View all Insights

Jul 28, 2022 / 2 min read

Atomic Search 3.0 unlocks +14% in conversion rate gains

We’re locked on to search conversion rate, and we’re pleased to announce a breakthrough.

Atomic is constantly iterating on various steps in the user journey to connect their work account. This journey begins with the crucial step of a user finding a system that they recognize. Here is a brief recap of our own journey, designing a better search.

Rewind to March, 2021: Reacting to search data

March, 2021 A/B Experiment

In early 2021, Atomic tested a hypothesis that a large segment of users desired to login to their payroll provider directly, rather than search for their employer. Although this contradicted pre-product survey data, user search query volume for payroll systems was unexpectedly high. Our approach was to add a carousel of payroll providers to the top of the page and watch the results.

Although the 12% increase to our total funnel was exciting, we found something even more interesting. Users that logged in to their payroll system instead of their employer were 2x more likely to know their credentials.


This search experience quickly became the industry standard, as other players in our market patterned their search after this design.

Search reset


Heading into 2022, the Atomic design team was looking for a win on search. After numerous adjustments to our current design without significant lift, we decided it was time to wipe the slate clean and rebuild. Our first experiment was to remove all context and observe how users behave with a Google experience: just a text box and a prompt. We quickly learned that context was important, as made clear by a 10% drop in converting users to the next step in our funnel during our A/B test.

Bringing context back


Our next iteration was to layer in the context, as minimally as possible. With the growing number of ways to earn income, Atomic made it a design priority to convey flexibility. No matter the user’s source of income, or mental model on how they could interact with search, we aimed to adapt. Although this iteration was a formidable competitor to our existing search experience, the A/B test results were not statistically significant, and we decided to look elsewhere.

Context + Familiarity


It is common for users of Atomic’s SDK to have multiple apps they frequently use to interact with work accounts. By leveraging familiarity, we created a search experience that mimics the user’s day-to-day experience, even down to the exact app icons they regularly see on their device. We also added a clear call-to-action to use our search feature if they do not see what they are looking for.

With this iteration, our A/B test is indicating:

  •  +6% increase to top of funnel search conversion.
  •  +10% increase in authentication attempt rate.
  •  +14% increase through the full funnel of completing a transaction.

What’s next?

We will increase traffic to our improved version after we make some design adjustments to improve the consistency of the end to end user experience. Furthermore, if you are an Atomic customer, you don’t need to do anything! To summarize, Atomic will continue to analyze our data and plan out the next iteration of Atomic search.