Media forecasting
An executive simulation tool that transforms media spend into forecast insights with an intuitive interface
Intro
One of the main planning pain points for executive users was the lack of a simple way to evaluate budget changes before applying them live. To estimate the impact, they had to rely on manual analysis, scattered reports, or real-world budget testing — a costly and time-consuming process that slowed planning down.
We designed a forecasting feature that lets users simulate different budget scenarios, compare projected outcomes, and understand the expected performance in seconds.
This project was created as part of my work on a large B2B software platform. The work presented here reflects my personal contribution and perspective as a designer.
Challenge
The challenge was to turn a complex forecasting model into a planning experience that felt simple enough for everyday use, while still giving users enough confidence to make budget decisions before launching changes live.
From a product perspective, the feature had to help executive users evaluate different spend scenarios, understand the projected impact on performance, and reduce reliance on manual analysis or costly real-world testing. That meant the experience needed to feel fast, intuitive, and actionable, without oversimplifying the underlying model.
At the same time, the design had to communicate the model’s limitations clearly. Users needed to understand not only the forecast itself, but also when the projection was reliable, when accuracy dropped, and when a result could no longer be calculated.
Also, there were no ready-made components in the design system, so the entire experience had to be designed from scratch while still fitting the overall platform’s visual language.
My role
I worked on this feature across 2023–2024, leading its design from early exploration through final UI delivery. I worked closely with product managers, engineers, and cross-functional stakeholders to shape the experience, define its core interactions, and turn a complex forecasting model into something that felt clear, intuitive, and usable.
My responsibilities included defining the overall structure, designing the main simulation flow, shaping the interaction logic across different states, and refining the experience to handle edge cases without adding unnecessary complexity to the UI.
The layout
Before any forecasting could happen, users first had to select a media plan as the starting point for the forecast, and then apply the relevant filters to view the campaign groups.
Since media plans could contain dozens of groups, we kept all of them collapsed by default except the first one. To preserve visibility without overwhelming the UI, we still surfaced key values in the collapsed state, but without showing the forecast curve. This helped users stay focused while allowing them to expand specific groups for deeper exploration only when needed.
To support both group-level and overall decision-making, we introduced a sticky footer that showed the total simulated impact compared with the planned one across all groups. The values in the footer were aligned with the values shown in each group, making it easier to compare local changes with the overall plan.
Making forecasts visual
At the core of the experience was an interactive forecast curve that visualized the relationship between spend and projected revenue.
Once the forecast was generated, the curve immediately displayed the planned budget point, allowing users to see whether it was positioned below or above the optimal point. To make that signal easier to notice, we introduced supporting badges that highlighted potential underspend or overspend
To support both precision and speed, we combined direct input fields with a draggable slider. Users could either enter exact values manually or explore scenarios more freely by moving along the curve and seeing the projected results update in real time. This made the simulation feel immediate, flexible, and easy to use during planning.
Monthly breakdown
The forecast also needed to support a monthly breakdown, since media plans were already structured by month and could span a full yearly view. To make the feature feel consistent with that planning model, the simulation had to reflect the same level of detail.
Bringing a full monthly view into the forecasting experience required careful design decisions. It risked overwhelming the interface, adding visual noise, and pulling attention away from the simulation itself, so we embedded the table inside each media group and kept it collapsed by default.
Each simulated cell had to remain both visible and editable. The column widths were designed to fit a full year into a single view for most users, while inner horizontal scroll helped preserve usability on smaller resolutions.
Edge cases
The experience also had to communicate the forecast’s boundaries clearly, so users could understand not only the projected result, but also how much confidence they should place in it. To support that, we introduced two states directly on the curve.
Out of range indicated that the selected budget was outside the model’s supported range, meaning a forecast could no longer be generated.
Outside historical data indicated that the forecast was still available, but based on extrapolation beyond past spend data, which made the result less reliable.
By visualizing both states directly on the curve and reinforcing them with tooltips and warnings, we helped users understand whether the forecast was available, less reliable, or no longer supported.
Bulk simulation
Bulk simulation was designed to help users apply budget changes across multiple groups in a faster and more controlled way. The flow supported four actions: setting a new budget, reallocating the current budget, increasing it by a selected value, or decreasing it by a selected value.
To reduce errors and make the process easier to follow, the interaction was structured as a guided sequence: users first selected the action, then chose the relevant groups, reviewed the supported forecast range, entered the value, and previewed the result before applying it.
After the change was applied, each affected input was marked with a “Simulated in bulk” indicator, making it easier to understand which groups had been updated through the bulk action.
Outcome
This feature helped users spot potential overspending or underspending before making live budget changes, simulate different budget scenarios, and see the projected impact in seconds. As a result, planning became faster, more informed, and less dependent on guesswork.
From a product perspective, the feature helped make budget simulations a more practical part of the planning workflow.
From a design perspective, this was especially interesting because the feature had no clear references. It was a chance to shape the experience from the ground up around both user needs and the logic of the forecasting model.
