Try a preliminary sift to streamline your card-sorts

A person pointing to a card as they sort them, watched by the user researcher
Photo by UX Indonesia on Unsplash

Here’s my favourite tip for card sorting. I get each participant to first of all put the cards into four categories:

  1. Things I definitely need or would do right now
  2. Things I have done or needed in the past
  3. Things I might do or need in the future, or that are relevant to someone else
  4. Things I don’t understand or I’m not sure about.

I always use category 1 above (definitely need or would do right now); the others vary a bit according to whim and circumstances but I do nearly always have category 4 as a ‘jargon busting’ category.

I call this a ‘preliminary sift’.

The preliminary sift helps to focus on what matters

I use this preliminary sift to work out what is actually important to the participant, because I’ve found that their grouping is much more accurate for  things they are truly interested in (category 1). People (mostly) come to websites or use intranets to solve the problem they have today; it’s less common to look up ‘something I need next week’ or ‘something I used a month ago’. It’s that ‘thing I need right now’ that they want.

A preliminary sift gets through lots of cards

Taking away all the other stuff and getting them to focus on the ‘definitely need right now’ means that I can include far more cards initially. The preliminary sift is easy and participants whip through it.

An extreme example: we had 200 cards, which would have taken ages and been quite boring for participants.

  • We put participants into a group of 4. Each participant got 50 cards and picked out any category 1 or 4 card. (Rationale: if one person needs it now, we need it. If one person is confused by it, that’s enough to tell us that we have naming/terminology work to do on that card).
  • Anything in categories 2 or 3 they handed on to another participant. (Rationale: another participant might put it into categories 1 or 2).

We rapidly whittled the whole thing down into maybe 20 to 30 cards that were genuinely and immediately interesting to at least one person.

And the participants enjoyed it. By the time we got to sorting the remaining cards, they were happily working together to get their best groupings sorted.

In another example: a client was convinced that people might want to do about 100 things. Participants didn’t: they chose a maximum of five of those things. Our client learned a lot about focus.

Bonus: fewer cards makes analysis of the final groups easier

In practice, I do very little analysis of the cards that went into groups 2 and 3 in the preliminary sift. Obviously I keep an eye on them: if a card persistently ends up in category 2 ‘I used this in the past but not longer need it’, that may signal content that has become tired and needs refreshing. If a card is constantly appearing in category 3 ‘I might use this in the future’, I look out for known future events to see whether we might need to make a feature of them. But I don’t worry too much about it.

The category 4 cards get close attention because, as I mentioned, they signify naming or terminology problems. Useful to know about, but if a participant doesn’t know what it is then I’m not going to worry about where they may put it. No grouping analysis for those.

And that leaves me with category 1 in a few solid groups of far fewer cards. Way easier to think about and to report on.

Or as I would put it: streamlined.