I treat douli.com as a product. That means it has a backlog, planning cycles, and a pre-planning data ritual — the same discipline I apply to product work professionally, tested first on something small enough that I can move fast and see the results quickly.
The ritual is 15 minutes with Claude and my Matomo data, before I decide what to write or change on this site. It sounds minor. It has changed which content I prioritise, which ideas I park, and which assumptions I stop making about my audience.
A content plan built without behavioural data isn’t a plan. It’s a list of assumptions with dates attached.
Why the dashboard isn’t enough
Matomo gives me a clean dashboard for douli.com. I can see sessions, bounce rate, traffic sources, top pages. It’s useful for monitoring. It’s almost useless for deciding what to do next.
The dashboard tells me what happened. It doesn’t tell me what it means for what I should build or write. The question I’m usually trying to answer before a planning cycle isn’t “what are the numbers?” — it’s “which problem is worth solving?” That requires a different kind of conversation with the data.
How the ritual works
I connect Claude to Matomo through the MCP server I built and described in How I connected Claude to my WordPress analytics. The connection means I can query my site data in plain language — no custom reports, no exports, no pivot tables. I ask a question, I get a specific answer.
The ritual runs before any significant planning decision: what to write next, whether to update an existing post, whether a page is underperforming or just receiving the wrong traffic. I want the data to challenge my assumptions, not confirm them — so I run it before I’ve decided anything.
The four questions I ask every time
Which content actually engaged readers — and which just got traffic?
I look at bounce rate and average time on page together, not separately. A post with 500 visits and 95% bounce is getting found and immediately abandoned. A post with 20 visits and 13% bounce is being read. Those are different problems with different solutions. The volume metric flatters the wrong thing.
Where did readers come from, and does it match what they found?
Traffic source matters because it tells me what expectation a reader arrived with. LinkedIn referrals to douli.com have a 41% bounce rate and over 10 minutes average session. Organic search has 85% bounce and under a minute. That’s two different audiences with different intent. Content written for one doesn’t serve the other. Knowing which source sent engaged readers tells me where to double down — and which content is currently mismatched to its audience.
Did anything I published change behaviour on the site?
This is the question most content reviews skip. Publishing frequency is easy to measure. Behavioural change is harder. I ask Claude to compare actions per visit and session duration before and after a post went live. If the numbers didn’t move, that’s useful information — it means the content isn’t doing what I thought it would, and the next piece shouldn’t follow the same pattern.
Where are readers going when they leave — and why?
Exit page data tells me where journeys end. The planning question is: is this an expectation gap, a content gap, or just a natural endpoint? Each has a different implication. A high exit rate on the homepage after someone’s read three posts is fine — they got what they came for. A high exit rate on a post mid-way through an argument is not. Claude helps me turn the exit pattern into a hypothesis rather than just a number.
What this practice is actually testing
Treating douli.com as a product is partly about the blog, and partly about proving to myself that the methods I advocate professionally actually work at small scale. If I can’t apply data-informed planning to a 150-visit-a-month blog, the argument that it scales to larger products becomes harder to make with conviction.
The four questions above aren’t specific to content. They’re the same questions I’d ask before any sprint: what worked, what didn’t, which audience am I serving, and where are they dropping off? The context is different. The discipline is the same.
Jeff Patton’s framing of outcomes over output applies here: the goal isn’t to publish more. It’s to publish things that change how readers engage with the site. The pre-planning ritual is how I keep that distinction honest.
TL;DR: I treat douli.com as a product and run a 15-minute pre-planning ritual using Claude and Matomo before any content decision. Four questions: which content engaged readers, where they came from, whether publishing changed behaviour, and where they leave. The same discipline applies to professional product work — this is where I test it at small scale.
Delphine Ragazzi is a Product Owner with 20 years in digital analytics and product delivery. About Delphine →
