Claude + Google Analytics + Search Console: steering your content without burning your life

For a long time I published by instinct. I wrote whatever crossed my mind, put it online, prayed. Some articles took off, others sank in silence. I rarely understood why.

Today, I spend twenty minutes on Monday morning and I know exactly what my readers are looking for, what they find, what disappoints them. Not because I became a data analyst. Because Claude reads the data for me.

The problem

My site has been publishing articles and recipes for years. The volume is too big to keep in my head. Search Console surfaces hundreds of queries every week. Analytics floods me with metrics. Facing that mass, I used to close tabs and keep publishing by feel.

The real problem is not a lack of data. It is the opposite. There is so much information that no human can read it every week without losing half a day. Result: articles that deserve an update collect dust, topics that are taking off go unnoticed, and we keep writing about what pleases us rather than what truly helps.

What I set up

A simple pipeline, running without me, delivering a summary every Monday.

Connect the two sources. Google Analytics tells me what people do once on the site: pageviews, time spent, paths, bounce rate. Search Console tells me what they searched for before landing: queries, positions, clicks, impressions. Together, that is the full picture.

A script that pulls the data. Every night, a small script calls both APIs and stores the last seven days of metrics in a local folder. I never open Google's interfaces, I no longer need to.

Claude that analyzes. On Monday morning, I run a command. Claude reads the files for the week, compares with the week before, and produces a three-part summary. Pages that are rising. Pages that are falling. Queries where I am on the second page, just one step away from the first.

A decision rule. Claude does not decide for me. It sorts. I choose: rewrite, update, leave alone, or redirect. But the triage is done. I no longer ask myself where to start.

The result

Before, I spent about two hours a week looking at stats, and reluctantly. Today I spend twenty minutes reading the summary and deciding. The two hours I reclaim go into writing.

More importantly, my decisions are better. On the last eight updates I ran from the summary, six climbed back in visibility within two weeks. When I used to publish by feel, the ratio was closer to one in three.

A pleasant side effect. Because I see rising queries, I catch topics that start to climb before they become saturated. Two articles I wrote this year came from that early detection. They now bring in more traffic than pillars I had worked on for months.

How you can replicate this

What you are reading is not a Rube Goldberg machine. Three building blocks, three days of setup if you start from scratch, and then autopilot.

First block, access. Open the Google Analytics and Search Console accounts, retrieve the API keys, put them in a secure environment file. Do not cut corners here: your keys are your data, they do not belong in a public file or a prompt.

Second block, the pull script. You can ask Claude for it, it delivers one in minutes. What matters is to tune it around metrics that make sense for you. Do not vacuum everything. Focus on fifteen to twenty well-chosen metrics.

Third block, the analysis prompt. This is where you really win. Give Claude the context of your activity, your goals, the questions you want answered. A lukewarm prompt produces a lukewarm summary. A precise prompt produces a sharp one.

One tip: do not try to automate everything in week one. Start with Search Console alone, read the summary, adjust the prompt. When it runs well, add Analytics. You will move three times faster than if you try to connect everything at once.

If you want to see what this looks like on your site, I can show you live.

Comments

Be the first to comment.