Chapter 4: The Mirror

Oscar shows up at MapleCo on a Wednesday morning at 8:15, fifteen minutes before Lisa arrives.

He brings a notebook. Not a laptop, not a tablet, a spiral-bound notebook with a pen clipped to the cover. Sam will later learn that Oscar always starts discoveries this way. Laptops create distance. Notebooks create conversation.

“I want to watch,” he says. “Not the process you’d describe if I asked you. The process that actually happens. There’s always a difference.”

Sam offers him coffee and a chair near Lisa’s desk. When Lisa arrives at 8:30, Sam introduces Oscar as “someone who’s going to watch how we work for a day. Just pretend he’s not here.”

Lisa gives Sam a look that says I will absolutely not pretend he’s not here. But there’s something else in the look too. Something harder. Lisa has been through this before. Three years ago, a “process consultant” watched her for a day, wrote a report that described her work incorrectly in fourteen different places, and recommended software that didn’t connect to QuickBooks. The software was purchased. It didn’t work. Lisa spent two months cleaning up the mess. Nobody apologized.

She sits down and starts her morning routine without a word.

Oscar watches.

At 8:37, Lisa opens her email. There are fourteen new messages since yesterday. She scans the subject lines, mentally sorting: purchase order, purchase order, tracking request, payment confirmation, spam, purchase order, question from a customer about grade availability, purchase order, supplier quote, spam.

She opens the first PO email. It’s from Henderson’s. A PDF is attached. She downloads it, opens it, and begins reading.

Oscar leans forward. “Talk me through what you’re doing right now.”

“Processing a purchase order.” Lisa doesn’t look up.

“Can you walk me through the steps?”

“I open the email. I download the PDF. I enter the data into QuickBooks. I update the tracker.” Her voice is flat. Efficient. This is the version she gives consultants. The clean version. Three sentences that describe a fifteen-step process.

Oscar is quiet for a moment. Then: “What about the pricing?”

Lisa pauses. She looks at him for the first time. That’s not a consultant question. Consultants ask about software and systems. They don’t ask about pricing.

“What about it?”

“Do you check it every time, or just for certain customers?”

Lisa glances at Sam. Sam nods.

“Every time,” Lisa says, and something shifts. Her voice is still careful, but it’s no longer the three-sentence version. “Henderson’s is on a contract price sheet, but the prices changed in September and not all the old POs in the system have been updated. So I cross-reference against the master price list.” She opens a Google Sheet in another tab. “This is the price list. It was updated last month. But some customers are still on the old prices because we haven’t sent them the new contracts yet.”

“How many price lists are there?”

Lisa pauses. “Well, there’s the master list in Google Sheets. There’s a version in QuickBooks that’s supposed to match but doesn’t always. And then Henderson’s has their own pricing agreement, which is different from the master list because they get volume discounts.”

Oscar writes something in his notebook. Sam can’t see it, but what he wrote is: Three sources of truth for pricing. None of them agree. The spreadsheet is not the issue. The fact that it has become a personality is the issue.

By 10:30, Oscar has watched Lisa process four purchase orders. Each one took between 12 and 18 minutes. Each one involved opening at least three different applications. Each one required at least one cross-reference to verify a price, a customer code, or a product description.

He’s also watched Lisa answer two phone calls (a customer asking about delivery status, which required Lisa to look up the order in her spreadsheet, then check the carrier website), respond to three emails (a supplier confirming a shipment date, a customer requesting an invoice copy, and Jordan asking where to find the exchange rate for today), and rescue a stuck printer.

In two hours, Lisa has processed four orders and handled six interruptions.

Oscar asks to walk the warehouse. Sam hesitates, but agrees.

Ray meets them at the loading dock with the warmth of a man being introduced to his fourth consecutive consultant. “Morning,” he says. He does not ask Oscar’s name.

Oscar doesn’t seem bothered. He watches the warehouse for twenty minutes without saying anything. He watches a picker read a packing slip, cross-reference it against a handwritten note on the whiteboard, walk to the wrong aisle, walk back, check the note again, and then pull the correct pallet.

“That note on the whiteboard,” Oscar asks Ray. “How often does it change?”

“Whenever someone moves inventory and remembers to update it.” Ray pauses. “So, sometimes.”

“And when it’s wrong?”

“Then somebody ships the wrong thing and I get a phone call about it.”

Oscar writes in his notebook. Ray watches him write and says, “You going to fix that with an app?”

“I don’t know yet,” Oscar says. “I’m still figuring out if you have a whiteboard problem or a systems problem.”

Ray looks at Oscar for a long moment. Something in his expression shifts from dismissal to cautious interest. “First guy who’s asked that,” he says.

Back in the boardroom, Oscar draws a timeline on the whiteboard.

“Here’s what I just watched,” he says. “One purchase order, start to finish.”

