to the final comparison; your sales team's call notes are a good source. Mark which you answer well, which badly and which not at all. The gaps are your publishing plan.

AI Overviews, AI Mode and GEO

Generative engine optimisation (GEO) is the work of getting your business named, and your pages cited, in AI-generated answers: Google's AI Overviews and AI Mode, and assistants such as ChatGPT, Perplexity, Copilot, Claude and Gemini. The term was introduced in a 2023 research paper (published at KDD 2024) whose authors found that optimising a page's content could raise its visibility in generative engines' answers by up to 40% in their tests, with results varying by subject.

For Google, GEO is mostly SEO. Google's guide says its AI features are built on its core ranking and quality systems and that SEO best practices still apply. It lists things you can ignore for Google: llms.txt files and other “special” markup, breaking content into small chunks, rewriting content for AI, and chasing inauthentic mentions. To be eligible at all, a page has to be indexed and able to show a snippet, and your site must be included in Search Console's Search generative AI setting, which it is by default.

For other assistants, less is published about how sources are chosen. What's clear is that they can only cite what they can reach, so check that robots.txt doesn't block their search crawlers. OpenAI, for example, says sites that block its OAI-SearchBot won't be shown in ChatGPT search answers, and that blocking GPTBot, its training crawler, is a separate choice. And because their answers draw on what other sites say about you, your standing in industry publications, reviews and directories matters as well as your own pages.

If you'd rather have this run for you, it's what our GEO service does.

What AI answers do to clicks

Being cited in an AI Overview doesn't always produce a click. In Pew's March 2025 data, people clicked a link inside the AI summary itself on 1% of visits to a results page that had one. The value is often exposure at the moment a buyer is forming a view of the category, which is why branded searches and enquiries, not clicks alone, are the numbers to watch.

Search Console now reports on it. Clicks and impressions from AI Overviews and AI Mode are included in the Performance report's Web totals, and since June 2026 a separate Generative AI report shows how often your pages appeared in those features, by page, country, device and date. It has been available to all sites since 31 August 2026. It shows impressions only: no clicks and no queries.

What makes a page worth citing

Nobody outside these companies knows exactly how each system chooses its sources, and Google warns that no third-party tool has access to its internal ranking or AI systems. These are the things that hold up, both in Google's guidance and in what AI answers visibly do:

• Something only you can say. First-hand experience, your own data and worked examples. Google calls this non-commodity content. A page that restates what's already online gives an AI answer no reason to cite you rather than the original.
• Clear, specific answers. Say plainly what something is, what it costs or how it works, with any numbers dated and sourced. Google says you don't need to write in a special way for AI, but you do need to be clear, for readers first.
• A consistent identity. Your business name, address, phone number and description should read the same on your site, your Google Business Profile and every directory. Google's guide notes that Business Profile information can help local businesses appear in AI responses.
• Others vouching for you. Links and mentions from industry publications, associations and genuine reviews are what an AI answer finds when it looks beyond your own site.

A worked example: BCS Broking

BCS Broking places commercial insurance and surety bond facilities for Australian construction and mining companies. Its rebuilt site launched on 29 April 2026, with a page for each insurance vertical and surety bond type, and an insights programme answering the questions a CFO asks before buying: bank guarantee alternatives, government contract requirements, mining rehabilitation obligations. Checked on 27 July 2026, three months later (case study):

• For “surety bonds broker australia construction mining”, BCS held the first organic result, and Google's AI Overview named BCS Broking first among the brokers it cited, with two of BCS's own insights articles among its sources.
• For “mining and resources insurance broker australia”, BCS ranked fourth organically, and the AI Overview named it among “prominent firms including Aon, Coverforce, and BCS Broking”.

Two things are worth noticing. The AI Overview cited two of BCS's insights articles as sources: the content programme is what earned the citations, which is why it keeps running. And these are spot checks of two searches on one day, and AI answers can differ from one check to the next. The next step in that engagement is rank tracking across the full keyword set, so movement shows per query.

Link building: authority over volume

Links still matter, for Google and, indirectly, for AI answers that look beyond your own site. What has changed is that a few relevant, editorial links are worth far more than volume, and schemes to manufacture them carry real risk. Google's spam policies treat links created mainly to manipulate rankings as link spam, including buying or selling links, excessive link exchanges and automated link building, and sites that break the policies can rank lower or drop out of results.

What makes a link valuable:

• Relevance: a link from a publication your buyers read is worth more than one from a big site in an unrelated field.
• Editorial placement: a link in the body of an article, given because the content deserved it, is worth more than a footer, sidebar or author-bio link.
• Diversity: links from many different sites count for more than many links from one.
• Real readers: a link on a page people actually visit is a sign of genuine editorial value.

The approaches that hold up:

1. Original research and data. Benchmarks, surveys or analysis nobody else has give journalists and other writers something to cite, and they feed the authority layer directly.
2. Expert commentary and digital PR. Regular, useful comment for journalists and editors covering your industry, tied to stories they're already writing.
3. Partnerships and associations. Joint content with complementary businesses, industry bodies and suppliers produces relevant links that are hard to fake.

What to avoid: link exchanges, paid placements on “news” sites that exist to sell links, guest posts on sites with no real audience, and any service promising a fixed number of links a month. And don't reach for the disavow tool by default: Google says most sites will not need it, and suggests it only when you have a considerable number of spammy or artificial links that have caused, or are likely to cause, a manual action.

