The Challenges of Driving CPG Demand
The Problem: Spend That's Hard to Measure
Marketing and Promotion investments are among the highest outflows in a CPG P&L. Whether looking at Consumer Marketing or Trade/Customer Marketing, these investments are among the most difficult to measure for ROI.
The AI Noise Problem
Leaders of Marketing/Sales teams are bombarded with messages from vendors who seem to promise magical results with no human intervention — the magic of AI! To add insult to injury, these messages also reach the CFO and the CEO, resulting in pressure to reduce the size of marketing teams and further compounding the difficulty in driving demand.
The Marketer would do well to step back in this situation, tune out the noise, and go to first principles. There are different parts of the problem that need to be solved by different tools.
Two Different Problems, Two Different Tools
The lower-hanging fruit — with correspondingly lower potential returns — is to drive efficiency using AI, so the team's output can be improved in both quality and quantity. LLM-based AI tools can play an important role here, helping the team increase their creative output and put together versions that can be objectively tested. With appropriate human oversight, these tools can help with the creation of consumer- or customer-facing content. Proper feedback loops can ensure the team increases their trust in these tools with increasing experience.
The bigger opportunity, though, is in determining the optimum marketing mix. There are several hurdles to overcome here.
Why Marketing Mix Optimization Is Still Broken
Comfort with history. There is a strong tendency to compare investment decisions with prior periods, which brings with it a discomfort with drastic change. As a result, investment "holy cows" are never challenged — a potential drag on overall ROI.
Watertight budgets. Investments to drive demand are controlled by different teams and sit in different budgets. Consumer marketing and sales/trade marketing teams zealously defend their own, locking investment into watertight compartments and limiting how far ROI can be optimized by shifting spend across them.
Limitations of traditional tools. The ideal tool for understanding the impact of different elements of the marketing mix is a marketing mix model (MMM). But as historically implemented, MMMs take months to complete — producing insights that are out of date by the time they arrive. They typically look only backward, with no predictive component, and they're expensive enough to eat into the very budgets they're meant to optimize.
Limitations of mindset. Depending on a category's growth stage and competitiveness, optimizing the marketing mix isn't enough on its own. Sometimes winning means recognizing when something is really working and going all-in on that rare opportunity — difficult to do when insight arrives six months after the fact.
Consider a brand that discovers, mid-quarter, that a regional promotion is outperforming forecast by 3x. Under a traditional MMM with a six-month reporting lag, that signal arrives long after the window to act on it has closed.
What's Actually Needed
It's important for CPG leaders to recognize that this is not a problem LLM-based solutions will solve. There are strong signals already present in a company's own data — promotional lift, price elasticity, channel response curves — that can guide decisions going forward. What's needed is a system that delivers the functionality of an MMM without its usual shortcomings.
A decision system like this would respond in real time to changes in the market, letting leaders objectively optimize their marketing mix to maximize revenue, margin, or any other metric of their choosing — all while operating within strategic guardrails that remain 100% under human control.
This is the right kind of AI tool for the job: one that operates strictly according to your rules and under your control, while effectively challenging your thinking.
This Is What EsperPulse Builds
EsperPulse's approach replaces the slow, backward-looking batch cycle of traditional MMM with real-time signal detection — surfacing what's working while there's still time to act on it, without the multi-month lag or the price tag.
If this sounds interesting, contact EsperPulse. A quick pilot is the best way to see the value of the approach.