How One Ad Agency Cut Media Plans Days to Minutes
An ad agency's strategy team bottled Binet & Field into a GPT-4o tool that splits brand vs. performance budgets — plus a voice CFO that lets marketers rehearse the pitch live.
Minutes not Days
For CFO-ready media budgets
Allister Hercus
Founder / Consultant
Stoop Studios
Connect ↓
The Challenge
Noble People, an advertising agency, faced a persistent challenge that CMOs across the industry share: determining the right budget split between brand-building and performance media — and then convincingly pitching that allocation to their CFO. Without a rigorous, evidence-backed rationale, brand marketing budgets were routinely undercut because they lack a direct correlation to sales. Marketers were spending excessive time preparing budget presentations, often from a blank page, with no structured framework to defend their decisions.
What They Built
Stoop Studios built Budget Balancer, an interactive AI-powered web tool that calculates the optimal brand vs. performance media split based on user inputs, grounded in the academic frameworks of Les Binet and Peter Field. An OpenAI GPT-4o Assistant synthesizes user data against those frameworks to generate a personalized budget recommendation with a written rationale. A role-play button then connects the user to a voice-enabled virtual CFO — powered by OpenAI's Realtime API — allowing marketers to practice defending their budget in a realistic conversation before the actual meeting. Built in approximately three to four weeks, deployed on Vercel.
Stoop Studios began by encoding the Binet and Field academic frameworks as structured knowledge within the GPT-4o Assistant context — giving the tool a principled foundation rather than relying on general AI reasoning about budgets. Users input details about their brand, category, and goals; the assistant synthesizes those inputs against the frameworks and produces a personalized brand vs. performance split recommendation, complete with a written rationale designed to hold up in a CFO conversation. The second stage addresses execution of that recommendation: a role-play mode connects the user to a voice-enabled virtual CFO powered by OpenAI's Realtime API. The CFO pushes back, asks hard questions, and lets the marketer stress-test their argument before the real meeting. The front end was built in Next.js and deployed on Vercel. Total build time was approximately three to four weeks. The result was a tool that compresses days of manual budget preparation to minutes — and converts a typically anxious budget defense experience into structured practice.
AI Role
An OpenAI GPT-4o Assistant analyzes user-provided campaign inputs against the academic frameworks of Binet and Field to generate a personalized brand vs. performance media budget split with a written rationale. A second AI layer — OpenAI's Realtime API — powers a voice-enabled virtual CFO simulation that engages in live conversation to help marketers stress-test their budget pitch before the actual meeting.
Infrastructure
Next.js (front-end framework) • OpenAI GPT-4o Assistants API (budget recommendation engine) • OpenAI Realtime API (voice-enabled virtual CFO) • Vercel (deployment)
Integration Points
GPT-4o Assistant configured with Binet and Field framework knowledge for budget synthesis • User input form connected to Assistant API for personalized recommendation generation • Realtime API voice session triggered from recommendation output for CFO role-play • Next.js front end deployed on Vercel with API route handling for both OpenAI endpoints
Impact
Academically Grounded Budget Recommendations in Minutes
Marketers who previously faced a blank page now receive a personalized brand vs. performance split — backed by Binet and Field's peer-reviewed frameworks — in minutes rather than days.
Real-Time CFO Roleplay Preparation
The voice-enabled CFO simulation gives marketers a live practice environment to stress-test and refine their pitch — a capability previously impossible to replicate without live role-play partners or coaching.
Significant Reduction in Presentation Prep Time
Allister designed the tool to compress work that would otherwise span multiple days of research and slide-building into a running-start session, getting marketers off a blank page fast.
Implementation Complexity
The tool required integrating OpenAI's GPT-4o Assistants API and the Realtime API with a Next.js front end deployed on Vercel, encoding the Binet and Field frameworks as structured knowledge within the assistant context. This goes beyond off-the-shelf tooling but is achievable in 3–4 weeks of custom development without complex backend infrastructure.
Best Fit For
Best for advertising, media, and creative agencies — and brand-side marketing teams — that need to justify brand marketing investments to finance leadership but lack a structured, evidence-backed process for determining and presenting budget allocations.
