# Performance forecaster
Vibe · 2026 · Lead Product Designer
Shows what a budget increase would deliver, and applies it in one click, wherever the advertiser already is.

Source : https://albancarmet.com/project/performance-forecaster

## Context

Budget increase recommendations were no longer applied. They were lost among other Dashboard cards unrelated to budget, with identical generic wording, and CSMs reported that a repeated 'increase your budget' with no reasoning did not build trust. The most frequent question in follow-up meetings was 'how much should I spend?'. Advertisers wanted a precise amount and what it would deliver.

## Process

There was no formal discovery: internal signals converged, and the projection model already existed on the data side, so we went straight into design and validated through prototypes. Three signals guided the decisions: interaction with Dashboard recommendations was low and declining, with Apply almost never clicked; 'how much should I spend?' was the top question in CSM follow-ups; and search and social platforms such as Google and Meta already show budget simulators with projected results, so advertisers expect that level of reasoning on CTV. The constraints: predict without over-promising (figures stay cautious estimates, modeled on recent performance), only strategies with enough delivery data can get a projection, and the recommendation had to stop being generic. I designed it with a PM and one engineer, in one month. AI was used for ideation and prototyping: building manipulable prototypes of the curve and slider, and generating variations of how to word and rank the gain, the gap to current budget and the recommendation. What we show or hide, the acceptable level of simplification, the tone of the alerts and the final wording stayed human.

## Solution

Performance projection shows what a budget increase would deliver, and applies it in one click, at four entry points, all in production. On the Dashboard, a Performance forecaster module highlights the best scaling opportunity with an estimated weekly gain, a curve with Current and Recommended markers, pagination across ready strategies, and Apply. In the Campaigns list, a 'Ready to scale' chip in the Performance column shows a mini projection on hover and opens the strategy. In campaign creation, an alert at the top follows the budget slider (below current budget, between current and recommended, at recommended, above it), and colored tooltips show the gap to the current budget. In the Agent, the existing 'Scale your results' card lets the advertiser approve or reject. Five decisions shaped it, all decided with the PM and the engineer: an integrated module rather than a separate 'Guided budget' mode; the alert at the top rather than under the slider; a tooltip that qualifies the impact rather than a bare number; keeping the Agent's existing card rather than rewording it (losing some wording consistency); and shipping the projection only, not the teasing for new campaigns, for development time (new campaigns get no projection and nothing announces one; a 7-day spend and impressions trend module fills the Dashboard when there is nothing to recommend).

## Results

Budget recommendations are now applied twice as often as other recommendation types within 14 days. But a short A/B test on a limited sample (mid-August to early September 2026) showed no measurable difference in apply rate with or without the forecaster (23% vs 25%, not significant), nor in the share of accounts raising a budget. About 15% of spending strategies have enough data for a budget recommendation. A first, provisional reading suggests the projection is optimistic: scaled strategies gain about half of what the curve announced; projected-vs-actual tracking is not in place yet. A CSM noted that many advertisers do not scale simply because they have no more budget. Open: make the projection accurate before making it more visible, and give new campaigns a reliability score that strengthens with data.