Profit Matrix
Manufacturing & Industrial, Accounting, Finance, Construction
Brian Albans partnered with us to grow Profit Matrix from a half-built idea into a production pricing-intelligence platform he can confidently sell and keep scaling - and today that growth shows up on both sides of the business. Brian now has the technical partner he needed to turn Profit Matrix into a real, sellable product instead of a side project stuck in his own backlog, while the 14 manufacturing and distribution companies now running their pricing and margin decisions through Profit Matrix have recovered an estimated $2.3M in margin that would otherwise have gone unnoticed, with every one of them renewing and already asking for the next capability.
Client Information
“RMM IT Consulting did a great job understanding our requirements, quickly developing code, and showing flexibility with working in a start-up environment.”
What Is Profit Matrix
Manufacturers and distributors lose an estimated $20B a year to hidden profit leaks - weak pricing, excessive discounting, cost overruns, and unprofitable customers hiding behind high revenue. Most don't have a dedicated pricing team; they have spreadsheets, gut instinct, and data scattered across ERPs, CRMs, and accounting platforms.
Profit Matrix set out to be the "virtual pricing manager" these businesses couldn't otherwise afford - but it needed to move from basic dashboards to something a finance or sales leader could actually trust and act on.
A Founder Who Needed a Partner, Not Just a Developer
Brian Albans had already built the first version of Profit Matrix with another developer, enough to prove the idea had legs. But that developer's attention got pulled onto other work, and Brian was left needing someone who could pick up a half-built, ambitious product and keep it moving - not from a spec, but from actually understanding why it existed in the first place.
Profit Matrix was never going to be a simple build. What Brian needed wasn't a developer who just takes tickets and writes code - he needed a partner who could understand his business and his vision quickly, who genuinely cared about what the product was trying to become, and who would stay with him through the whole journey: development, marketing, ongoing maintenance, and eventually turning it into something he could sell.
Understanding the Business Before Touching the Code
The first weeks weren't about shipping features - they were about understanding. We spent hours in meetings with Brian learning how pricing managers actually think, what a sales leader needs to trust a number enough to act on it, and where the existing codebase already had the right instincts versus where it was improvising. That groundwork mattered more than any single feature, because it's what let every decision after it move fast without guessing.
Once we understood the app, we didn't try to rebuild everything at once. We identified the core of what made Profit Matrix valuable - the ability to explain why margin moved, not just that it did - and focused there first, before expanding outward into the modules that built on top of it.
Along the way we made a habit of choosing boring-but-reliable over impressive-but-fragile: batch-scoring models into the existing database instead of standing up live prediction endpoints, absolute business thresholds instead of shifting statistical segments, a direct tool-calling loop instead of a black-box AI agent framework. Every one of those calls traded a flashier demo for something a small team could operate, explain, and trust in production - exactly what a founder selling into finance and sales leaders needed underneath the product.
What We Built
Over the course of the partnership, that focus turned into eight connected modules - each one solving a specific problem Brian's customers were feeling, not a feature for its own sake.
1. Data Integration: One Coherent Model of the Business
None of the analysis works without clean, current data, and Profit Matrix's customers run on a patchwork of ERPs, CRMs, and accounting platforms that were never built to talk to each other. We built the ETL pipelines and the underlying database design (PostgreSQL on AWS Aurora) that pull those systems into one coherent model of the business - the foundation every other module reads from.
That foundation is also multi-tenant by design: every customer's data is logically isolated from every other's, so the same infrastructure can serve all of Profit Matrix's customers securely without any of them ever seeing another's numbers.
2. Reporting: YoY Comparisons That Actually Mean Something
Knowing where you stand this month means little without knowing whether you're actually improving. We built year-over-year reporting across the metrics leadership already watches - net revenue, PPM, and the rest - so a number isn't just a snapshot, it's a trend line Brian's customers can hold themselves accountable to.
3. PVM Analysis: Seeing Exactly Why Margin Moved
A report that just says "margin is down 2% this quarter" doesn't tell a finance or sales leader what to do next - the right response is completely different depending on whether volume dropped, prices slipped, costs rose, or the product mix shifted toward lower-margin items. We built a price-volume-mix (PVM) analysis engine that bridges margin change period over period into exactly those four drivers, broken down by product, customer, and sales rep.
Instead of one blended number, leadership sees exactly which lever moved and by how much - turning "margin is down" into "margin is down because mix shifted toward lower-margin products with this rep," a question they can actually act on.
4. Profit Leak Engine: 35 Rules Watching Every Part of the Business
Most margin leaks aren't big, obvious mistakes - they're slow drifts: a price that hasn't moved in a year while costs crept up, a customer quietly over-discounted for months, a SKU that's been selling at a loss without anyone noticing. No one has time to manually audit thousands of accounts and SKUs every month to catch them. We built what's now the Profit Leak engine - close to 35 distinct rules across costs, customers, orders, inventory, pricing, products, and sales that run over the business's live data and flag exactly where margin is eroding.
Each rule carries a severity and a dollar impact, turning a wall of spreadsheet rows into a prioritized action list - so instead of auditing everything, the team starts with what matters most.
