Checking the “Feel” of Your UI with an Interaction Audit – Web 2.0 Expo Berlin

The author presented the process that in Ebay leads to interaction audit. But, first of all, what is “feel” and why should i matter?

It’s the feel in look and feel, while many times we just consider the look of the interface. The feel can be considered how you operate the user interface with your own hands.

Feel affects:

  • learning curve
  • mental bandwiith needed to operate the user interface
  • user success (or errors)
  • site personality
  • brand promise
  • adoption (or abandonment)

While developing the Ebay interaction audit the team created a flash demo of the different pages and sections of the Ebay sites. What is interesting to note is that by playing this series of screenshots a little fast it’s possible to highlight several inconsistencies, especially in the structure (eg. headers, logos of different size and in different positions). Creating this sort of animations is very valuable but it’s not expensible or complex to do. They use such animation as a sort of benchmark.

Developing an interaction audit was not like following a strict blueprint. It happened that, based on new findings, they noticed how to improve the process.

Project phases

  • strategy
  • data collection
  • analisisy
  • recommentdations


Improvement of user experience on ebay: they made a compelling artifact. They wanted to collect data around the flows. So they imagined flows based on their user behaviour, such as “new users finds an item, bids for it, register as a member”, “new seller lists item for sale, create sellers account”.

They eventually put all on a spreadsheet to see with tasks were common to several flows

Data collection

They used a simple db (filemaker pro). They also captured graphical data.

Database fields used:

  • task and subtask
  • step description
  • page and url
  • action (syntacrtic)
  • screen shot close up
  • instructional text
  • click/keystroke record


They printed each record (flows as storyboards) so they could made comparisons. It was like storyboarding in the comics industry.

Ideas for presenting the findings

  • Radial charts to track feel metrics
  • Emotional flow to track feel effects


Here is a list of some of the analysis results

Affordance: a visual cue that some interaction is offered
Affordance inconsistencies
: inconsistencies: hyperlink, tab

Task: a path to accomplish an immediate goal
Task Inconsistencies: a simple goal is accomplished via multiple path: filtering data, enable/disable section of a form

Data objects: a representation of a data record or other piece of data
Data objects inconsistencies
: a single data tye object representede multiple ways: ebay member

Interaction audit goals:

  • low learning curves
  • consistent clues for actions
  • predictable behavior of affordances
  • instant recognition of interface elements
  • allow ebay member content to shine

After being presented the results, the design team was divided in several group (clean-up teams) based on the inconsistencies found:

  • The clickers: links and buttons
  • The swappers : tabbs and toggles
  • the submitters: forms
  • shufflers: sorting
  • disclosers

They work of “cleaning-up” ended creating reusable components thanks to patterns

Clean-up process

  • find problem area
  • recommend simpler set of interactions
  • document design patterns

Feel metrics

objective: page dimension, number of elements, density….
semi-objective: number of syntactic actions, reloadiness, number of tools switches (keyboard, mouse), popup,
subjective: number of different interaction styles, simplicity, flatness, cognitive load

What we’ve learned:

  • it’s important to check feel
  • an interaction audit can be compelling, actionable and part of a real environment
  • audit should focus on flows and be representative of real experience
  • simple tools work
  • audit for inconsistencies in afforfance, task, and data object
  • clena up objects
  • harder problems require site specific values