Improving trust and clarity in forecasting for CTV ad publishers

TL;DR

Forecasting in Publica by IAS was a critical, but unreliable experience for publishers planning future ad inventory. Inputs were fragmented, outputs were hard to interpret, and users lacked confidence in the results.

I led the redesign of the forecasting experience into a centralized, transparent tool that unified inputs, clarified outputs, and improved trust in planning decisions.

Impact

  • Centralized forecasting inputs into a single, coherent workflow

  • Improved clarity and trust in forecast outputs

  • Increased usability for publishers planning quarterly revenue and campaigns

  • Enabled more confident decision-making in high-stakes planning workflows

Context

Publica by IAS is a CTV ad server platform that helps publishers manage ad delivery and optimize streaming revenue through data-driven insights.

Forecasting plays a key role in this ecosystem, it helps publishers estimate future inventory and revenue before committing to campaigns.

However, the existing experience was fragmented and difficult to trust.

The problem

Forecasting required users to:

  • Navigate multiple disconnected inputs across the product

  • Make decisions without full visibility into what influenced results

  • Interpret outputs that lacked transparency or clear structure

As a result:

  • Users struggled to understand what was included in forecasts

  • Confidence in forecast accuracy was low

  • Planning decisions were delayed or second-guessed

A key concern repeatedly surfaced in user feedback:

“We don’t fully understand what’s included in the forecast, especially audience segments.”

Why this mattered

Forecasting is not just a UI problem, it directly impacts:

  • revenue planning

  • campaign commitments

  • publisher confidence in the platform

When forecasts are unclear or untrusted, users stop relying on them altogether.

My role

Lead Product Designer, Publica by IAS

I owned the forecasting redesign end-to-end, including:

  • research and synthesis

  • UX strategy and information architecture

  • design execution and iteration

  • cross-functional alignment with product and engineering

Strategic Approach

Rather than treating this as a UI redesign, I reframed it as a trust problem in a decision-making system.

The goal was not just better usability—it was better confidence in outcomes.

Key decisions

1. Centralize forecasting into a single workflow

Instead of distributing inputs across multiple surfaces, we consolidated all key parameters into one coherent experience.

2. Make uncertainty visible, not hidden

Rather than abstract outputs, we prioritized clarity around what influences forecasts and what is excluded.

3. Expand input coverage to match real decision-making

We added missing targeting parameters (e.g. audience segments, frequency capping) to align forecasts with how users actually plan campaigns.

4. Structure outputs for interpretation, not just display

Forecast results were redesigned as scannable, comparable insights not raw values.

5. Align product + engineering early on feasibility

We defined a clear V1/V2 split to balance ideal experience with implementation constraints.

Research & Insights

Through interviews and analysis, I identified:

  • Publishers need to understand reach before committing spend

  • Forecast outputs lacked explainability and context

  • Missing input parameters created a gap between expectation and reality

  • Users were actively cross-checking forecasts elsewhere due to low trust

These insights directly shaped both UX structure and product decisions.

Solution

We designed a unified forecasting experience with:

1. Centralized inputs

All targeting parameters (dates, inventory, audience, constraints) in one structured workflow.

2. Transparent outputs

Forecast results presented with clear breakdowns and supporting context.

3. Interpretable data views

Flexible breakdowns (e.g. by channel, tier) to support different planning needs.

4. Clear system feedback

Loading and processing states designed to set expectations and reinforce reliability.

Design Evolution

  • Wireframes focused on restructuring information hierarchy

  • Iterations refined balance between flexibility and clarity

  • Component exploration ensured consistency with existing system

  • Engineering collaboration validated feasibility and scope boundaries

Tradeoffs

  • Completeness vs clarity
    Not all variables were surfaced equally, prioritized decision-relevant inputs

  • Flexibility vs structure
    Too much flexibility reduced trust; too much structure reduced usability

  • Speed vs accuracy
    Chose interpretability over overly complex modeling exposure

Outcome

The redesigned forecasting tool:

  • Provided a more coherent and centralized experience

  • Improved clarity around what drives forecast results

  • Increased user confidence in planning decisions

  • Reduced ambiguity in interpreting output data

Product impact

Beyond the UI, this work:

  • Shifted forecasting from fragmented inputs → unified decision workflow

  • Improved alignment between product assumptions and user mental models

  • Strengthened forecasting as a core planning tool within the platform

Loading State

Populated

< Design System

What I’d improve next

  • Introduce explainability layers for forecast drivers

  • Improve scenario comparison (what-if planning)

  • Expand transparency around confidence and variability ranges

Auctions Insights >