He maps it out, step by step:

1. Email arrives with PO attached
2. Lisa opens email, identifies it as a PO
3. Lisa downloads the PDF
4. Lisa opens the PDF and reads it
5. Lisa opens QuickBooks
6. Lisa creates a new sales order
7. Lisa types the customer name
8. Lisa types the PO number
9. Lisa types each line item (product, quantity, price). For Henderson’s, that’s three line items
10. Lisa cross-references the price against the Google Sheet
11. Lisa checks the exchange rate (Henderson’s pays USD)
12. Lisa saves the sales order
13. Lisa opens the Google Sheet tracker
14. Lisa adds a new row with the PO details and marks it “Received”
15. Lisa moves the original email to a “Processed” folder

“Fifteen steps,” Oscar says. “And that’s a clean one. No exceptions, no questions, no phone calls in the middle.”

Sam stares at the whiteboard. She knew Lisa entered POs, but she’d never seen it mapped out like this.

“Now multiply that by forty to sixty orders a week,” Oscar says.

Sam does the math in her head. At 15 minutes per PO, and 50 POs per week, that’s 12.5 hours. Just for initial entry. Not counting tracking, invoicing, reconciliation, or any of the dozen other manual steps that follow.

He circles steps 7 through 11. “This is where errors happen. Not because Lisa makes mistakes. Because everyone makes mistakes when they’re copying data for the twentieth time that day while also answering the phone.”

Sam thinks about Mrs. Chen. The wrong grade. A single field.

“You don’t have one workflow,” Oscar says quietly. “You have six workflows in a trench coat pretending to be one.”

Lisa hasn’t said anything since Oscar started drawing. She’s been watching him reduce her entire day to fifteen lines on a whiteboard. Each step clean and numbered and obvious. She’s spent eight years learning to do this fast, to do it right, to be the person the business can’t run without. And now a man with a notebook has turned it into a diagram that looks like it was designed to be handed to a machine.

She doesn’t say anything to Sam. She goes to lunch, sits in her car for ten minutes, and comes back. She processes the next PO. But something has shifted, and she doesn’t have a name for it yet.

Oscar flips to a new section of the whiteboard.

“I need to show you something that’s going to be uncomfortable.”

He draws five boxes in a row, like steps on a staircase.

Stage 1: Manual. Everything done by hand. Spreadsheets, email, paper. This is where MapleCo was fifteen years ago.

Stage 2: Tool-Assisted. You have software, QuickBooks, maybe a CRM, but humans do all the connecting work between systems. “This is where you are right now. You have tools, but you’re the glue.”

Stage 3: Semi-Automated. Some workflows are automated, but you still need to kick things off or handle exceptions manually. “This is where most of my clients want to get to first.”

Stage 4: Systematized. Core processes run on their own. Humans handle exceptions and strategy. “This is where the magic happens. Not because of the technology, because the business owner stops being the bottleneck.”

Stage 5: Optimized. The system learns and improves. Process mining, predictive analytics, continuous optimization. “This is where large enterprises are. You don’t need to worry about this yet.”

“Where do you think MapleCo is?” Oscar asks.

Sam considers. She has QuickBooks. She has a Google Sheet tracker. She has a carrier website. She has a supplier portal. She has software. She’s not running things on paper.

“Stage 3,” she says. “Maybe late Stage 2. We have all these systems. They just don’t talk to each other well.”

Oscar puts down the marker. “That’s the answer I get from almost every business owner. And it’s almost always wrong by at least one stage.”

“What do you mean?”

“Having software isn’t the same as being tool-assisted. Being tool-assisted means the tools are actually helping. Your tools are creating work. Lisa opens five applications to process one purchase order. She’s not assisted by those tools. She’s employed by them. She works for QuickBooks, not the other way around.”

Sam opens her mouth to argue. She wants to say that’s not fair, that they have systems, that they’re not some paper-and-fax operation.

But she thinks about the six browser tabs. She thinks about the three price lists that don’t agree. She thinks about Marc’s sugar shack in Quebec, where three people run twelve thousand taps because the monitoring system actually works. And she thinks about her own business, where forty-five people run five hundred orders a week and she still can’t take a vacation.

“Stage 2,” she says quietly. “Early Stage 2.”

“That’s right. And there’s no shame in that. The shame would be thinking you’re at Stage 3 and building for Stage 3 when you’re really at Stage 2. That’s how you end up spending $30,000 on software that doesn’t fit.”

He taps Stage 3. “This is where we’re going. Semi-automated. Your highest-pain processes run with minimal human involvement. Lisa goes from data entry to exception management. You go from firefighting to approving.”

“The next step,” Oscar says, “is documenting exactly how your business actually works. Not the version on your website. Not the version you’d describe at a networking event. The real one.”

Sam shifts in her chair. “I kind of thought that’s what you just did. Watching Lisa.”

“That’s one process. You have dozens. And here’s what I’ve learned from doing this with thirty-plus organizations: every single client tells me their processes are simple. ‘Oh, it’s straightforward. We get an order, we process it, we ship it.’ Then we start mapping it, and it’s never simple. There are always exceptions, workarounds, judgment calls, and steps that only one person knows about.”

