There is a specific revenue band where product businesses start to suffocate. It is not at launch. It is not at 1 million. It is somewhere between 3 million and 10 million, when the operation has outgrown the founder's ability to hold everything in their head but has not yet built the systems to hold it for them.

We have seen it inside food brands, consumer goods companies, and contract manufacturers. The symptoms are always the same. Orders slip. Inventory numbers stop matching reality. Cash gets trapped in places nobody can see. And the founder, who built the business on intuition and hustle, starts spending more time firefighting than building.

This is the Build Log for the three systems we see break first, every single time.

The Problem Nobody Warns You About

Between 3 million and 10 million, something changes. The volume of transactions crosses a threshold where manual tracking becomes a liability instead of a scrappy advantage.

At 1 million, you can track your SKUs in a spreadsheet. You know your top products by feel. You can eyeball your inventory levels and be close enough. Your order flow is manageable because the volume is manageable.

At 5 million, that same spreadsheet has 300 rows. Your top products shift seasonally and you do not notice until you are out of stock on your bestseller. Your inventory numbers are off by 15-20% because three people are updating the same file and none of them are updating it the same way. Orders come in from five different channels and each one has a different format, a different timeline, and a different set of expectations.

At 10 million, the spreadsheet is a liability. Every decision based on it carries risk. The founder is spending 15 hours a week on tasks that should take 3, not because the tasks are complex but because the systems underneath them are broken.

Here is what nobody tells you: the problem is not that you need better tools. The problem is that the three core systems every product business depends on were never designed for the volume you are doing now.

The Failed First Move

Almost every founder we work with has already tried to fix this. Their first move is always the same: hire people.

They bring on an operations manager. A bookkeeper. A warehouse lead. Sometimes all three at once.

And for a few weeks, it feels better. The new hires absorb some of the chaos. The founder gets a few hours back. Things seem like they are stabilizing.

Then the new hires start asking questions the founder cannot answer cleanly. "What is the process for this?" "Where does this data live?" "Who is supposed to approve this?"

The answers are: there is no process, the data lives in four places, and the founder approves everything because nobody else has the full picture.

Hiring people to operate inside broken systems does not fix the systems. It adds payroll to the problem. Now you have the same broken processes, the same data gaps, and the same bottlenecks, plus you are paying three more salaries.

The fix is not more people. The fix is better systems. Specifically, these three.

System 1: SKU Tracking + Sales Forecasting

This is the system that breaks first and costs the most when it does.

At 3 million, most product businesses are tracking SKUs in a spreadsheet or in a basic inventory system that nobody updates consistently. Sales forecasting is done by feel: the founder looks at last year's numbers, adjusts for what they think will happen this quarter, and places purchase orders based on gut instinct.

That works until it does not. And when it stops working, it stops working fast.

The Historical Data Trap

Here is the pattern we see. A brand has 18 months of sales data. They look at the last 12 months and use that as their forecast baseline. Seems reasonable.

But the data is lying to them. Here is why.

Last year's sales include stockout periods where they could have sold more but did not have the product. They include promotional spikes that will not repeat unless they run the same promotion. They include channel-specific trends that are invisible in the aggregate numbers.

If your bestselling product was out of stock for 6 weeks last summer, your historical sales data shows a dip. If you use that data to forecast this summer, you will under-order. You will stock out again. The cycle repeats.

We call this the historical data trap. The data tells you what happened, not what would have happened if your systems were working. And if your systems were broken during the period you are using as a baseline, your forecast inherits those broken patterns.

Three Forecasts, Not One

The fix is not more sophisticated software. The fix is running three forecasts instead of one.

Forecast 1: Baseline. What will sales look like if nothing changes? Same channels, same marketing spend, same product mix. Use your historical data but adjust for known stockouts and non-repeating events. This is your floor.

Forecast 2: Growth scenario. What happens if your planned initiatives work? New retail accounts, a product launch, an expanded marketing budget. Be specific. "We are adding 3 retail doors in Q2" is a growth scenario. "Things will probably grow" is not.

Forecast 3: Downside scenario. What happens if a key account pulls back, a supplier raises prices, or a channel underperforms? This is your stress test. It tells you the minimum inventory you need to hold and the maximum cash you need to have available.

