LigoSocial went from 997 clicks to 1,869. Here are the loops behind it.

The first version of this page was written in summer 2025. Its headline numbers came from two Search Console screenshots, and we have kept both. They are the “then”. Nothing about them has been redrawn.

Now the same three measures, pulled on 9 October 2026 for the latest complete 28 days, 9 Sept to 6 Oct 2026.

Two snapshots can hide a lot, so here is everything in between that we can still pull. From 10 December 2025 to 6 October 2026, 301 days, ligosocial.com earned 21,571 clicks from 4,940,489 impressions in Google Search.

It is not a straight line, and we would not trust it if it were. Clicks climbed through the spring, and the best 28 days were 23 Apr to 20 May 2026 with 3,179 clicks. The best single day was 19 May 2026, with 171. Then the summer was soft: July brought 1,544 clicks at an average position of 10.8.

Switch the chart to average position to see what the latest passes of the loops line up with. Position went from 11.3 in August to 7 in September, and the latest 28 days are the best the site has recorded. Clicks are following: 1,544 in July, 1,802 in August, 1,960 in September.

LigoSocial's Search Console is connected to Murkuz once. After that the work happens inside Claude. Murkuz hands Claude the real rows for the site, with the window and the source attached, and Claude does the reading: which queries, which pages, what changed, what to do next. We draft the articles and the edits in the same conversation.

The drafts then go into the Murkuz CMS, which renders LigoSocial's blog. On 9 October 2026 it held 280 published articles for the site. That part of Murkuz is an add-on we enable on request, and it is covered further down. The loops themselves need only the connector.

There is no single trick here. There are three questions, asked again and again of fresh data. Each one is a loop because the answer changes after you act on it: publish the article, merge the pages or tighten the paragraph, wait for Search Console to catch up, then ask again.

The first loop looks for searches we have no proper answer to. Claude reads the queries LigoSocial already gets impressions for, finds the ones where no page addresses the search directly, and widens each into its neighbouring terms. The output is a short list of articles worth writing, each with a target query and the questions it has to answer.

We then draft those articles in the same conversation and publish them through the Murkuz CMS. The three most recent went live on 8 October 2026, the day before this page was rewritten.

Publish steadily for long enough and your own pages start competing with each other. Google has to choose between them for one search, and often ranks none of them as well as it would rank one clear page. The second loop finds those collisions.

On 9 October 2026 the connector found 105 queries where two or more LigoSocial pages appeared for the same search in the last 28 days (queries with at least 50 combined impressions). Here is one of them, exactly as returned.

Five pages, 1,149 impressions, no clicks. Two of the pages carry almost all of the impressions and sit at positions 6 and 7, close enough to trade places. The fix is a decision, and Claude proposes it with the rows in front of it: pick the page that should own the search, fold the best of the others into it, redirect what is left, and point the internal links at the winner. This one is on the list for the next pass, which is the point of showing it. The loop does not run out of work.

The third loop is the cheapest and usually the fastest to pay back. Take a page that already ranks, list every query it ranks for, and look for the ones it is weak on. A page that sits first for ten phrasings of a question and fifth for the eleventh does not need a rewrite. It needs that eleventh phrasing worked in.

Here is a real one. Over the last 28 days this page earned 108 clicks from 19,993 impressions at an average position of 2.5:

/blog/how-to-add-links-to-linkedin-posts-and-make-them-stand-out

Seven of the eight sit between position 1.0 and 2.3. One does not: “add link to linkedin post”, the short imperative form, is at 4.4. That is the weak one. The edit is small: use that exact phrasing in a subheading and answer it in the first sentence under it. Then leave the rest of the page alone, because the rest of the page is working.

The same loop also runs across the whole site. The connector found 95 queries sitting at position 15 or better with at least 100 impressions in 28 days. These are the six with the most impressions.

For context, these are the pages doing most of the work right now. Every one of them is on page one on average.

None of this depends on a person remembering to do it. Murkuz gives you the skill file for it: a prompt that Claude runs as a loop of its own, on your machine or wherever you run your agents. Connect your sites, hand Claude the prompt, and it works through all three loops and comes back with one prioritised list.

Here is a starter version you can paste into Claude today, with the Murkuz connector attached.