Agentic MMM in Action: The Future of Autonomous Media Planning and Buying – in Real Time is now available on demand.
In this 45-minute live session, we showed how Sellforte AI plans, optimizes, and buys media in real time, using incrementality-based measurement and agentic decision-making.
You’ll see how conversational MMM works in practice, how miROAS guides every optimization decision, and how we’re moving toward fully autonomous media optimization.
Agentic MMM in Action: Watch the full webinar recording
Live AMA: Full Q&A from the Webinar
Below is the complete Q&A from the live session, covering incrementality, optimization, AI reliability, experimentation, and data security.
How do you define “incremental”? Is it different from paid media contribution?
There are two main ways to measure how paid media contributes to your sales.
Incrementality-based measurement:
This refers to the concept of measuring the true incremental revenue lift generated by a specific marketing channel, campaign, or ad set. Incremental revenue is the amount of sales you would lose in the short term if you stopped the marketing activity. Measuring the true effectiveness of a campaign requires understanding its true incremental sales impact.
Attribution-based measurement:
Attribution uses pre-defined rules to assign credit to channels and campaigns. For example, last-click attribution assigns all credit to the last touchpoint before the purchase. Attribution does not take incrementality into account, meaning that attributed ROAS can be significantly higher or lower than incrementality-based ROAS.
Are Sellforte models at ad set or campaign level? How are you getting ROI by campaign or ad set? Is this platform ROAS or business ROAS?
Sellforte measures Incremental ROAS (iROAS) at:
-
Campaign level for Google
-
Ad set level for Meta
This ensures Sellforte supports marketing decisions at the level where they are actually made.
iROAS is based on actual incremental revenue generated by a campaign or ad set — not revenue attributed by an advertising platform. iROAS is estimated using Sellforte’s analytics, combining Marketing Mix Modeling, experiments, and attribution data.
You can also measure Incremental ROAS and optimize at a higher summary level, such as:
-
Channel or tactic level (Meta Advantage+, Meta Prospecting, Meta Awareness)
-
Ad platform level (Meta, Google, TikTok)
What feeds historical performance into the model? Do you run daily or weekly MMM?
Sellforte models run daily (in most cases) to stay responsive to the fast-changing digital advertising environment.
Historical performance is measured using Sellforte’s analytics, which combine Marketing Mix Modeling, experiments, and attribution data. Data inputs include:
-
Paid media data (spend, impressions, clicks) across digital advertising platforms
-
Offline media data (where applicable)
-
Sales and revenue data
-
Promotions, pricing, and seasonality signals
-
External and contextual factors, where relevant
Does future planning take into account seasonality, spend curves, and fixed budgets?
Yes. Sellforte has the most comprehensive planning tool in the market.
Some features include:
-
Seasonality patterns by default, enabling realistic weekly allocation recommendations
-
Response curves and saturation effects for each channel
-
Flexible budget constraints
-
Optimization for either a fixed budget or targeted sales growth
How does the model balance awareness and conversion channels?
First, the model measures the actual incremental revenue impact of each channel, ensuring that awareness channels — which may receive little credit with traditional attribution — are measured properly. This ensures each channel gets a fair spend allocation in default optimization scenarios.
Second, users can configure optimization scenarios by funnel stage. For example, if the objective is to build awareness, you can fix the total budget for top-of-funnel channels and let the optimizer allocate spend within that budget, while optimizing mid- and bottom-funnel channels separately.
How do you account for rising media costs?
Rising media costs are captured directly in the data. Because Sellforte runs models daily and uses response curves, it detects:
-
Diminishing returns
-
Cost inflation effects
-
Shifts in efficiency
Recommendations adapt automatically as economics change.
Does Sellforte provide visualizations?
Yes. Sellforte provides flexible dashboards for:
-
Historical effectiveness analysis
-
Forecast charts
-
Response curves
-
Scenario comparisons
-
Budget allocation visuals
Sellforte AI complements these by explaining and navigating insights conversationally. In the coming months, charts will also be available directly within AI responses.
Is the model reliable for short-term optimization?
Yes. Sellforte is designed for short-term tactical decisions as well as quarterly and annual planning. The accuracy of future-looking forecasts depends on factors such as data quality, historical variance, business dynamics, and unexpected events.
Can the AI propose and rank incrementality tests to improve the model?
Yes. Sellforte’s Experiments Agent, launching in Q1, can:
-
Detect natural experiments
-
Propose and rank tests
-
Recommend where testing will most improve model confidence
This capability forms the foundation of the Experiments Agent in Agentic MMM.
Are budget recommendations based on marginal ROI or ROI?
Recommendations are based on miROAS (Marginal Incremental ROAS or Marginal ROI), not average ROI.
This is critical because average past effectiveness does not answer the question:
“Where will the next dollar have the greatest impact?”
Do you provide accuracy measures like MAPE?
Yes. Sellforte provides model diagnostics, including MAPE and other validation metrics, ensuring transparency and trust in recommendations.
Do you give recommendations for offline media like TV, radio, and OOH?
Yes. Sellforte supports many large retail customers with significant offline media budgets, including TV, radio, OOH, and print. Sellforte measures their effectiveness and provides optimization recommendations.
Can this work with multi-channel campaigns? Do you ingest offline data from third parties?
Yes. Sellforte measurement covers all channels, including:
-
Digital media
-
Offline media
-
Owned media (email, SMS)
Sellforte also provides flexible data integrations and can ingest offline data from third parties.
Does Sellforte AI replace junior or entry-level data analysts?
No. Sellforte AI amplifies productivity rather than replacing roles. Tasks that once took hours can now be completed in minutes, allowing analysts and marketers to focus on strategy, experimentation, and growth.
Which LLM and agentic framework do you use?
The current version of Sellforte AI uses Anthropic models via AWS Bedrock. The agentic framework is Strands.
How do you ensure data safety if LLMs are involved?
Sellforte takes data security very seriously. The platform is regularly audited, and product development has followed strict enterprise-grade IT security policies from day one.
As a core security principle, each customer’s data is isolated in a dedicated cloud environment (typically AWS) within a specific geography. When using an LLM, the model is deployed and operated within that same environment, ensuring data is never sent to external servers. Only LLMs that can run in isolated environments are used.
Author
Edward Ford is Marketing Director at Sellforte. He has over 15 years of marketing experience in B2B SaaS and Tech with specialization in marketing measurement and intelligence. Before joining Sellforte, Edward spent over 6 years at Supermetrics where he joined as an early-stage employee. Follow Edward on LinkedIn for more about marketing and growth.