CASE STUDY ‧ Lightning AI

Leading design at an AI company

Project Background

I met Colette, the founder of Lightning AI, when she was a customer of ours at Roadmap. They were a big fan of the design and experience we created at Roadmap, and entrusted me to do the same for Lightning AI.

After briefly contracting, I joined as Head of Design, taking ownership over the visual identity and experience of Lightning AI from marketing to product to branding and everything in between.

"Allison has an incredible eye for design. She built our brand assets from the ground up, and has been fundamental in the way we think about marketing our business."

– Colette Nataf, CEO Lightning AI


Brand Research
Understanding the problem

Brands spend an enormous amount of money advertising on social networks. But Facebook limits access to their audience groups, leaving brands to advertise to the same people over and over again.

After awhile, conversion rates go down along with their return on ad spend (ROAS). Talking to marketers proved this was a huge pain point for them. When Colette realized how much manual work people were doing only to get dissapointing or inconsistent results and at the expense of a massive marketing budget, she was determined to build something better.

Understanding their technology

Lightning AI's technology experiments with audience groups outside of the ones they can typically reach (or even think to do so), teaching Facebook's algorithm as new ones are found. It also automates the process of launching and managing campaigns. Lightning AI takes an overall budget and decides how to spend it and which ads to show to who in order to get the best possible return on ad spend.

Brand Audit

Before designing anything I wanted to learn more about what they hated and loved about their existing brand and website. Two things needed to stay: the lightning bolt, and the color teal. Everything else was up for grabs.

Logo Before
Website Before
Discovery & Design

We had a relatively tight deadline of a few weeks, so they could unveil the new brand during demo day after completing their accelerator program.

To begin, I experimented with dozens of variations of lightning bolts and explored typography options to compliment the mark. Once we were confident with a direction I started working in color. Although teal was a constraint, the value of that color and a complimentary palette needed to be defined.

A flexible system

There are 3 variations of the logo. The full logo and two options for logo marks. This was intential so we could have a brand that would work great under any circumstance at any scale.

Logo variations

The primary logo, stacked in full color makes the most impact.

An alternative horizontal logo, for use cases where vertical height is limited.

A mark with the bolt standing in for the word "Lightning" alongside AI.

The bolt on it's own, only to be used alongside the full logo unless inside of a LAI product.

Designing and building the website

I designed, coded, and wrote content for Lightning AI's new website. We launched a brand and website within a month just in time for demo day.

After going live I wanted to push our website a step further and breathe even more life into our brand with some custom illustrations. I knew what we needed was beyond my capabilities, so we worked with an illustrator to bring my concepts to life.

The sketch I created for the illustrator

The final illustration created by FireArt Studio

Website design

The final illustration created by FireArt Studio

Results are a big emphasis for our product and we wanted it to be the same for the website. I created some graphs to go alongside real results from our customers.

"Allison manages both brand design and product design, and she excels in both. Given a high-level vision, she can turn ideas into user flows and prototypes for user research and engineering."

– Colette Nataf, CEO Lightning AI


Unique challenges with AI

There were a few steps the user needed to do in order to spin things up with Lightning AI like Facebook OAuth & permissions to manage campaigns. They also need to give us a spend budget and create a funnel they want to optimize towards.

But after that, the algorithm runs on its own behind the scenes doing all of the hard work and heavy lifting for the user.

Initially the plan was to build a self-serve SaaS product on top of that algorithm. We started working through flows and design, and even built out some functionality.

A pivot from Self-Serve SaaS

We ultimately decided to take the axe to our SaaS product.

By putting a UI on the product we were actually doing the AI a disservice. The bread and butter of what makes Lightning AI great is how it just works, and that's what our customers loved about it. We took something complicated and made it simple, they saw real results. Plus they enjoyed working with an account manager who helped them succeed – why would we take that away? What we realized was they didn't want another app in their workflow, or to have to learn how to manage campaigns outside of Facebook. And from a business perspective, people were already paying us for how we currently operated with managed services and the product development was a huge expense. So we pivoted.

Back to expirimentation

Instead of building a self-serve product around onboarding and capturing those must-have inputs needed for the algorithm, we decided to handle that for them. We started to envision something as simple as a toggle to turn on and off the algorthm, and simple email or Slack notifications for the results of the AI working for them. Our entire proposition is to create less work and free up their time to do more creative work – so this pivot would empower them to do just that.

We decided to break apart what we had initially seen as one huge product into tiny AI tools that work for marketers. This would allow us to experiment and continue learning from our customers, and provide some great lead generation for our website.