Breaking the guess-and-check loop
A cross-team hack-day initiative that let users preview the fields and data they'd get from a source — before committing to a connection.
- Role
- Lead designer
- Team
- Connect, Data-In, Key
The field & data preview design in Figma
The challenge
Users found it difficult to understand what they'd actually get when connecting a new data source. They didn't know which fields a given configuration would produce, or what the underlying data would look like — so the standard workflow was: connect a source, check the Data Explorer, reconfigure, repeat. A slow, frustrating guess-and-check loop that also drove avoidable support load.
The idea
Break — or drastically shorten — that loop by surfacing a preview inside the connect flow itself. Two candidate approaches: a live data preview using the current configuration, and a searchable field listing based on that configuration.
Research
I partnered with Solutions, Support and the PLG team to gather insights, sifted through Slack and Intercom for recurring requests, and looked at how competitors handled the same problem. I built early prototypes in Figma ahead of the hack-day so we could react to a concrete design rather than a blank page.
Preparations
Before hacking, we reviewed the research together, walked through the prototypes and mapped pros and cons for each problem space — quota concerns, per-connector effort, backend readiness — so the split into two focused teams was grounded in trade-offs, not guesses.
Group 1 — Data preview
We tested with Instagram Insights: a limited download entry point returned a live sample directly to a new preview step between configuration and connect, complete with pretty field names pulled from the connector schema.
Group 2 — Field listing
Inspired by our Google Analytics field picker, we explored a full field picker for Google Ads — technically possible, but custom fields, attribution windows and filters made it too heavy for the time-box. We pivoted to a simple, configuration-driven field list that shipped the same insight with a fraction of the complexity.
Launch & conclusions
We shipped both a data preview and a searchable field list, A/B tested which one users saw first, and instrumented the flow with lightweight tracking. Fullstory sessions confirmed the feature was being used as intended, which in this case was enough signal to be confident it was solving the problem without creating a heavy measurement setup. For the data sources which this new feature was implemented for, we saw an 80% decline in relevant support tickets.
Reflections
After reviewing user sessions in FullStory, it was clear that field listing and data previews aren't competing solutions — users flick between them. We decided to update the order to fields first and the data preview second, as this best suited the user's mental model.
A reusable preview framework live across three connectors and rolling out to the rest — turning a guess-and-check loop into a confident, one-shot connection.