Page MenuHomePhabricator

Computer-aided tagging usability study
Closed, ResolvedPublic

Description

Usability test

Testing script: https://docs.google.com/document/d/1sWUs9HFPwR4wP2Tb1h6LM38yNCPYfpgS3o1uz4yiiPI/edit#

Clickable prototype: https://wikimedia.invisionapp.com/share/BZTUWXU3E95

Testing goals

The primary goals of this study are to gain insights on the following aspects of the computer-aided depicts tagging tool:
Identify usability improvements the tag review flow
Gauge the user’s impression of suggested tags
Measure the understanding of the benefits of this feature

Research questions

First impressions and general understanding of depicts tags

  • Are the provided onboarding screens sufficient for understanding this tool?
  • Can users successfully describe what the difference is between “Uploads” and “Popular images” in the navigation?
  • What are external users’ first impressions of computer-suggested tags?

Performing key tasks

  • Can users perform the task of confirming and rejecting tags?
  • Can users successfully ignore a tag if they are unsure whether to add it?
  • Can users successfully undo a confirmed or a rejected tag?
  • Can users successfully “zoom” in on an image and get back to the tagging tool?
  • Do they have difficulty performing the overall task?

Gauging general sentiment

  • Do users find reviewing tags to be friendly and easy to use?
  • Is the quality and appropriateness of tags acceptable for users?
  • Do users feel the need to confirm tags before publishing changes?

Findings from prototype v1

Onboarding screen

✅ All users had a correct understanding of the information on the onboarding page
✏️ Two users wondered if they will be able to manually add tags

Personal uploads vs popular

✅ Five users correctly guessed that Personal uploads meant a user’s uploads
✅ Three users correctly guessed that Popular meant uploads by others
❌ One user incorrectly guessed that Popular meant a brand or popular seller
❌ One user incorrectly guessed that Popular meant what’s important right now on social media or things that are being currently searched for

Tags: first impression

✅ All users correctly guessed that tapping on the ‘X’ of each tag would reject those tags
✅ Three users correctly guessed that they could manually tap a tag for approval
✏️ Two users thought that they’d approve the tags by just publishing them (and rejecting ones they didn’t want)
✏️ Two users wanted to be able to add their own tags and thought that was missing
✏️ One user noted that, based on the onboarding screen mentioning it, that they were expecting an ignore option
❌ One user incorrectly thought that tapping on a tag would show a different suggestion

Skipping images

✅ All users correctly guessed that skipping images would take them to the next image
✅ Three users had no expectation that they’d see a skipped image again
✏️ Two users hoped they would see skipped images either “down the road” or in some “skipped images” area

Zooming in and getting more info

✅ All users correctly guessed that tapping on the image would open up the image in a larger view
✅ All users correctly guessed “Details” would take them to a page that gave them media info (size, format, uploader, etc.)

Unfamiliar tags

✏️ Four users would ‘X’ out a tag if they were unfamiliar with it
✏️ One user would google the name of any tag if they were unfamiliar with it

Interacting with tags

✅ All users were able to successfully confirm a tag
✅ All users were able to successfully reject a tag
✅ All users were able to successfully undo a rejected tag

Publishing

✅ All users correctly guessed what would happen if they were to publish the tags they reviewed

Finalizing tags before publishing

✅ All users correctly guessed what this screen is for (confirming tags before publishing)
✏️ Two users stated that they would prefer to see the image on the Confirm screen
✏️ One user did not expect this screen and thought it would be sufficient to double-check their own work on the tagging screen
❌ One user didn’t like the layout of this screen as it didn’t fit the previous one

Cleared queue

✅ Four users guessed that tapping Tag popular images would enable them to tag images uploaded by others
✏️ Two users thought it would be good to be prompted to upload more images
✏️ Two users thought it would be useful to see the work they had already done on this screen
✏️ One user hoped that Popular images could suggest similar images to the ones they uploaded, since they would likely be more informed about those particular types of images
✏️ One user thought that this screen needed more color
❌ One user thought popular meant what was trending that week

Suggested updates

Onboarding screen
  • Iterate on stand-in graphic
Tag interaction
  • Since users mostly treated unfamiliar tags by rejecting (vs ignoring) them, consider simplifying the interaction to only suggested and confirmed tags. This would mean removing the X's, and tapping on a tag will either confirm it, and tapping a confirmed tag would unconfirm it (making it suggested again)
Confirmation screen
  • Add image to Confirmation screen
  • If tag interaction is simplified, include only confirmed tags on this screen
Cleared queue
  • Iterate on stand-in graphic
  • Consider one of the following as a secondary CTA:
    • link to user's Uploads page
    • link to user's Contributions page
    • CTA to upload more images

Event Timeline

PDrouin-WMF renamed this task from [stub] MAD tagging usability test to [stub] MAD tagging usability study.Aug 21 2019, 2:29 PM
PDrouin-WMF renamed this task from [stub] MAD tagging usability study to Computer-aided tagging usability study.Sep 10 2019, 5:29 PM
PDrouin-WMF updated the task description. (Show Details)
PDrouin-WMF updated the task description. (Show Details)
PDrouin-WMF updated the task description. (Show Details)
PDrouin-WMF added a subscriber: Ramsey-WMF.

Updates in the works:

Things I changed or added:

  • Subhead changed from “Images to review” to “Image X of Y for review”
  • De-emphasized the image title size
  • Removed the “X” functionality of the tags and simplified the tag design
  • Added an image to the “Confirm tags” dialog
  • Card design adjustment: moved away from the iOS card style to one that resembles the card-like boxes on Content Translation to fit better with mobile/desktop web
  • Please note the “toast” notification — not sure how we’re able to do that with what we have to work with, but something like that will be needed

Things needing iteration in the browser:

  • how to handle mobile to desktop layout changes (especially max width/height for images)
  • tap target of tags: though I’ve made them fit the styling more (see Recent Changes on Wikipedia), I am concerned about the height of the tags, and will need help keeping it within the style guide but also make it mobile friendly
  • Publish/reset/skip buttons sticky on mobile?