Aurora Energy: electrical outage communications

The brief

Aurora Energy is a lines company — when a powerline goes down, they are who fixes it. They manage the electrical distribution infrastructure for over 87,000 people in Otago, New Zealand, and were under intense public and media scrutiny for infrastructure reliability. The public had lost trust, and it was costing them money.

A power outage is the only moment most customers ever deal with the company directly — and there was no public tool for them to self-serve when it happened. I led research and designed the outages map from scratch, as lead designer on a wider website redesign.

The problem

78% of customer service calls were outage-related — spiking sharply during widespread events, overwhelming the team at exactly the moment they needed to focus on repairs. Customers had nowhere to go for accurate information, so they called.

What I designed

A mobile-first map built around one operational reality: if the power is out, the laptop is unavailable. Outages on the homepage — one page load, no navigation required during a crisis. A drawer interaction that preserved map context while revealing outage detail. An accessible list view alongside the map for users who couldn’t use it.

Results

These are indicators of success rather than definitive proof — the survey can’t be attributed to the website alone, and I’d have liked more absolute numbers on the call reduction.

“The only time I have interacted with Aurora was when I had a planned outage.”

Aurora Energy customer

Research: two things that changed the design

Eight interviews spanned residents, business owners, contractors, farmers, people working from home, and an older adult who was also an advocate for that group. Alongside them: website analytics, internal search data, and 526 customer service calls from the previous 12 months. Moderated card sorting with the same participants and tree-testing (n=50) informed the information architecture. My colleague and I ran a persona workshop with internal stakeholders to build shared alignment on Aurora Energy’s diverse users before moving to design.

Accuracy over speed

Customers don’t expect instant fixes — they expect honest information. Inaccurate updates, disappeared alerts, or missing context cause significant frustration.

Knowing that something has changed — even for the worse — is preferable to uncertainty.

Planned = plannable

Planned outages are acceptable because they’re manageable. Unplanned outages without information are not.

Framing the problem

How might we support decision making for customers before, during, and after electrical outages so public trust is restored?

Improved information around outages would reduce customer service calls and ultimately improve perception of Aurora. Success would be measured as a reduction in the percentage of customer calls regarding outages.

When the power goes out

Almost every participant in the research described mobile as their default during outages for one reason: if the power is out, the laptop is unavailable.

Competitors typically opened maps in a separate application — a full-screen experience, but one that required an extra navigation step at exactly the moment a user with no power can least afford it.

Why the drawer won

The outages map needed to function within Aurora’s website as a content block — not a standalone application. This created a specific interaction problem: how do you reveal outage detail without losing map context?

Modals disrupt spatial orientation

The user loses their place on the map — in a crisis context, this creates unnecessary re-orientation.

Popups create nested complexity

An embedded map already has limited interaction surface; a popup within that space compounds the problem and risks obscuring the surrounding map context.

Drawer preserves orientation while revealing detail

The map doesn’t disappear — it narrows. For a crisis decision, this matters: the user can verify they’re looking at the right area while reading the detail.

Accessibility as a design decision

Embedded maps don’t provide an accessible experience for all users. I included a list view — not primarily because users preferred it (testing confirmed the map was the default) but because it was the right call for the users who would need it most and couldn’t use the map.

This decision had structural consequences beyond the accessibility case: including a list view required a map/list toggle, which then influenced where every other UI element lived — search, outage tabs, and legends all fell into place around that toggle.

What testing changed

I brought six participants from the original research panel back for usability testing, over video call, against Figma prototypes covering three key scenarios.

Zoom reset button

Users were getting lost at high zoom levels with no easy way to return to their area. We added a zoom-out control.

”See more” cue

Users weren’t discovering that the drawer could be expanded. Adding text alongside the expand icon particularly helped older participants less familiar with the icon pattern alone.

Suburb labels

The list view was rendering as an undifferentiated list of street names. Adding suburb context to each entry gave users enough orientation to identify whether an outage was near them. This required co-ordination with outages API development.

Area polygons

The outage markers didn’t show the specific affected areas. We couldn’t add address level information due to customer privacy, so we added area polygons instead.

The changes I wasn’t able to stop

This was a highly collaborative process with the technical lead from Effect and Aurora Energy’s API developer, to ensure we served the right information and could create a reliable experience.

The orange brand colour was integrated into the active state, which isn’t accessible. That was a decision at C-suite level, and the project stakeholders didn’t want to challenge it — a bit of managing up that I didn’t win. I do wish I had fought back on that.

Final designs

Reflection

Customer service calls are a lagging indicator. I should have measured a more direct leading indicator — the percentage of users successfully finding their outage information on first visit. I’d also expand the user testing pool outside of previous interviewees to ensure a non-biased test sample, and quantify usability testing success: task completion, time to complete, error rate.

What this project changed about how I approach information hierarchy: when someone’s power is out and they’re on a phone, they can’t evaluate three equally-weighted options. The drawer wasn’t just an interaction pattern — it was a decision to give the user one clear thing rather than a choice between things.

If I find myself giving users two equally prominent paths, I ask whether I’ve actually resolved the decision or just deferred it to them.