How AI marketing platform redesign helped Fieldstream secure $3,8M in pre-seed funding

Based on the UX audit, we redesigned the dashboard as a system-level layer of the product by developing a unified design system to minimize visual clutter and make the screen predictable to scan.

Period and dependent-variable controls sit between the two layers, so a single filter change propagates across both, and nothing has to be reset when the question shifts.

The dashboard now works as a decision-making tool. Summary metrics illustrate how the business is performing, the tabs explain why, and the recommendations surface what to do next.

We rebuilt the logic for working with marketing data so the relationships between controls and the charts they affect are visible at a glance. Filters and Grouping were consolidated into a panel at the top of the screen, which means the marketer immediately sees which settings shape every visualization on the page. 

We removed hidden dependencies between blocks to enforce one grouping principle across the screen. Every chart now responds to the same input, lowering both the cost of a wrong read and the cost of explaining the dashboard to a new team member.

The Effect Uplift block was rebuilt as a metric-aware analytical surface:

1. The block was split into separate tabs to allow the user to focus on one metric at a time instead of decoding several layers stacked into a single chart.

2. An Uplift group selector narrows the view into a channel family like Social Media, and an inline explainer beneath each chart demonstrates the analytical interpretation that used to require a call with an analyst.

We structured planning as a sequential workflow with three named stages (Create planning → Add constraint → Choose the plan) visible at the top of the screen. The complex logic of the underlying model becomes a workflow the marketer can follow without prior modeling knowledge.

The planning flow now moves through three controlled stages:

1. Create planning asks only for what the model needs to start. The user defines what to optimize for without first learning how the model works.‍

2. Add constraint turns business rules into structured rows. All rules are entered in the same format and can be added or removed individually, which keeps even multi-channel, multi-period scenarios legible.‍

3. Choose the plan presents the model's options side by side, letting the marketer compare alternatives before committing. 

A "Create custom plan" path lets the user fork from any system-generated option and reallocate the budget manually, with the projected effect updating against every move. Planning is now scenario selection, with each option backed by a probability band and a per-channel allocation the marketer can defend to a CFO.