Programmatic SEO and the city-swap trap

Programmatic SEO means generating many pages from structured data. It works when every page carries information that's genuinely different: Wise, for example, publishes a page for each currency pair, each with its own rates. It fails when the only thing that changes between pages is a word in the template.

The version that tempts service businesses is the city swap: a “[service] in [suburb]” page for every suburb or city, with the same copy and the place name changed. Google's spam policies describe doorway abuse as pages created to rank for specific, similar searches, including pages targeted at specific regions or cities that funnel users to one page, and scaled content abuse as many pages generated mainly to manipulate rankings rather than help users. A city-swap grid fits both descriptions.

Location pages hold up when they reflect reality: you genuinely serve the place, and the page says something specific about the work there, such as the jobs you've done, local requirements, who does the work and how quickly you can get there. If you can't write that for a place, it doesn't need a page.

If you do build pages programmatically, the technical basics still apply: a clean URL structure, correct canonicals, pages linked from somewhere a visitor can reach, and a sitemap that updates as pages are added or removed.

SEO and paid search: share the data

SEO and paid search work best as one programme with shared data, not two budgets reporting separately.

The most useful thing paid search gives SEO is conversion data by query. Search Console tells you which searches bring clicks, but not which of them turn into enquiries; Google Ads' search terms report shows exactly that for the searches you bid on. A few months of well-structured campaigns will show which terms produce enquiries and which only produce traffic, and that should decide which pages the SEO programme builds or strengthens first.

It runs the other way too. Paid and organic search need the same thing from the site: a specific page for each service and each place. MLA Traffic's old ad account sent every ad to the homepage because there were no service pages to send them to. The page-per-service rebuild addressed both at once: Google got specific pages to rank, and paid traffic now lands on the page that matches the search.

In practice, share search terms reports, conversion data and landing-page results between whoever runs SEO and whoever runs paid search at least monthly, and set page priorities together. If your paid search needs attention first, start with why broad match is the most expensive cheap traffic you can buy.

Measuring SEO across search and AI answers

Measuring SEO now takes more than rankings and organic sessions, because some of the value never produces a click. Measure by intent, and look at AI visibility and attribution alongside traffic:

• Organic traffic by intent: separate informational, commercial and branded searches. Informational growth that never leads to commercial traffic is a content-strategy problem, not a win.
• Click-through rate on queries with AI Overviews: Search Console doesn't flag which queries showed an AI Overview, but many rank trackers now record it. Tag those queries and compare their CTR with similar queries that don't show one. Lower CTR on informational queries with an Overview is expected; falling CTR on commercial queries is a problem to fix.
• Branded search: some people who meet your name in an AI answer, or anywhere else, will search for it later, so branded searches are one of the few visible signs of awareness built off your site. Search Console's branded queries filter splits branded from non-branded queries automatically for eligible sites. Growth in non-branded clicks shows new buyers finding you.
• Visibility in AI features: Search Console's Generative AI report for Google's AI Overviews and AI Mode; Bing Webmaster Tools' AI Performance report for citations in Microsoft Copilot and Bing's AI summaries; and a monthly check of a fixed set of buyer questions in ChatGPT and Perplexity, which neither report covers.
• Attribution: in GA4, the Attribution paths report (Advertising → Attribution) shows how often organic search appears in journeys that end in a key event, not just at the last click. The Attribution models report compares data-driven attribution with paid-and-organic last click; GA4 removed the first click, linear, time decay and position-based models in November 2023.

What not to use as a headline KPI: average position. It blends thousands of queries of different intent into one number that's easy to move without changing anything that matters, for example by losing low-ranking, high-volume queries. Track positions per query, grouped into bands, for the searches you care about.

For the AI side in detail, including a GA4 channel for AI assistants and a monthly citation audit, see how to measure GEO success.

Your SEO strategy checklist

Not an exhaustive technical audit: these are the decisions that determine whether the investment compounds. Work through them roughly in this order.

Technical foundation

• Core Web Vitals in the good range on key templates (LCP 2.5 s or less, INP 200 ms or less, CLS 0.1 or less)
• XML sitemap submitted and clean in Search Console
• No accidental noindex or robots.txt blocks on pages you want found
• Canonical tags correct site-wide
• Organization, Article and BreadcrumbList structured data (plus LocalBusiness if relevant) deployed and validated
• No orphan pages: every page you care about is linked from somewhere

Structure and coverage

• A page for each service you sell and each place you genuinely serve
• The questions buyers ask mapped to the stages of the purchase, each with a page that answers it
• Content gap analysis done against your top three organic competitors
• Key pages reviewed at least once a year, with dates changed only when the content changes

AI visibility

• Key pages indexed and able to show a snippet, and the Search generative AI setting left on in Search Console
• robots.txt allows the search crawlers of the assistants you want to be cited by
• Business details consistent across your site, Google Business Profile, LinkedIn and directories
• Named authors with real bios on content pages
• A fixed set of buyer questions checked monthly in Google and the main AI assistants

Authority

• At least one piece of original research or data in production or published
• An expert commentary or digital PR programme running
• Link profile reviewed, with the disavow tool kept for spammy links that have caused, or are likely to cause, a manual action

Conversion and measurement

• Calls, forms and bookings tracked as key events in GA4
• A source recorded against every enquiry in your CRM, including a “how did you hear about us?” question
• Search Console checked at least monthly, including the Generative AI report and branded versus non-branded clicks
• Organic traffic and key events reported by intent
• Search terms and conversion data shared between SEO and paid search

To see where your own site stands, start with a free website audit, or talk to us about our SEO service.