SEO stands for Search Engine Optimization. In practice, it means: make your site easy for Google (and Bing and DuckDuckGo) to understand, so the right people land on the right page.
The librarian analogy
Think of Google like a librarian for the entire internet. Every day it sends out little robots (called crawlers) to read websites. The crawlers report back: «this page exists, it's about prompts, and here are the links it contains». The librarian writes it all into a giant catalog (called an index). When someone searches, the librarian picks the books that match best and sorts them into a specific order. That order is the ranking.
Your job is to make sure the librarian can find your page (crawl), understand it (index), and decide it's worth showing first (rank). That's what it all comes down to.
How it actually works under the hood
Crawl. A crawler visits your URL, downloads the HTML, follows the links it finds, and queues up new URLs. Two files tell crawlers how to behave: robots.txt, a short text file at yoursite.com/robots.txt that says «you can read this, skip that». The other is sitemap.xml, a list of every page you want indexed.
Try it yourself: open https://kenove.no/robots.txt in your browser. The rules are right there. Then try https://kenove.no/sitemap.xml. No magic, just a text file.
Index. Once a crawler has your HTML, it reads all the visible text, the headings, the link text, and a chunk of invisible-but-readable information in the page's <head>. The most important elements:
<title> is the blue link in Google. <meta name="description"> is the gray summary underneath. <link rel="canonical"> means «if this page has duplicates, this is the original». Open Graph tags like og:title and og:image decide what shows up when someone pastes your link in Slack. JSON-LD is structured data that says «this is an Article, by this author, published on this date», in a format machines prefer.
Rank. Google then asks: is the page about what the searcher typed? Is the site trustworthy? Do other sites link to it? Is it fast on a phone? You can't game this, but you can avoid the mistakes that trip you up.
How Lovable handles this for you
Most of the technical setup is handled by Lovable's built-in SEO & AI search review. It checks the elements that matter most and fixes most of them in one click:
Per-page titles and descriptions, so no two pages share the same text. Open Graph and Twitter cards, so a shared link shows a real preview instead of Lovable's own branding. Canonical links per route so duplicate URLs don't compete. Hreflang when you have multiple languages (this site uses it for EN/NO). A sitemap.xml and a sane robots.txt that lets crawlers in. Structured JSON-LD data where it makes sense.
On top of that, the SEO & AI search tab in the Lovable sidebar runs a full analysis on demand: titles, descriptions, semantic HTML, alt text, mobile, and performance. Most red findings come with a one-click fix. Open it, click Scan, wait about a minute. You'll see green, yellow, and red rows. Click a red row to read what's wrong, then either fix it yourself or let Lovable run the fix.
What you still need to do yourself
Lovable can't decide what your site is about. That's still on you:
Positioning: «Practical AI in Norwegian» is a sentence Lovable can't write for you.
Target keywords: Pick 3–5 phrases people actually search for. Lovable's Semrush integration (set up once under Connectors) gives you volume and difficulty.
Real content: A page with 100 words of filler ranks below a page with 800 words of substance. No tool fixes this.
Internal linking: When you write a new note, link it to related pages. That's how the librarian learns your site's structure.
One canonical URL per topic: Don't write three thin posts about «what is a prompt». Write one good post and keep it updated.
What to remember
SEO is three steps: crawl, index, rank.
Lovable's SEO & AI search review checks the technical framework (titles, sitemap, robots, JSON-LD, hreflang) and fixes most of it in one click.
Your job is positioning, real content, and linking your own pages together. The tool can't do that for you, and that's the part that actually makes the difference.