Latest dispatch How we ran a $40M pricing study in eight weeks — and what the last consultancy got wrong
Commercial
Intelligence Agency
Request briefing
Our discipline

Reasoning systems,
built for commercial decisions.

Agentic AI. Causal inference. Bayesian methods. We do the math that pricing committees, S&OP, and trade-spend leaders have to defend — faster than your last consultancy thought possible.

The market gap

Most enterprises are buying advice when they need answers.

Strategy decks don't tell you whether to drop the price 4%. Quarterly dashboards don't tell you which retailer drove the variance. A consultancy readout doesn't quantify what a $2M trade investment actually returns. We do that work — and we defend the math.

$0 $1M $2M Q1 Q2 Q3 Q4 Same quarter. Same insight. Different outcome. Advice Math Advice held flat. Math returned +$1.84M.
Our services

What we model.

Six commercial-decision practices. Each one produces real numbers a real operator has to defend in a real meeting. Elasticity coefficients. Forecast intervals. Causal lift estimates. Optimization frontiers. If we can't put a number on it — and stand behind that number — we don't ship it.

S-016–10 weeks

Pricing intelligence

Price elasticity forecasting. Promotion lift quantified causally. Competitive monitoring on autopilot.

S-028–14 weeks

Marketing intelligence

Modern marketing mix. Causal lift by channel. Synthetic controls in place of fragile last-touch.

S-0310–16 weeks

Demand intelligence

Sell-in and sell-out forecasting, wired into planning. Variance monitored. Demand scenarios simulated.

S-048–12 weeks

Customer intelligence

Probabilistic lifetime value. Churn calibrated to retention spend. Segment unit economics.

S-056–10 weeks

Trade-spend intelligence

Forensic accounting of trade investment. Causal lift by program and retailer.

S-0610–18 weeks

Field intelligence

Earlier detection of velocity shifts, distribution gaps, and execution inconsistencies. Signals before the lagging KPI catches them.

Our methods

How we reason.

Four methodological commitments. Modern commercial decisions don't yield to A/B tests and dashboard math — they require causal inference, hierarchical priors, and the kind of statistical craft most consultancies have stopped doing. The techniques below are what your last vendor will tell you are "too academic" for your use case. They are why our numbers hold up.

M-01
Reasoning agents
Custom-tuned LLM systems — QLoRA, SFT against your domain — that ingest, synthesize, and reason over heterogeneous commercial data at scale humans cannot match.
M-02
X Y
Causal modeling
Beyond correlation. Synthetic controls, difference-in-differences, instrumental variables. The things that give you the answers you need when A/B testing isn't possible.
M-03
Bayesian inference
For small samples, strong priors, and decisions under uncertainty. We quantify what you don't know.
M-04
Operational handoff
Models that live past the engagement. Documented assumptions, defensible re-fit cadence, and clean handoffs to the operators (or platforms) that run them in production. We work with FieldGoal where deeper system integration is needed.
The approach

How we engage.

Four phases. It starts with a conversation, not a model. We translate your business expertise into math that actually understands cause and effect — and we stay engaged when the world changes.

01 / Conversation

Build the foundation.

We work with you to build a custom, bespoke model of how your business works — not a generic, black-box machine learning system. It starts with a conversation.

02 / Modeling

Build the model.

We translate your business expertise into mathematical models that actually understand cause and effect. They don't pattern-match. They don't chase correlations. They reason.

03 / Decision

Decide, don't report.

Stop building dashboards. Start building decisions. Static narratives become conversational exploration. Manual analysis becomes AI-assisted decision support.

04 / Support

Continuous support.

When the world changes, we don't leave you hanging. Models drift. Markets move. Distributions shift. We stay engaged so the math stays operational.

Our team

The operators.

Six operators from FieldGoal, our sibling company — the platform side of commercial execution. Data science, statistics, AI engineering, intelligent design, systems architecture, commercial operations. No analysts in pre-MBA training. No partner-and-pyramid model. Every engagement runs through a senior with their name on the work.

Vahag Dudukgian

Founding Partner

15+ years in commercial systems for field-sales operations. Ex-Rockstar Energy (operations through the $3.85B PepsiCo acquisition). Early career at Raytheon. Founder & CEO of FieldGoal.

Patrick Bjornstad

Founding Partner

Data scientist with 10+ years across statistics and applied software development. Previously at NASA JPL. Leads LLM and agentic AI integration at FieldGoal; brings the same stack to our modeling work.

Spencer Boucher

Principal · Quantitative Methods

Quantitative methods and applied statistics. Previously a data scientist at Uber. Causal estimation, experimental design, and the model-to-decision handoff that most analytics work never makes.

Axelle Sarray

Senior · Decision Interfaces

10+ years designing AI interfaces for enterprise teams. Builds the layer where analytical output becomes operator action — the difference between a dashboard and a decision.

Yury Lysov

Senior · Systems Architecture

15+ years architecting production applications. Built FieldGoal's frontend foundation and AI experience layer. Bridges the modeling work and the systems it has to run inside.

Adeel Cheema

Senior · Field Systems

Software engineer combining React Native and AI for cross-platform applications. Builds the field execution layer where forecasts and pricing decisions get tested against reality.

Our thoughts

Dispatches.

Methodology, case studies, and the occasional rant. What most consultancies wouldn't publish — the work, the math, and where the industry is wrong. For readers who want the math.

Working session

Bring us your hardest
commercial questions.

Sixty-minute working session with a partner. We'll sketch the analytical approach, the math, and the data we'd need. No deck. No phased engagement plan. If we can't help, we'll tell you who can.