Configuring sensors' orientation on the installed machine is part of Augury's Halo installation process. This step was so poorly constructed in the app, it was nearly impossible to get it right.

I owned this project end to end - from recognizing and defining the problem, to initiating and planning the research, to executing it, to analysing the findings, to communicating recommendations with all stakeholders.

To learn more about Augury's Halo installation, check out how I created the holistic IoT solution.

User research

Human centered design

The problem

On a mission towards a human centered design

1. The instructions were very technical. People struggled understanding what they were supposed to do.

2. The letters-combo options in the menu were visually indistinguishable and error prone.

3. There were three types of letter sets. H-V-A stand for Horizontal, Vertical and Axial. OK-ish. How about V-R1-R2? Now that's baffling.

4. Users had to compromise or guess the answer when in doubt.

5. Internal users and account managers repeatedly complained about this step. I didn't know exactly what was wrong, but evidently something was not right.

The objective

Creating a product that works the way people work

Configuring sensors' orientation successfully was a challenging task even for the experts. Mistakes meant bad data would be fed to the algorithms from day 1 and forever, leading to bad diagnostics, thus wrong insights for customers, and damage to credibility for Augury.

The objective I set for myself was to transform this technical POC into a human centered product.

  • Bring users a sense of achievement instead of frustration.
  • Incorporate an educational onboarding experience for new users.
  • And of course reduce mistakes significantly.

Scoping a research plan

Minimal effort, extraordinary impact

This problem wasn't a priority, with major parts of the workflow still missing. I put it on the back burner, and hijacked a research when opportunity allowed.

Ethnographic research

I performed observations and contextual interviews on a few occasions with several user types.

"Eat your own dog food"

I participated in an installation. The experience was paramount to gain empathy and deep understanding of the problems.

Prototype and Usability testing

I gathered feedback through prototypes and interviews. Since this task was to be performed in extreme physical environment, it was essential to test it with the real thing as well.


Cognitive mega challenge

I knew the task wasn't intuitive, but boy, was I surprised! Here's what I found out during the ethnographic research.

1. Users' workflow varied based on physical factors: e.g a machine was accessible for a short time window; Installers wore gloves when using adhesives and set sensors' orientation in batches. Sometimes sensors' orientation wasn't visible after installation.

2. Matching the sensor's orientation in relation to the machine required high level visual perceptual skills: Three-dimensional visualisation; Visual memory.

3. What's more, it resulted in cognition overload: by the time you recognized the third dimension, you forgot the first one.

4. Installers start their day at the break of dawn, following an evening flight. They work long hours of hard labor under very hot or freezing cold weather. They are bound to complete the work on time for flight, before the machines are back on, before employees leave, etc. All of these factors put them under enormous stress that narrows their focus and hurts their performance.

First time user

After doing this work for a whole day, a few days on a row, you feel like your brain is left at home.

Early adopter

When I select the sensor's mounting plane, one of the three letter couples is known. Why doesn't the app automatically remove the irrelevant options? Isn't it supposed to be smart?


Interaction and Visual exploration

Since configuring the sensor's orientation required challenging three-dimensional visualisation, I recommended to make it easier by adding a visual aspect to the task. I based the instructions on visuals, with minimal descriptive text. I also changed the voice and tone from very technical to friendly and human.

I explored an image-flipping mental model, familiar from photo editing apps such as VSCO and Instagram. Intuitively, the sensor would visually flip on every click, perhaps animated. It felt like an excellent direction. However I dropped it at this rough stage, since the development effort was too high.

In these two options I examined simpler variations to display the orientation. Still visual, but cleaner and generic. With the busy user in mind, I reduced clicks by making the most common orientation selected by default. Having a visual upfront also served as an educational tip for the new user.

In order to explain what the orientation means I split the action to two steps. It also contributed to reducing the cognitive effort. The down side was not being able to see the entire selection at once, and adding an extra click.

After many trials I reached a good solution: All info is laid out on the main screen. There's no need to go even one click further. Big and simple elements are easy to select and read in dim light. An indication guides the non-expert.

Diving deep into science

Vibration analysis and algorithms

I consulted with algorithm developers and vibration analysts within the company, about recommended mounting orientation. We also discussed edge cases and technical debt, that caused confusion, mistakes in configuration or mounting sensors in bad locations.

In addition, I designed the flow's logic. Once the user selected a mounting plane, the second field is updated: impossible options are removed. Further more, very rare and unrecommended options are also hidden from the main flow. Now the user would select 1 of 2 options instead of 1 of 6!

Usability tests and feedback

Desperate times call for Guerrilla research

As mentioned, this project wasn't a priority. I managed to squeeze in an improvised lab in the office to ran a simple usability test on the workflow. The selected persona was a non-expert new user. I wanted to challenge the app and see if it's "dummy proof".

I summarised my observations in a Rainbow spreadsheet. When tested with the beta app - the flow was easily completed, with the orientation configuration being the only exception. My assumption for the required focus area was verified. After a couple attempts, users got the principle and successfully completed the task.

The response to the new design was good. Users were excited! Surprisingly, when experienced users tried it out, they mentioned they are missing the letter combo. My recommendation was to leave out the letters despite their feedback, and collect data, assuming this was just an old habit, and relying on illustrations would form into a new habit soon enough.

Onboarding guidelines

Making the most of it

As intuitive as it can be, we're still talking about a very professional task performed by non-professionals. They needed some education. Understanding the main principle was now easy, and the users had a much more pleasant experience. It was time to handle the next level of expertise, guiding a user who's no longer a beginner. At the time I also worked on a big initiative to create the company's onboarding program and help center. It was a good timing. The orientation guidelines could be embedded both in the app and the emerging help center. I used it first to train Augury's Support and Customer Success managers, and they could make use of it when assisting end customers remotely. It was also useful for internal onboarding of new field employees.