Earned on the Side: October update, early rankings and a change of tools

The most useful change is average position. In the first 19 days it was 46, meaning most of the site's visibility sat several pages away from the leading results. Across the full period since launch it is now 22, and the trend through September and October was one of gradual improvement rather than a single jump.

Several referral-code pages have started to rank for the searches they were written for. Most are not yet high enough to attract regular clicks, and many still sit below the first page of results. That is frustrating, but it is a far better starting point than position 46.

There was also a short spike in impressions in early October, which fell back within a few days. One-off movements like that are worth noting but not worth building plans around. The steadier improvement in position is the more meaningful signal.

I am still not publishing individual search terms or page-level performance. The figures above are aggregate and directional.

The site has received 21 clicks from Google since launch. That is roughly two a week rather than a few a month. It is progress, but it is still far too little traffic to judge how readers behave once they arrive.

The click-through rate has fallen from 1.29% to 0.6%. That looks like a step backwards, but it mainly reflects impressions growing faster than clicks. The site is now appearing for many more searches, often in positions where few people click. If pages continue to move up, the click-through rate should follow. If they do not, that will tell me something useful about the titles, descriptions or relevance of the pages concerned.

Earned on the Side was launched using Hostinger's AI website builder. It was a quick way to get a working site live, and it allowed me to focus on content rather than code in the first few weeks.

The problem is cost efficiency. A site like this needs a steady stream of small changes: updated dates, corrected offer terms, refined headings and new pages. Each change uses AI credits, and because the credits expire, any unused allowance is lost rather than carried forward. For a self-funded project that makes frequent small edits, that model does not make sense over the longer term.

I have therefore decided to move the site away from the builder and manage it directly by connecting it to Claude. This gives me more control over the content and avoids paying for credits I may not use.

Alongside the move, I have started setting up rule-based automations. The aim is not to hand the website over to AI. It is to make the repetitive parts of maintenance faster and more consistent, so that limited time goes into judgement rather than routine checking.

The rules are designed to:

Changes are still reviewed by me before they go live. That matters most on pages that include personal experience, financial conditions or information that can change. More detail on how AI is used is set out in the AI and editorial policy.

The first real test of the new workflow came at the start of November. Every offer and review page was rechecked, review dates were updated and headlines referring to the month were refreshed.

The Tide pages were also verified against Tide's live pricing and the public review platforms. Two details had changed since the last check and were corrected: the Google Play rating had moved, and Tide had renamed its free plan. Neither change was dramatic, but both are the kind of small inaccuracies that build up on referral pages when nobody is checking.

All 16 offer and review pages were checked and re-dated. No offers were removed.

Approximately four hours were spent on the project in October. That is well below the intended 10-15 hours per week, and I do not want to disguise it.

Part of the reason for moving to automated checks is to make limited time go further. Even in a month with only a few hours to spare, the site should stay accurate, although little new work may be produced.

Search visibility is developing in the right direction. Average position has improved from 46 to 22 and more pages are appearing for relevant searches.

The maintenance process is also stronger than it was at launch. The November refresh showed that offer pages can be checked and updated consistently, and that small changes made by providers can be caught and corrected.

Traffic is still very low. Twenty-one clicks in just over two months are not enough to understand reader behaviour or commercial performance.

Measurement is also still incomplete. Website users and sessions have not yet been confirmed, and referral-link clicks are not tracked reliably enough to report. That was a priority after the first update and it remains one now. Without it, the middle of the journey between a search result and a partner application is still largely invisible.

Time is the other weakness. Four hours in a month is enough to maintain the site, but not enough to grow it.

The immediate objective remains to strengthen the existing foundation and collect better evidence.

Priorities for the next reporting period are:

After just over two months, Earned on the Side has 31 published pages, 3,750 search impressions, 21 clicks and an average position of 22. The site is still small and the traffic is still modest.

The difference from September is direction. Pages are moving up, more of them are being seen and the process for keeping them accurate is becoming more efficient. The next update should show whether that direction continues and whether better positions begin to turn into consistent visits.