12 SEO tips you can apply today

Most SEO advice starts with a tool costing a hundred euros a month. This article doesn't. You can do everything below this afternoon with free resources, and for each tip I tell you how to measure whether it worked. The tips come straight from the scans we run on the sites of business owners who come to us: it is always the same twelve things that aren't in order.

Measure first: know where you stand (tips 1–2)

Tip 1: run a free scan. Before you change a single title, you want to know where the gaps are. Our free site scan checks your titles, metas, structure, speed and AI readability and ranks the findings by impact. Five minutes of work, and you have a list instead of a hunch.

Tip 2: install an on-page checker in your browser. We built our own Chrome extension for this, Neuramerce SEO: it analyses every page you are on (title, meta description, heading structure, alt texts, structured data, hreflang and internal links) right in your browser, without any of your data leaving it. That way you don't just check your own pages, you also see straight away how a competitor does it. The extension is in its final testing phase; until it is in the Chrome Web Store, the site scan does the same job from our side.

On-page: the five classics (tips 3–7)

Tip 3: give every page a unique, descriptive title. The page title is still the heaviest on-page signal. Put the term the page should be found for at the front, keep it under sixty characters, and check that no two pages share the same title. Duplicate titles make pages compete with each other.

Tip 4: write meta descriptions that earn the click. They aren't a ranking factor, but they are your ad copy in the search results. Put the core message in the first 120 characters and end with a reason to click.

Tip 5: one H1 per page, headings as a table of contents. If your page has a logical heading structure (H1 → H2 → H3), search engines and language models both understand how it is built. Rule of thumb: anyone who reads only the headings understands the article.

Tip 6: fill in alt texts for images. Describe what can be seen, in plain language. Good for accessibility, Google Images and AI systems that parse your page.

Tip 7: answer the question at the top. Open every article with the direct answer and only then go into detail. Visitors drop off less, and AI assistants prefer to cite pages that answer the question straight away.

Structure and links (tips 8–10)

Tip 8: add internal links between related pages. Internal links are the cheapest ranking gain there is: they tell search engines which pages belong together and which matter most. Rule of thumb: every article links to at least two related pages, with anchor text that says where the link goes ("our comparison of SEO software", not "click here").

Tip 9: fix broken links and old URLs. Every 404 is a dead end for visitors and crawlers alike. When a page moves, set up a 301 redirect from the old address to the new one.

Tip 10: check your speed on mobile. Google measures mobile. Run your site through PageSpeed Insights and tackle the two biggest culprits, which are usually images that are too large and scripts that do nothing. Compressing images alone often saves seconds.

Don't forget AI (tips 11–12)

Tip 11: write answers, not keywords. A growing share of your visitors ask their question in ChatGPT or get an AI Overview above the search results. Those systems cite pages that answer a question directly and factually. Rephrase your most important headings as questions and put the answer right below them. For Google this is not a separate trick: according to Google itself, AI Overviews and AI Mode run on the same ranking and quality systems as the regular search results. Tips 1 to 10 therefore count there in full.

Tip 12: add an llms.txt to your site, for the assistants outside Google. One text file in the root that tells AI assistants what your site is and which pages matter. Google doesn't use it: its guide to generative AI features (May 2026) says llms.txt is treated there as an ordinary text file. Crawlers from ChatGPT, Claude and Perplexity do demonstrably visit your site. It takes half an hour and can't do any harm, but don't expect anything from it in Google.

Want to go deeper after this? We will cover findability in AI answers and llms.txt in more detail on this blog over the coming weeks, and on a Neuramerce site this whole list (scores, metas, internal links, llms.txt) is simply built into the platform.