I've done these card sort sessions in the past, the websites were massive, the content really needed a strategic and sensible direction. I wasn't sure a card sort exercise would make sense for a small site, a personal endeavour. However, I wanted to give it a try, I wanted to be methodical. This site build process is a true way to put every nuanced knowledge I've gained to real use, and build in public.

How I started

I wrote down thirty things that could go on this site - the bio, the timeline, each case study I could potentially ofeer, the CV, the mentorship invitation, all of it - put each one on a card, and asked three people to group them however made sense to them. No categories from me. No labels. An open card sort, run in Miro.

I picked three people who would disagree. Lucy, a seasoned digital experience professional and a former colleague. Priyanka, an HR recruiter, who's day often goes by screening CVs and verifying credentials. Amnika, a junior colleague standing in for the people who might want to collaborate.

I went in expecting to learn where things go. What I got was a list of the places my own labels stopped making sense. More importantly, I understood how each one thinks so differently (obvious when the intent differs).

The pile that taught me the most

Amnika made seven groups, and then she made an eighth that I had not accounted for. She called it "I am not sure", and she put two cards in it.

The first was Oxford Saïd Business School. Her note: unclear whether this represents a completed course, an ongoing learning experience, or a larger part of the professional journey.

The second was Virtual gallery. Her note: confused what the gallery would showcase and what role it plays within the site experience.

Neither of those is a placement problem. You cannot file a card you cannot read. I had written both labels knowing what I meant, which is the least useful way to write anything. The Oxford card became "currently studying AI for Business". The virtual gallery did not survive - if a card needs a paragraph before it earns a place, it has not earned a place.

That is the finding I keep coming back to. A card sort is supposed to test structure. This one tested my vocabulary, and my vocabulary lost.

Where the opinions split

The disagreement I cared about most was whether case studies and articles are the same thing.

Priyanka and Amnika separated them cleanly. Articles in one pile, case studies in another. Priyanka went further and reframed "SEO and AEO architecture" as a case study rather than an article, which clarified how she was thinking about topics that suit each content type.

Lucy put all of them in one group she named Article Hub. CRO programme, personalisation, AI creative production, all sitting alongside the blog posts. For Lucy, a case study is proof of thinking, not a portfolio showcase.

Two against one, so I kept them separate. But Lucy's version is the one that nags. If a case study is not visibly, structurally different from an article, then she is right and the separation is decoration.

Note: You may find me switching my content pieces on this site, changing labels or placement while I continue to validate these points.

The recruiter wanted things I had not written

Priyanka added two cards that were not in my inventory: a hard skills matrix and a list of brands worked with.

This is the part of research that stings a little, because it is not a preference. It is a gap. She was not saying she liked skills matrices. She was saying that without one she cannot do her job in the ten seconds she gives a site when screening folks. Recruiters need fast credibility signals, and I had built an inventory with none in it.

Lucy added five cards of her own, including location, which she made into its own category. In this market, whether someone is local is not a detail.

The best new idea came from the most junior person

Amnika added a card called Founder Advantages, and wrote her reasoning on it:

Just to help balance this section out. The other two cards already focus on challenges and things you'd change in hindsight, so adding a more positive lens as a separate card would make the narrative feel more rounded and intentional.

My founder content was two cards: lessons from scaling, and three things I'd do differently. Both looking backwards, both slightly apologetic. A colleague spotted the imbalance in a sorting exercise and proposed the fix. That is not a card sort finding. That is content design, offered for free, by the person I had cast as the least experienced in the room.

What I would do differently

Three participants is enough to see patterns and not enough to trust them. Where two agreed and one did not, I have a hypothesis, not a result. I recorded those splits as splits rather than rounding them into consensus, which is the only honest thing to do at this sample size.

I would also stop treating the "missing card" mechanic as a footnote. Every one of the three added cards. Between them they asked for location, a skills matrix, brands, a positive founder piece, and more demonstrable artifacts rather than more write-ups. The things people add when given a blank card are worth more than the things they sort, this was leading to my content design. The sort told me how to arrange what I had. The blank cards told me what I was missing.

I did the simplest thing after, I called in for a few more friends to join and help me out in this exercise. My site now is the culmination of a few more inputs, and lot of incremental changes.