Analytics

Collecting data on the use of interactives is a requirement and will ensure we can make more confident and informed decisions in the future to improve our digital outputs.

Page URLs within the application should be human readable and describe the content of the page, so the analytics are easier to interpret. E.g. For a portrait you could display: artist-name-of-work-1971

More stringent documentation on analytics will be written as these are implemented, tested and revised in the near future. In the meantime, use the following as a guide of what data might be tracked and how it could be used.

Examples of metrics to be recorded:

  • Views (pages, video plays, images in a gallery, etc)
  • Navigation paths
  • Failure points
  • Specific tasks/events, for example:
    • Language switching
    • Subtitle use (which languages are used most)
    • Gloss use
  • Social sharing (eg, how often, which platforms)
  • Email sign ups
  • General error logs
  • Session length

Examples of what we could learn from these analytics:

  • Usage of interactives
    • What is popular? What isn’t?
    • Do certain areas of the museum perform above or below average? Does moving them improve the engagement?
  • Trends of exiting interactives
    • Do users exit interactives at predictable points (eg, when first seeing the menu)?
  • Which features to support
    • Subtitles – we can learn which languages get used the most to and focus efforts on them
    • Social sharing – which platforms should we support and which ones don’t get used
    • Accessibility menu – is it discoverable and usable?
  • Which design patterns promote interaction and which don’t
    • For example, comparing two design patterns over time with A/B testing to see which gets the best engagement

Things to keep in mind when using analytics

  • Be wary of confirmation bias. Be sure you aren’t just confirming what you want to see. Don’t stretch analytics beyond what they are really telling you. For example:
    • More click-throughs but a higher abandonment rate is not better
    • Increase in ‘clicks’ but decreased time spent on content is not better (eg, users are just ‘playing with the technology’ not engaging with the content)
  • Remember what analytics can and can’t measure. Many of the factors we want to track are impossible to do so with just analytics. For example:
    • Did they actually watch the video, or did they just press the Play button?
    • Did they enjoy this article? Did they care? Did they understand it? Were they moved by it?
  • Supplement analytics with other research. Use the analytics in tandem with other research, such as observations, surveys and more sophisticated visitor tracking solutions.