Running all three takes about two hours per quarter. It is not elegant. It does not require a data scientist. But it forces you to think in ranges instead of single points, and that is the difference between a forecast that protects you and one that misleads you.

Forecast Accuracy

Here is the number that matters: forecast accuracy. Not whether your forecast was right, but how wrong it was.

Measure it monthly. Take your forecasted units by SKU, compare to actual units sold, and calculate the percentage variance. Industry benchmarks for mid-market product businesses sit around 70-80% accuracy at the SKU level. If you are below 60%, your purchasing decisions are essentially guesses.

Track it over time. If your accuracy is improving quarter over quarter, your system is working. If it is flat or declining, something in your data or your assumptions is off and you need to dig in.

The principle: Your forecast is not a prediction. It is a planning tool. It needs to be wrong in ways you can manage, not in ways that surprise you.

The test: Can you tell me, right now, your forecast accuracy for last quarter by SKU? If you cannot, this system is broken.

What to build:

System 2: Order Entry Consolidation

This is the system that creates the most daily friction and is the easiest to fix.

At 3 million, orders come from maybe two or three places. Your website. A couple of wholesale accounts. Maybe a marketplace. You can manage them manually because the volume is low enough.

At 5 million and above, orders come from five or more channels. Your own e-commerce platform. Amazon or another marketplace. Wholesale accounts that email POs. Distributors with their own portals. Direct accounts that text or call. Maybe a retail partner with an EDI feed.

Each channel has its own format. Each one has different payment terms. Each one has different fulfillment expectations. And each one feeds into your operation at a different point, in a different way, with a different level of data quality.

The 5-Channel Problem

Here is what the 5-channel problem actually looks like on a Tuesday morning.

Your e-commerce platform processed 47 orders overnight. They auto-import into your fulfillment system. Good.

A wholesale account emailed a PO as a PDF attachment. Someone needs to open it, read it, and manually enter the line items into your system. That takes 15 minutes if the PO is clean. It takes 30 if there are questions.

Your distributor placed an order through their portal. You need to log in, download the order, cross-reference the item numbers (because their SKU system does not match yours), and enter it manually. Another 20 minutes.

A retail buyer texted your sales rep with a reorder. The sales rep forwarded it to your ops person. The ops person has to figure out the account details, the shipping address, the payment terms, and whether this order has already been entered somewhere else. Another 25 minutes, plus a back-and-forth thread.

Amazon orders auto-import but the fulfillment requirements are different: specific labeling, specific packaging, specific ship-by dates. Someone has to flag and separate these from the rest of the queue.

Total time spent on order entry that morning: 90 minutes. And that is a clean day. On a messy day with questions, corrections, and duplicate entries, it can hit three hours.

Multiply that by 250 working days a year. You are looking at 375 to 750 hours annually spent on a process that adds zero value to your product, your brand, or your customer experience.

Standardize the Entry Points

The fix is not a single magical system that integrates everything. That system does not exist for most mid-market brands, and the ones that claim to do it cost six figures and take a year to implement.

The fix is standardization. You cannot control how your customers send you orders. But you can control how those orders enter your system.

Step 1: Create one order entry template. Every order, regardless of source, gets entered in the same format. Same fields. Same sequence. Same validation rules. If a field is missing, the order does not move forward until it is complete.

Step 2: Assign channel owners. Each channel gets one person responsible for entry. Not "whoever gets to it first." One person who owns that channel's orders from receipt to entry. They know the format. They know the quirks. They catch the errors.

Step 3: Set entry deadlines. All orders received before noon get entered by 2 PM. All orders received after noon get entered by 10 AM the next day. No exceptions. This creates a rhythm that the warehouse can plan around.

Step 4: Track entry errors. Every order that has to be corrected after entry gets logged. Track which channel, which type of error, and how long the correction took. After 30 days, you will see the pattern. One channel will be responsible for 60% of your errors. Fix that channel first.

The principle: Order entry is not a creative act. It is a data entry process that should be identical regardless of source. Standardize the process, not the customer.

The test: Time how long it takes to enter orders from your five biggest channels. If any channel takes more than 5 minutes per order on average, the entry process for that channel is broken.

What to build:

System 3: Invoicing + Cash Collection

This is the system that does not feel broken until you look at your bank account and realize you are profitable on paper but short on cash.

