ON-DEMAND RECORDING
Agentic MMM in Action: The Future of Autonomous Media Planning and Buying
Most measurement tools focus on the past, leaving a gap between seeing data and taking action. This session bridges that gap by demonstrating how AI can finally close the loop.
In this 45-minute session, Sellforte CEO Juha Nuutinen and VP of Marketing Edward Ford walk through a single workflow that connects three things most teams keep separate: measuring the impact of past campaigns, planning future budgets against a causal model, and pushing those optimizations directly back into Google Ads and Meta.
Juha runs it live against a realistic ecommerce scenario.
The characters are a CMO and a Success Manager at a $200M ecommerce brand. It's time to review the "Peak Season" (November) performance. They need to know which channels actually drove incremental growth, how to pivot the budget for January 2026, and how to execute those optimizations across Meta and Google without manual entry.
What you'll see:
- Agentic MMM. The convergence of MMM’s predictive power with autonomous AI agents (Planner, Buyer, and Experiments) to automate the manual work of planning, forecasting, and buying.
- The MAGIC Framework. A breakdown of why autonomous buying must be Marginal, Actionable, Granular, Incremental, and Cross-platform to capture real-time demand.
- Change Intelligence. A live pre/post analysis of November performance, identifying top-performing channels like Meta and Direct Mail while uncovering specific underperforming "bottom" campaigns.
- Dynamic Scenarios. AI-driven forecasting for January 2026 that suggests budget shifts to maximize incremental ROAS (iROAS) based on predicted demand.
- Productivity & Safety. A deep dive into data security and a discussion on how Agentic MMM boosts analyst productivity by automating "grunt work" rather than replacing strategic roles.
Built for Heads of Performance Marketing, Marketing Analytics, and Growth at ecommerce and retail brands looking to take MMM out of the boardroom and into everyday execution.
00:00:00 – Introduction and Technical Setup
Edward: All right hi everyone, thanks so much for joining. We just hit the hour so I think we will just give people a couple moments still to find their way in. But welcome to Agentic MMM in action: the future of autonomous media planning and buying in real time. Pretty big topic, super excited to dig into this with you today. There is actually a lot to cover. We have 45 minutes, so it is going to be a lot of fun.
Juha: Hello everybody. My name is Juha. I am the CEO and one of the co-founders of Sellforte. I am also responsible for our product management. This is my proud moment to be able to share the crown jewel from our R&D kitchen. I have been really looking forward to this one.
00:01:28 – The Role of Sellforte
Edward: For those that do not know, we are a provider of MMM and incrementality specifically for ecommerce, DTC, and retail brands from all around the world. Sellforte helps marketers answer the questions you have been looking to get answers on: Which media channels drive the most sales? How should I allocate my media budget for the next four weeks? How effective are my promotions?
00:04:05 – Defining Agentic MMM
Edward: What is Agentic MMM? We have a very simple view. It is an advanced approach to marketing measurement that combines the predictive power of marketing mix modeling with autonomous AI agents. It is the combination of two superpowers. You have MMM measuring ROI and predicting revenue impact, and now you have AI agents helping us do things they are super good at: automating the planning, forecasting, and buying in real time. It is taking MMM out of the boardroom and into your everyday workflow.
00:06:19 – The Three AI Agents
Edward: We see Agentic MMM running upon three agents:
- The Media Planner Agent: Your super smart analyst who builds precise, hyper granular budget recommendations at the campaign and adset level.
- The Media Buyer Agent: Your 24/7 media buyer executing real time buying directly in platforms like Meta, Google, and TikTok.
- The Experiments Agent: Your ever curious data scientist automatically designing and analyzing experiments.
00:07:30 – The MAGIC Framework
Edward: To make autonomous media buying happen, you need a little bit of MAGIC:
- M is for Marginal: Looking at the marginal incremental return on ad spend. If you invest one more dollar, what do you get back?
- A is for Actionable: Translating recommendations into specific bid values.
- G is for Granular: Operating at the campaign and adset level.
- I is for Incremental: Understanding the true revenue impact.
- C is for Cross-platform: Integrating into all major advertising platforms.
00:11:15 – Live Demo: The CMO Scenario
Juha: All right, let me start beaming my screen. Edward, could you play the role of a CMO of a 200 million annual revenue ecommerce retailer? I am your Sellforte customer success manager. What would you like to ask?
Edward: Okay, a good starting point: how did last month go? How was November?
00:13:30 – Historical Performance Deep Dive
Juha: The AI already knows what the last month is. It is calling the MMM analysis agent now. For November 2025, you invested 1.9 million in media. Total revenue was 10 million, with an incremental ROI of 5.3. Your top channels were Meta and Google Ads. Surprisingly, Direct Mail also produced a high ROI.
Edward: How does this compare to previous years?
Juha: We invested 27 percent less this year but achieved the same level of incremental sales. Efficiency in Meta ads doubled.
00:19:00 – Tactical Recommendations
Edward: Thinking ahead to November 2026, what should I do differently?
Juha: It is identifying top and bottom campaigns. For example, one specific Meta campaign had an ROI of 15.0, so it recommends scaling that. Meanwhile, it found Google PMax campaigns that were underperforming compared to our target, so it recommends an increase in target ROAS to force better efficiency.
00:23:45 – Forecasting for January
Edward: Can you optimize my budget for January?
Juha: It is identifying that this request is about optimization. Based on the current plan, it wants you to spend more on Meta Advantage plus and Google YouTube. I would be a bit careful on the Direct Mail recommendation here, as a great November doesn't always mean a great January in fashion. This is where the human still sense checks the AI.
00:28:45 – Handling Audience Questions
Edward: We have a question: is this based on mIROAS?
Juha: Currently, the planning module provides incremental ROAS. Adding marginal ROI specifically to this table view is a great suggestion that our developers will note down.
00:31:00 – Executing Media Buying in Real Time
Edward: Let us do the deep end. Which campaigns should I focus on for the last 30 days?
Juha: It is looking at the live target ROAS and budget values. It recommends decreasing target ROAS for YouTube USA to increase spend, while increasing target ROAS for Search Brand to pull back.
Edward: I like those. Please execute your recommendations.
Juha: For Meta, it is done. For Google, it is done. We are maybe two or three months away from doing the real change in the actual live platforms in the demo environment, but this is how it must feel in the future. It communicates the changes it has done, why it did them, and proves the contribution to the target KPI.
00:41:00 – The Future of Marketing Roles
Edward: Does this replace junior analysts?
Juha: It gives you tools to be more productive. It makes insights available even for the CMO without needing a learning course on specific dashboards. Analysis that used to take many hours now takes 10 to 15 minutes. It automates the manual tasks but leaves the strategy to you.
00:46:50 – Closing Remarks
Edward: We are coming up to time. We had so many questions in the Q&A. We will follow up with an email containing the slides, recording, and the answered questions. If you want a deeper dive, scan the QR code. All Sellforte signups by the end of the year are bundled with Sellforte AI.
Juha: I hope the audience had enough surprising moments. I hope we did not embarrass ourselves too badly.