He tells Sam about a construction company he worked with. The owner said their invoice processing was “basic.” It turned out they had seven different approval workflows depending on the project, the contractor, and whether the amount was above or below $10,000. Two of those workflows existed only in the office manager’s head.

Sam is quiet for a moment. “You’re saying Lisa’s workarounds are actually the process documentation.”

“Lisa’s workarounds ARE the business. Everything else is just what you assumed.”

Oscar gives Sam homework.

“Before our next meeting, I want you and Lisa to do something. Pick your three most painful processes, the ones that eat the most hours and create the most errors. For each one, I want you to write down every single step. Not what the software manual says. What actually happens.”

He gives her a framework:

Who does each step? What system or tool do they use? What data are they moving, and from where to where? How long does it take? Where does it break? What are the common errors or exceptions? What’s the workaround? When something goes wrong, what does Lisa actually do?

“Don’t try to make it pretty,” he says. “I want the ugly version. The one where step four is ‘call Sam because she’s the only one who knows the password to the supplier portal’ and step seven is ‘check if the exchange rate looks right based on vibes.’”

Sam laughs for the first time in weeks.

“I’m serious,” Oscar says. “The ugly version is the valuable one. If you give me the clean version, I’ll build the wrong thing. Every contractor who’s ever disappointed you built the clean version. I want to build the one that actually works for the way you actually work.”

Three days later, Sam calls Oscar.

She’s not calling to chat. She’s frustrated. Henderson’s Q3 report took Lisa two and a half days. Jordan processed a batch of POs with three errors because he was rushing. A supplier in Quebec missed a shipment and nobody noticed until a customer called to ask where their order was. And Sam has spent the last week answering questions, showing Oscar around, and doing homework instead of fixing any of these things.

“Oscar, I need to be honest with you. I’m paying you to fix my business and so far all we’ve done is talk about my business. I’ve answered more questions in the last two weeks than I answered in my entire due diligence when I took over MapleCo. When does the fixing start?”

There’s a pause. When Oscar speaks, his voice is calm but direct.

“What would you like me to fix first?”

“The PO processing. Obviously.”

“OK. Which customers send POs as PDFs? Which send them as inline email text? Which send them by fax?”

“I… Lisa would know that.”

“What’s the most common exception type? Pricing mismatch, unknown product code, or format the system can’t read?”

“I don’t know.”

“If the system encounters a PO with a product code it doesn’t recognize, should it reject the order, flag it for review, or try to match it against similar codes?”

Sam is quiet.

“That’s why we’re still asking questions. You want me to build the right thing. The right thing requires knowing the answers to about two hundred questions like those. Every contractor who disappointed you skipped this part. They built the clean version. The obvious version. And it broke the first time it encountered something that wasn’t obvious.”

Sam’s jaw tightens. She knows he’s right. She hates that he’s right. She hates that being right means another week of questions before anything changes.

“How much longer?” she asks.

“We’re close. One more meeting with Lisa, and I’ll have what I need to start building.”

Sam hangs up. She calls Gerald.

“Your automation guy is driving me crazy. All he does is ask questions.”

“Is he wrong?” Gerald says.

Sam doesn’t answer. Because he’s not.

That night, Sam and Lisa stay late.

They start with purchase order processing, the obvious one. Lisa describes each step while Sam writes it on a legal pad. It takes longer than expected. There are steps Sam didn’t know about. There are judgment calls Lisa makes unconsciously. She can tell from the formatting of a PO which customer it’s from before she even reads the header. She knows that one supplier requires orders placed before noon EST to ship same-day. She knows that the Golden grade has a longer lead time than Amber or Dark.

None of this is written down anywhere. It lives in Lisa’s head.

They map out invoicing next. Then the bank reconciliation. By 8 PM, they have three legal pads full of messy, honest process documentation.

“This is kind of terrifying,” Sam says, looking at the pages. “If you got hit by a bus tomorrow, none of this knowledge would survive.”

“That’s called key-person risk,” Lisa says. Then, quieter: “I can keep fixing this manually, but that’s not the same as this being fine.”

They both sit with that for a moment. Sam makes a mental note.

The next day, Sam sends Oscar a photo of the three legal pads.

Oscar replies: “This is perfect. This is exactly what I need. Most people send me a three-bullet summary. You sent me the truth.”

Then he adds: “By the way, that thing you said about Lisa and the bus? That’s not just a risk management problem. That’s actually the strongest argument for automation I can make.”

“How so?”

“When you automate a process, you’re not just saving time. You’re extracting institutional knowledge from people’s heads and encoding it into a system. Every rule Lisa follows, every judgment call she makes, every workaround she’s invented, we capture that. We make it visible. We make it repeatable. If Lisa goes on vacation, gets sick, or, god forbid, decides to leave, the system doesn’t forget.”

Sam thinks about this for a long time.

“So it’s not about replacing Lisa.”

“It never is. It’s about making sure the business doesn’t depend on any single person’s memory. Including yours.”

Especially hers.

And once you see clearly, you’re ready for your first win.