Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Statistically significant taxonomy

E6968532ebe51ca317a4074869ebd2a8?s=47 Matt Jukes
February 18, 2017

Statistically significant taxonomy

Talk at World Information Architecture 2017 in Manchester by Jonathan Porton and myself.


Matt Jukes

February 18, 2017


  1. A statistically significant taxonomy.

  2. Matt Jukes ..formerly known as ‘Head of Product, Office for

    National Statistics’
  3. Jonathan Porton ..formerly known as ‘UX Architect, Office for National

  4. 1.What we faced 2.What we did 3.What we learned 4.Questions

  5. 1.What we faced 2.What we did 3.What we learned 4.Questions

  6. The Office for National Statistics is the UK’s largest independent

    producer of official statistics. @jukesie @iauxbod
  7. The website gets approximately 500,000 visitors per month. @jukesie @iauxbod

  8. The website had approximately [shed loads] pages & documents. Publishing

    an additional 
 [boat loads] a week. @jukesie @iauxbod
  9. The annual list of baby names is the single most

    popular release. @jukesie @iauxbod
  10. @jukesi e @jukesi e @jukesie

  11. The site was called
 “..a national embarrassment” 
 in the

    Financial Times. @jukesie @iauxbod
  12. @jukesi e According to the Parliamentary Administration Select Committee;

    Office for National Statistics website makes figures hard to find and statistics are often presented in a confusing way..”
 @jukesie @iauxbod
  13. A respondent to the website satisfaction survey ended their response

    “The ONS website makes me want to cry..”
 @jukesie @iauxbod
  14. The previous taxonomy for the website had 1400 categories that

    went down five levels.
 @jukesie @iauxbod
  15. A number of the categories were in fact empty and

    reflected statistical publications that the ONS did not actually publish.
 @jukesie @iauxbod
  16. The site search was referred to as like 

    Google on LSD”
 @jukesie @iauxbod
  17. 1.What we faced 2.What we did 3.What we learned 4.Questions

  18. We had to persuade a risk-averse organisation to trust us.

    @jukesie @iauxbod
  19. Outlined and agreed the approach and methodology with ONS methodology

    team. @jukesie @iauxbod
  20. Web analytics - analysis to identify patterns within the taxonomy

 @jukesie @iauxbod
  21. None
  22. Previous research - reviewed every minute of every usability test

    conducted on the website. Not the binge watch of choice!
 @jukesie @iauxbod
  23. Competitor analysis - just as applicable as private sector organisations.

    Lots of mini standardisations, but no real consistency @jukesie @iauxbod
  24. Content audit - what was there, why was it there

    and how the hell did it get there? Reviewed and consolidated structure and content. @jukesie @iauxbod
  25. Card sort - closed, compromised and complicated. Yet still a

    step forward. 200 participants, moderated and unmoderated.
 @jukesie @iauxbod
  26. Tree tests - proposed vs status quo and benchmarking the

    hell out of things. 100+ participants, moderated and unmoderated. Tested and iterated...a lot
 @jukesie @iauxbod
  27. None
  28. None
  29. None
  30. The new (not quite a) taxonomy for the website has

    173 categories and three levels.
 @jukesie @iauxbod
  31. 1.What we faced 2.What we did 3.What we learned 4.Questions

  32. Everyone has a plan until they get punched in the

 Mike Tyson
  33. Real life gets in the way of implementing research no

    matter how comprehensive.
 @jukesie @iauxbod
  34. There was consensus as to what made up the main

    navigation categories - unfortunately nobody agreed what fitted under them.
 @jukesie @iauxbod
  35. Usability tests on live website show top level is essentially

    surplus to requirements. Exposing the 2nd level makes the user’s life so much simpler! @jukesie @iauxbod
  36. Content confuses as often as clarifies and specialised terminology is

    a burden even to expert users. 
 @jukesie @iauxbod
  37. In removing friction from publishing we had failed to capture

    sufficient metadata. This became a major issue.
 @jukesie @iauxbod
  38. Internal politics creep back in and the requirements of the

    ‘business’ start to crowd out the needs of the user.
 @jukesie @iauxbod
  39. If I had my time again...

  40. None
  41. It should have been an open card sort.

  42. 1.What we faced 2.What we did 3.What we learned 4.Questions

  43. @jukesi e Thanks.
 @jukesie and @iauxbod