Michael Dell

The pattern they actually represent

Dell's 1998 annual report names five reasons the company was winning, and not one of them is about the computer. No network of wholesale and retail dealers to support. No dealer mark-ups. No inventory cost sitting in that channel. No competition for shelf space. Less risk of parts going obsolete while they wait to be sold. Five clauses, and every one describes either the route to the customer or the money moving along it.

The machines were assemblies of components anyone could buy. At Comdex in 1986 the pitch was comparative rather than novel — a 12 MHz 286 at $1,995 against IBM's fastest 6 MHz 286 at $3,995, or in his own summary, twice as fast and half the price. He did not own the machine. He owned the road it travelled and the order in which money moved along that road.

That second half is what gives the pattern teeth, because it landed on the balance sheet. In the same filing, days of supply in inventory fall from thirty-one to thirteen to seven across three fiscal years. Seven days. A company selling hardware was holding a week of it, and everything after that follows: little capital tied up, little stock left to lose value on a shelf, and cash from buyers arriving before the bills from suppliers went out. Revenue went from $2.9bn to $12.3bn in four years, funded in large part by the customers.

Then the filing adds the half most people skip. Direct contact also produced a running record of what customers were asking for, and that record shaped what got built next. The channel is not only where the product goes out. It is where the specification comes back in.

The blockage it speaks to

This is the entry to read if you believe a good enough product sells itself.

The belief rarely states itself outright. It arrives as sequencing: distribution is a downstream chore, bolted on once the build is good enough to deserve it. So the hours go into the half of the business you already know how to do, and the route to the customer stays a placeholder marked later.

What keeps the belief alive is that it never gets tested. Nothing has failed, because nothing has been offered. The build is not finished, so the market has not answered, so the theory stands. The pattern keeps you permanently upstream of the one conversation that could disprove you.

Dell is the cleanest available counterexample, and his own company put it in writing. He won a hardware market without ever having the best hardware. When the business had to state its advantage to shareholders, it named a dealer network it did not have to pay for and inventory it did not have to hold, then described collecting from customers before paying suppliers as the thing that let it grow that fast without much capital.

Read as an instruction, that is uncomfortable, which is the point. The distribution decision was the design decision. He drew the route first and let the machine be ordinary, and the route paid for the company.

Three moves you can steal

Compute your cash conversion cycle this week, on one line, from real numbers. Days you wait to be paid, plus days your stock or unbilled work sits, minus days you take to pay your own suppliers. That is the formula Dell's filings use, and the answer names who is funding your growth right now. Positive, and it is you. Negative, and it is your customers.

Draw the route to your customer as a list of hops and put two numbers on each hop. What that party takes, and how many days it adds to information coming back. Founders inherit a channel rather than choose one, so the hops turn into furniture nobody sees. Deleting one hop usually improves margin, cycle time and product decisions at once.

Sell one unit of the next thing you plan to build, before it exists. One buyer, one price, one commitment. Not a waitlist and not a letter of intent — an order. Build-to-order gets filed as a manufacturing technique; it is really a sequencing rule. The order comes first and the cost comes second, and everything downstream follows from that order.

Where the pattern breaks

An operating-model advantage is a lead, not a moat, and Dell's filings record the lead closing. Between fiscal 2006 and 2008 the cash conversion cycle narrowed from negative forty-three days to negative thirty-six, inventory rose from five days of supply to eight, and the 2008 annual report states it flatly: the company lost 1.9 points of share during 2007. The repair cost part of the advantage: the same filing blames the narrowing cycle partly on a larger presence in the retail channel, the channel the model existed to skip. In April 2007 Michael Dell told employees the direct model had been a revolution but was not a religion. By June the machines were in Wal-Mart.

What to do this week

Take one order for something you have not finished.

Pick the next item on your build list, name a price, and find one person who will commit to it this week. Not a waitlist, not a call to discuss it later — money, a signature, or a purchase order. If that feels early, look at what the feeling protects: while nothing has been offered, nothing can be refused, and the theory that the product sells itself stays intact.

Whatever comes back is the cheapest information you will buy all year. An order says the route works, and hands you cash before you spend any. A no says which part of the offer is wrong while it is still cheap to change. Silence says the route itself is the problem, which is the finding you have been postponing.