Consumer brands spend roughly $500 billion on trade globally every year, making it one of the largest controllable investments for many CPG manufacturers.
Yet a familiar problem persists: brands know some promotions work better than others, but often struggle to determine what actually drove the result, and what they should do differently next time.
Most teams assume the answer is more data or newer AI. It isn't.
Many brands are already drowning in POS files, syndicated data, retailer portals, competitive pricing data and internal reports. They don't necessarily lack data. They lack the ability to connect it, explain what matters, and turn it into a recommendation a commercial team can actually use.
That gap is where better decisions, and better returns on trade investment, get lost.
AI is rapidly becoming table stakes. The important question is no longer whether a tool uses machine learning. It is whether the system understands the commercial reality well enough to help someone make a better decision on Monday morning.
Many analytical solutions fall short for four reasons:
- They lack retail context. A mathematically optimal recommendation is useless if it doesn't survive contact with the realities of the category, retailer or shopper.
- They are difficult to explain. If commercial teams cannot understand why a recommendation was made, they are unlikely to trust it, and even less likely to defend it internally.
- They tell you what happened, but not why. Identifying a sales decline is easy. Determining whether it came from price, promotion, distribution, seasonality, competitive activity or some combination of those factors is much harder.
- They stop short of the decision. Sales and Revenue Growth Management teams don't need another dashboard. They need to know what action to take and why they should take it.
As Jeff Becker, General Partner at Antler, recently put it: "Stop pitching the technology. Start pitching the transformation."
For CPG brands, that transformation is straightforward to describe and difficult to execute: make better commercial decisions, faster, with enough evidence and transparency to act with confidence.
This is the problem Invisible Hand Technologies was built to solve.
We combine granular sales, pricing, promotion, availability and competitive data to help CPG brands move through three steps: understand what happened, identify why it happened, and determine what to do next.
Ingest. Start with the data commercial teams already have: store-level sales, pricing, promotions and distribution, and combine it with external signals such as competitive prices, promotions and out-of-stocks. The goal isn't simply to create a bigger data lake. It is to connect signals that usually sit in separate reports and systems.
Model. Predictive models estimate relationships such as price elasticity, promotional response and competitive effects, and allow teams to evaluate alternative pricing and promotion scenarios before committing budget. Instead of waiting until the end of a quarter to discover whether a decision worked, teams can compare potential outcomes before they act.
Act. The final output should not be another chart. It should be a recommendation: what happened, why it happened, which variables mattered, what action to consider next, and what impact that action is expected to have. Explainability matters because commercial decisions don't happen inside an algorithm. They happen in meetings, among Sales, Revenue Growth Management, Category, Finance and ultimately the retailer. A recommendation that a Category Manager can understand and defend in front of a buyer is far more valuable than a black box that simply produces a number.
Invisible Hand Technologies was built by people who have worked on both sides of this problem: commercial operators making decisions under pressure and technologists building the models behind them.
Max Pedró, CEO, brings experience across Walmart, McKinsey and consumer-facing businesses. Mitch Freeman, CPO, brings experience from Bain & Company and Wayfair, with a focus on translating analytical capabilities into tools people actually use. Fuchang Yin, CTO, brings deep expertise in data science and predictive modeling.
That combination matters.
The hardest part of commercial AI isn't producing another prediction. It is translating complex data into a decision that makes sense in the real world of pricing, promotions, category reviews and retailer negotiations.
The objective isn't simply to forecast more accurately.
It is to create a repeatable decision loop:
Discover what happened → diagnose why → design alternatives → decide based on expected impact → deploy → learn and repeat.
For a promotion, that might mean separating truly incremental sales from baseline demand, identifying the role of discount depth, competitive activity, seasonality and retailer execution, and then using those insights to design the next event.
For pricing, it might mean understanding elasticity and competitive response before recommending a change.
Over time, the value comes not from one prediction but from making each commercial decision with better information than the one before it.
The brands that get the most from AI will combine two things that rarely come together naturally: models that can understand increasingly complex market signals and people who understand how retail decisions actually get made.
AI provides the analytical power. Commercial context makes it useful. Explainability makes it trusted. And a clear recommendation makes it actionable.
The goal isn't another dashboard. It's knowing what to do next, and why.
