01 | Choose a niche: SaaS, local businesses or ecommerce.
02 | Build a website or find a business willing to let you gain practical experience.
03 | Set up Google Search Console and GA4.
04 | Understand how Google crawls, indexes and ranks pages.
05 | Learn how robots.txt, canonical tags and XML sitemaps work.
06 | Use Ahrefs or Semrush to analyse five competitors.
07 | Collect 100 keywords your target customers actually search for.
08 | Organise them by search intent and topic cluster.
09 | Search your target keywords on Google and study the top five pages.
10 | Select ten low-competition keywords with strong commercial intent.
11 | Create genuinely useful pages around those keywords.
12 | Learn on-page SEO: titles, H1 headings, URLs and meta descriptions.
13 | Build useful internal links between related pages.
14 | Use original screenshots, practical cases and real data.
15 | Learn AEO: make your content answer questions clearly and directly.
16 | Add Schema structured data where appropriate.
17 | Use Screaming Frog to conduct a technical SEO audit.
18 | Fix indexing issues, broken links and page-loading problems.
19 | Analyse competitors’ backlinks and earn relevant, high-quality links.
20 | Learn GEO: understand how ChatGPT, Gemini and Perplexity discover and cite information.
21 | Track which brands are frequently mentioned and cited in AI responses to your target questions.
22 | Identify information missing from AI answers and create content that fills those gaps.
23 | Build credible brand mentions on relevant industry websites and third-party platforms.
24 | Use AI to support research, not to mass-produce low-quality content.
25 | Keep monitoring keyword positions, clicks, conversions and AI-search visibility.
26 | Improve pages with high impressions but low click-through rates.
27 | Double down on content that genuinely brings customers and revenue.
28 | Document the entire process, including actions, results, data and screenshots.
P.S. It really comes down to this.
Learn first, then put it into practice.
Get results first, then document your case studies.
When you can demonstrate real traffic, customers and revenue for a website, you have a stronger basis for charging clients.
SEO builds expertise through practice. AEO answers questions through content. GEO uses credible information to pursue AI visibility.
Your real competitive advantage is not how many concepts you know, but the results you can deliver.
Frequently asked questions
Must I analyse five competitors and collect 100 keywords before starting?
No. These figures organise the exercise; they are not search-engine requirements or performance thresholds. Start with relevant competitors and a manageable set of real customer needs, validate the page direction, then expand. Quantity does not replace relevance or quality.
Do SEO, AEO and GEO require three separate content sets?
Usually not. They can share accessible pages, clear facts and credible sources. AEO emphasises answers; GEO also examines AI citations and brand understanding. Google says foundational SEO remains relevant to its AI features. This does not require duplicate pages for each label.
What do robots.txt, canonicals and sitemaps do?
Robots.txt manages crawling, canonicals express a preferred URL, and sitemaps help discovery. None guarantees indexing. Robots.txt does not replace noindex or access controls. Private drafts require appropriate access and public-route protection, not merely exclusion from a sitemap.
What should I check when a working page is not indexed?
Inspect the URL in Search Console and distinguish discovery, crawling, indexing and canonical selection. Check status codes, access, noindex, rendered content, internal links and distinctiveness. HTTP 200 is one technical condition, not an indexing guarantee. Non-indexing does not automatically mean a penalty.
Does every keyword need its own article?
No. Cluster by user task, intent and page type. One page can answer several expressions of the same need. Separate pages when the task or audience meaningfully differs. Improve an existing answer before adding near-duplicates.
Will adding FAQs or Schema secure AI citations or rich results?
No. FAQs should answer unresolved reader questions, and markup must match visible content and current supported features. Validation does not establish rankings, display or AI recommendations. Do not add unrelated questions merely to increase markup.
Should I immediately change titles when impressions are high but CTR is low?
First segment queries, positions, devices, markets and periods. Check audience relevance, page-type fit and whether the search result answers the question without a click. Then decide what to change. Preserve the baseline and change dates instead of changing everything at once.
How can I measure AI visibility without confusing crawls with citations?
Use a stable set of questions, platforms, modes, languages and sampling times. Record mentions, citations, accuracy, referrals and qualified enquiries separately. Retain answers and sources. Crawls prove access activity, not answer visibility, and self-tests should not count as customers.
How can I demonstrate customer and revenue outcomes?
Combine Search Console, appropriately configured GA4, customer-reported discovery and CRM records. Separate clicks, visits, enquiries, qualified opportunities and sales. Account for windows, deduplication, brand demand and other channels. Before-and-after change is not automatically incremental SEO impact. Document costs and scope, obtain permission for cases and protect personal data. The checklist does not promise results within a fixed period.
Source：Google: Search technical requirements
Source：Google: AI features and your website