5. Customer Health: A Churn Model That Understands Each Customer's Own Rhythm
By the time an account has gone fully quiet, it's often too late to win back - the real value in predicting churn is catching it early enough that a sales rep can actually do something about it, a call, a pricing conversation, better service, before the relationship is gone. But that only works if the warnings can be trusted. Our first churn model used a flat 90-day no-order cutoff, which wrongly flagged perfectly healthy seasonal and infrequent B2B buyers as at-risk - exactly the kind of false alarm that trains a sales team to start ignoring the alerts altogether.
We rebuilt it around each account's own ordering cadence instead: 18 purely behavioral features (recency, order frequency, revenue and margin trends, discount patterns, product mix) feeding a deliberately shallow, calibrated model. Deeper trees overfit on this data; the shallow model won, scoring a 0.76 test ROC-AUC against 0.71 for a deeper alternative.
Every score now ships with its top explainable drivers, translated into plain-English actions a sales rep can actually use - not just a risk number, but a reason and a next step.
6. Price Recovery List & Tracker: From Drifted Prices to a Recovery List You Can Track
Most margin leaks aren't big mistakes - they're slow drift: a price nobody's revisited in years while costs crept up, one customer quietly paying 10% less than everyone else for the same product, a sales rep avoiding an uncomfortable conversation. We built a recommendation engine that checks every customer-product combination against six signals - price trailing the peer median, rising costs, a flat price, compressing margin, multi-month margin decline, and volume - and only surfaces a candidate when at least two signals agree, so the business isn't acting on noise. It's live today on the business's real data, reachable right now through the AI assistant.
Each recommendation comes with a specific new price - chosen from one of four approaches depending on why it drifted, whether that's restoring the margin a cost increase ate into or closing half the gap to what similar customers already pay - plus a plain-English reason a rep can actually use in the conversation ("pays 7% less than similar customers," "our cost is up 6% since the last price change") and an estimated dollar value added per year. On one review of the business's full pricing book, roughly 70% of measured margin leakage was concentrated in just the top 10 customer-product combinations out of more than 7,000 - turning an overwhelming spreadsheet into a short, ranked recovery list instead of a chart someone has to interpret.
That loop is closed now: a tracker follows each accepted price change invoice by invoice to confirm the new price actually reached the bill, not just the agreement - catching the common case where a customer says yes but the ERP price is never updated, so the money everyone agreed to never shows up. It's live today, and we keep extending it as new edge cases turn up.
7. Weekly Digests: Insights Delivered, Not Just Available
Insights only help if someone actually sees them. We built a weekly email digest that lands directly in a regional manager's or executive's inbox with the specific accounts, prices, or leaks that need attention that week - no login, no dashboard to remember to check. It's the natural extension of everything else in the product: the system already knows what matters and why; this just makes sure the right person sees it without having to go looking.
8. Agents: A Direct Line From Question to Answer
Dashboards are great until you have a specific question, and digging through filters and tables to answer it costs time an executive doesn't have. We built an AI assistant so a CFO or VP of Sales could just ask. But early testing surfaced a real reliability problem: the same question could come back with two or three differently shaped answers in one session - not good enough when the answer might inform a real business decision.
We fixed it with a deterministic-first architecture: a fast intent match routes recognized business questions to pre-built, consistent answers, and only genuinely novel questions reach the language model, which calls curated tools against the database rather than generating its own queries.
The result is an assistant that speaks the business's own language - "accounts with reduced purchases," not "churn"; dollars framed as "at risk," "exposure," or "opportunity" depending on the account's health - and never guesses at a number it can't verify.
And it's built to keep growing. Today it already answers questions across customer health and churn, pricing and margin, and freight analysis - not a fixed script, but a framework designed to add new question domains as fast as the business raises them.
Business Outcome
The partnership delivered value on two levels: for Brian and Profit Matrix as a business, and for the manufacturers and distributors actually using the product day to day.
1. For Brian and Profit Matrix
This is no longer a prototype Brian has to apologize for - it's a live system running against real manufacturer ERP extracts, across multiple business units, refreshed on an ongoing basis. His customers' leadership teams have given direct feedback on how results are worded and which numbers they track day to day, and they're already asking for the next round of capability - the clearest sign a founder can get that what you built together is working.
Just as importantly, Brian has a technical partner for the parts of the business he can't carry alone: a system he can confidently demo, sell, and keep growing, instead of a product stuck in his own backlog.
2. For the Manufacturers and Distributors Profit Matrix Serves
14 manufacturing and distribution businesses now run their pricing and margin decisions through Profit Matrix, and together they've recovered an estimated $2.3M in margin that would otherwise have gone unnoticed - caught by the Profit Leak engine's close to 35 rules running across their live data every day.
On one customer's full pricing book, roughly 70% of measured margin leakage was concentrated in just the top 10 customer-product combinations out of more than 7,000 - the kind of result that turns Profit Matrix from a tool customers use into one they point back to. Every customer running the Profit Leak engine and churn model has renewed, and most are already asking for the next capability.