Website with AI Content Engine: How new posts are continuously created from search data

An AI content engine is a cycle in which the website, search data, and AI work together: the AI reads from Google Search Console, what your customers are searching for, suggests suitable topics, writes drafts, and improves pages that are getting too few clicks. A person reviews and approves every post. After that, results are measured, and the cycle starts again.

Most SME websites are built once and then hardly changed for years. New content is created when someone has time, so rarely. A content engine reverses that: the website keeps getting better because data and AI prepare the work, and the human only decides and adds to it.

The cycle in six steps

1. Collect data: Search Console (search queries, positions, clicks), Google Analytics (what visitors do) and Google Business Profile (calls, directions).
2. Find opportunities: The AI looks for search queries for which the website already appears but ranks weakly, and pages with many impressions and few clicks.
3. Check topic:Is there already a page for this keyword? Two pages on the same topic are detrimental to both, which is why every idea is checked against all existing pages.
4. Write draft: Research with sources, text according to the rules for AI-citable texts, title image, internal links, on request translation into additional languages.
5. Human approval:Technical review, add your own experience, approval. Nothing is published without approval.
6. Measure and improve: After a few weeks, Search Console shows whether the post is being found. Weak titles are revised, and new questions are incorporated into the next round.

What AI handles and what the human

Human approval is not a side step. Google recommends checking AI content for accuracy before publication, and the EU AI Act ties transparency obligations to editorial responsibility. More on this in Are AI texts bad for SEO?.

Why this works for SMEs

• Topics from real data instead of gut feeling: We write what customers are actually searching for.
• Low time commitment: Instead of writing texts, you review drafts and add your expertise.
• No duplicates: Checking against existing pages prevents your pages from competing with each other.
• Measurable: Every week, the data shows what works.

Example: bettersorted.de

This exact cycle is running on bettersorted.de. Within ten weeks, Google clicks increased from 6 to 122 per week. A practical example: the analysis showed that the funding page for BAFA consulting was competing with posts around the same terms. The pages were then restructured, around 230 internal links were added, and titles were revised. The figures are shown in the SEO case study.

Requirements

• A website with a system in which posts can be maintained in a structured way.
• Access to Search Console and Analytics.
• A person in the company who reviews drafts for accuracy, usually one hour per post.
• Clear rules: which topics, which tone, which statements never.

We offer the AI content engine as part of AI visibility from bettersorted: from connecting data sources to ongoing content creation and monthly evaluation. This turns your website step by step into your most important sales channel. For funding: a consultation on online and visibility strategy can be funded as business consulting through BAFA consulting grant can be subsidized, in the new federal states at 80%, in the old federal states at 50% (applications until 31.12.2026). The implementation itself, i.e. website development or copy, is not subsidized. What is eligible for your business is shown by the funding check.

Frequently asked questions

Does the AI write the posts entirely on its own?

No. The AI analyzes data, suggests topics, and writes drafts. A person reviews each post for accuracy, adds their own experience, and approves it. Nothing is published without approval.

How many posts are created per month?

That depends on the topic and the time needed for review. For many SMEs, four to eight good posts per month are realistic. More important than the quantity is that each post addresses a real search query.

What data does the content engine need?

At minimum, Google Search Console. Ideally also Google Analytics for behavior on the website and the Google Business Profile for local queries. Personal customer data is not required for this.

Does it work with every website?

In principle, yes, if posts can be structured and access to the Search Console is available. For very old systems, a relaunch is often the better starting point.

As of October 2026.