At 3 million, invoicing is manageable. You have a handful of wholesale accounts. You send invoices manually. You follow up when something is overdue. The volume is low enough that you can keep the status of every open invoice in your head.

At 5 million and above, you have 30, 50, maybe 100 active accounts. Each one has different payment terms. Each one pays differently: some by check, some by ACH, some by credit card, some by carrier pigeon if you let them.

And here is what happens: invoicing becomes the thing that gets done "when there's time." Which means it gets done late. Which means your customers pay late. Which means your cash position is always worse than your P&L says it should be.

The 30-Minute Test

Here is a test we run with every client. We ask one question: "Can you tell me, right now, your total outstanding receivables, broken down by current, 30 days, 60 days, and 90+ days?"

If they can answer in under 30 minutes, their invoicing system is functional. Not great, but functional.

Most cannot answer at all. Not because they do not care about cash. Because the data is scattered across QuickBooks, a spreadsheet, an email thread with their bookkeeper, and a Post-it note on someone's monitor.

The average time to answer this question across the last eight brands we have worked with: 4.5 hours. Some could not answer it in a day.

That is not an invoicing problem. That is a visibility problem. And visibility problems become cash problems faster than almost anything else in a product business.

The Cash Collection Fix

Step 1: Invoice on shipment, not on a schedule. Most brands invoice weekly or biweekly in batches. This means a product that shipped on Monday does not get invoiced until Friday. That is 4 days of free float you are giving your customer. Invoice the day the product ships. Automate it if your system allows it. If it does not, make it a daily task with a specific owner.

Step 2: Implement a 3-touch follow-up cadence. Touch 1: A friendly reminder at 7 days past due. "Just checking that you received invoice #1234." Touch 2: A firmer note at 21 days past due. "This invoice is now 21 days past terms. Please confirm payment date." Touch 3: A direct conversation at 30 days past due. Phone call, not email. "We need to resolve this before the next order ships."

Most brands do Touch 1 inconsistently and never do Touch 2 or Touch 3. The result is that late payers learn they can pay late without consequences. And once a customer learns that, they will always pay you last.

Step 3: Build an aging report you actually review. Weekly. Every Monday morning. 15 minutes. Look at every invoice over 30 days. Assign follow-up actions. Track them. This is the single highest-ROI habit in cash management and almost nobody does it consistently.

Step 4: Set credit limits. If a customer has more than 60 days of outstanding invoices, new orders go on hold until the balance is resolved. This feels aggressive. It is not. It is basic cash management. You are not a bank. You should not be financing your customers' operations.

The principle: Revenue is not cash. An invoice is not money. Cash is only cash when it is in your bank account. Manage the collection process with the same rigor you manage your production process.

The test: What percentage of your invoices are paid within terms? If you do not know, or if the answer is below 80%, this system needs work.

What to build:

The 10x Friction Test

Here is the framework we use to determine whether a system is ready for the next revenue milestone.

Take your current process for any of the three systems above. Now imagine your volume is 10x what it is today. Not 2x. Not 3x. 10x.

Ask three questions:

1. Does this process still work at 10x volume?

If you are manually entering orders from PDFs and it takes 90 minutes a day at current volume, what happens at 10x? That is 15 hours a day. It does not work. The process breaks.

If your answer is "we would just hire more people," you have failed the test. Hiring is not a systems fix. Hiring is a payroll increase that inherits the same broken process.

2. Does this process depend on a specific person?

If one person leaving would cause this system to fail, you do not have a system. You have a dependency. At 10x volume, that dependency becomes a single point of failure that can shut down your operation.

3. Can a new hire operate this process on day one with written instructions?

If the process requires weeks of training, institutional knowledge, or "you just have to know how we do things here," it is not a system. It is tribal knowledge wearing a system's clothes.

Any system that fails any of these three questions is not ready for growth. Fix it before you scale. Because scaling a broken system does not fix it. It makes it break louder.

The Scaling Self-Audit: 12 Questions

Run through these honestly. Score yourself 1 (yes, we have this) or 0 (no, we do not). Anything below 9 means you have system debt that will slow your growth.

SKU Tracking + Sales Forecasting

  1. Do you have a single source of truth for SKU-level inventory that is updated at least weekly?
  2. Can you produce a forecast accuracy report for last quarter by SKU category within 30 minutes?
  3. Do you run at least two forecast scenarios (baseline + one other) before placing major purchase orders?
  4. Do you reconcile system inventory against physical counts at least quarterly for your top 20% of SKUs?

Order Entry Consolidation

  1. Do all orders, regardless of channel, get entered using the same template and format?
  2. Does every channel have a single named owner responsible for order entry?
  3. Are there published entry deadlines that the warehouse can plan around?
  4. Do you track order entry errors by channel and review the data monthly?

Invoicing + Cash Collection

  1. Are invoices sent within 24 hours of shipment?
  2. Do you have a documented follow-up cadence for overdue invoices with assigned owners?
  3. Do you review an aging report weekly?
  4. Do you have credit limits or order holds for accounts past 60 days due?

Score 10-12: Your systems are solid. Focus on optimization and automation.

Score 7-9: You have gaps that are costing you time and cash. Prioritize the lowest-scoring section.

Score 4-6: You are operating on manual processes that will break under growth. Fix these before scaling.

Score 0-3: You do not have systems. You have habits. Start with System 1.

Build This Yourself

You do not need a consultant or a six-figure software implementation to fix these three systems. You need a spreadsheet, a calendar, and 90 days of discipline.

Month 1: Pick the most broken system. Use the self-audit scores. Whichever section scored lowest, start there. Spend the first two weeks mapping the actual process (not the ideal process, the real one). Spend the second two weeks building the templates, assigning owners, and setting deadlines.

Month 2: Implement and measure. Run the new process for 30 days. Track the metrics: forecast accuracy, order entry time by channel, days to collect by account. Do not optimize yet. Just measure. You need a baseline.

Month 3: Fix the second system. Take the same approach. Map, build, implement, measure. By the end of Month 3, you have two of the three systems operating on documented processes with measurable baselines.

The third system gets fixed in Month 4. By Month 6, all three are running, measured, and improving.

This is not fast. It is not glamorous. It does not involve AI, machine learning, or any technology more sophisticated than a spreadsheet and a weekly calendar reminder.

But it works. And it works because it fixes the infrastructure that everything else depends on.

Ops Intel

A few things that crossed our radar this week that are worth your attention.

McCormick and Unilever both reported supply chain margin expansion this quarter. McCormick's cost savings program delivered $85 million in annual savings, driven largely by procurement consolidation and manufacturing efficiency. Unilever's productivity programs hit 1.8 billion euros in cumulative savings. The takeaway for mid-market brands: the big players are getting leaner. If your cost structure is not improving, the margin gap between you and your larger competitors is widening.

Coca-Cola's new CEO is an operations guy. John Murphy, who has been the company's CFO and President, takes the top job this month. His background is in supply chain and financial operations, not marketing. This is a signal. The era of brand-led CPG leadership is giving way to operations-led leadership. If you are building a product business, your operations capability is becoming your competitive advantage, not just your cost center.

Tariffs are reshaping sourcing decisions faster than anyone expected. New tariff structures on Chinese imports are pushing brands to re-evaluate their supplier base. We are seeing brands that sourced 80% of their packaging from China scrambling to find domestic or near-shore alternatives. If you have not stress-tested your supply chain against a 25-50% tariff scenario, do it this week.

DHL partnered with Locus Robotics to deploy autonomous mobile robots in their North American fulfillment centers. They are targeting a 50% increase in pick productivity. For mid-market brands using 3PLs, ask your provider what their automation roadmap looks like. The gap between automated and manual fulfillment costs is going to widen significantly over the next 24 months.

MIT researchers published a paper on multi-agent robotics for warehouse operations. The short version: coordinated robot swarms can now handle mixed-SKU picking in unstructured environments. This is still 3-5 years from mid-market deployment, but it is worth watching. The warehouse labor model is going to change fundamentally.

Toyota's "Swarm" logistics concept is being piloted in Japan. Instead of fixed conveyor systems, they are using networks of small autonomous vehicles that self-organize based on demand. The concept reduces fixed infrastructure costs by 40% and increases throughput flexibility by 60%. This is the future of manufacturing logistics, and it is closer than most people think.

The systems that break between 3 million and 10 million are not mysterious. They are predictable. And the fixes are not expensive. They are disciplined.

Build the infrastructure now. Your future revenue depends on it.