The six checklists

Where this comes from: I opened Rwanda on a $150,000 order while my team wanted to wait six months for a tender worth twenty times more. The small order made me a supplier who had already been paid, on time, before the big one came. The first order I ever took from one of the largest energy companies on earth was $250,000. That was the point.

1. Take the small order.

   A small order is not revenue. It is a receipt. It buys you the right to be considered for the second one.

2. Work out what the first order is auditioning.

   With the largest buyers the size is irrelevant and the delivery record is everything. They are not testing your price. They are testing whether you do what you said.

3. Check whether you are being given an order or a vendor number.

   These are not the same thing. The vendor number is worth more, and it is the one that is hard to get.

4. Enter where you are genuinely different, not where your catalogue is biggest.

   It is useless to convince a substation contractor to buy building wire. You are wasting your time. Sell him switchgear and your probability goes from zero to eighty. Widen from inside the account afterwards.

5. Deliver the first one perfectly, even if it looks thin on paper.

   You are not pricing a transaction. You are buying your way onto a list, and an approved competitor beats a better product every time.

6. Register the customer with your principal in writing, with acknowledgement, before you need to.

   An agency is only as strong as the paper trail behind it. Protect the channel while everyone is still friendly.

7. Plan the second order before you ship the first.

   One account compounding twelve times beats twelve accounts that never repeat. I grew one gas-sector account in southern Iraq from $500,000 to $6 million in two years, and it started as a small first order.

Where this comes from: When Syria began reopening, everyone wanted to be first to sell. I refused to quote until the banking and clearing path was real, because a quote is a commitment. In a new country the first quote is never about price. It is about whether the deal can clear.

1. Who wrote the specification.

   Not who is buying. Who wrote it. By the time a tender is published the specification is written and the shortlist is informal but real.

2. How the buyer got paid last time.

   If their customer pays late, you will be paid late. Their problem becomes your cash flow problem.

3. Whether anyone local is already approved.

   An approved competitor beats a better product. Every time. You are not selling a product, you are buying your way onto a list.

4. What happens at the port.

   Clearance delays have destroyed more margin than discounting ever has. You cannot quote a lead time you do not control.

5. Who I can call at 9pm.

   If there is nobody, I am not in that market yet. I am visiting it. Coverage is a list of names. Presence is one person who answers outside office hours.

Then, before the number leaves your desk

• A cost is only a cost at the point of delivery.

  One dollar at the factory in India is not the actual cost. The actual cost is what it is costing me standing in Tanzania, next to the market price.

• Confirm the HS code with the manufacturer, never assume it from the agent.

  One character is the difference between 25 and 35 percent duty. That single character can be your whole margin.

• Model the money timeline against the goods timeline, then size the advance to the gap.

  Invoice in ten days, paid in thirty to forty, cash back around fifty, against a shipment landing in three weeks. That gap is the reason the advance moves from 50 to 70 percent.

• Rebuild your competitor's cost stack before you decide your own price.

  Currency buffer, then the financing cost of holding inventory, then margin. Their price is an output, not a mystery.

Where this comes from: One of the largest industrial customers in the country, on thirty-day terms. Settled invoices from that period ran four to five months. One invoice passed 370 days overdue and still carried a balance. Their size made the risk look like no risk at all.

1. Look at how they paid last time, not at how big they are.

   Size is not creditworthiness. The biggest name in the country can still be the slowest payer you have.

2. Pull the actual settlement history before you agree the terms.

   Not the agreed terms. The real dates. The gap between the two is the number you are actually underwriting.

3. Price the delay into the deal, or decline it.

   Their payment cycle becomes your cash flow problem, not theirs. If the terms cannot carry the delay, the deal is not the deal you think it is.

4. Run profitability and liquidity as two separate calculations.

   A job can be profitable and still be one you regret for cash flow. Run both before accepting anything large.

5. Read concentration as a risk number, not a performance number.

   Six orders from one salesperson with eighty percent from a single customer is a warning, not a result to celebrate.

6. Measure what it costs to serve the account before you judge the revenue.

   If eighty percent of your effort goes into one account and generates one million of profit, it may not even cover your own cost to the company.

7. Treat a market-wide payment reputation as a rumour until you have assessed the customer yourself.

   Payment-risk labels get applied to whole countries and then to individual companies nobody has actually assessed. That is where the opportunity usually is.

8. Solve where the money clears before you agree to sell anything.

   In a reopening market everyone races to be first to sell. The ones who survive are first to solve how they get paid.

Where this comes from: I recruited and trained more than 200 people in a country where I knew nobody, and structured the organisation from zero. In a new market you are not hiring for skills that already exist there. You hire for attitude and build the skill yourself. That is a training budget decision, not a recruitment one.

The four questions

1. "Tell me about something you did not know how to do. What did you do next?"

   You are listening for whether they went and found out. Not knowing something is acceptable. Not finding out is not.

2. "Walk me through something you finished last month, start to end."

   You want evidence of what was done, not a description of what will be done. Anyone can explain how they would perform.

3. "What went wrong, and what was your part in it?"

   Someone who cannot locate themselves inside a failure will not be honest with you on the day it costs them something.

4. "Who did you tell, and how quickly?"

   The gap between noticing a problem and saying it out loud is the number you actually care about. Put the difficult thing on the table the same day you feel it.

The rules I run afterwards

• One person, one manager.

  Two sources of instruction produce a confused employee and a ready-made excuse.

• Every scope has one owner, A to Z.

  How it gets done is theirs. Whether it gets done is not negotiable.

• Do not manage the people who report to your managers.

  If you have to, the layer between you has no purpose.

• Keep rotating people until the arrangement fits.

  Your job is to find the best in every person, and to keep adjusting until it works. That is not indecision, it is the work.

• Every open item leaves the room with a deadline and an agreed exit.

  Find someone within twelve hours. If we do not find anyone, we decline the job and move on rather than let it drift.

• Internally we solve, externally we speak with one voice.

  A manager who routes the customer onward has not managed anything.

• Customer data belongs to the company, not to somebody's phone.

  An organisation that runs on personal contacts and personal WhatsApp is not an organisation yet.

Where this comes from: A $30 million sole-supplier award on a national airport terminal, the largest single award of that period. Sole-supplier status is decided long before the tender. By the time the document appears, the specification has usually already chosen.

1. Find out who wrote the specification.

   If you are reading the document for the first time on the day it is published, you are already late.

2. Read the specification for fingerprints.

   Clauses that happen to match one manufacturer's standard range are not a coincidence. They are a signature.

3. Count how many suppliers can actually comply.

   If the honest answer is one, the tender has already chosen and you are being used to make the file look competitive.

4. Ask when the budget was approved, not when the tender was issued.

   The decision lives with the budget. The document is the paperwork that follows it.

5. Test the counterparty's authority to buy, not the warmth of the relationship.

   A ten million signing limit at a thirty percent win rate means thirty purchase orders a month for that target to be real. Do the arithmetic before you invest the travel.

6. Work the six months before the document, not the two weeks after it.

   Intelligence is only worth anything if it arrives before the document does. I watch for the signals rather than wait for the tender.

7. If you cannot win on price, change who is bidding.

   A partnership rewrites the comparison the buyer is making. I won a $22.5 million three-year supply agreement as a joint offer rather than by competing alone on price.

8. Know what your own factory is short of this month.

   Otherwise you are quoting lead times you cannot keep, and promises you cannot deliver cost more than orders you did not win.

Where this comes from: Fifteen people, two continents, and about five hundred dollars a month of tooling that I built myself while running the company. Weekly sales reporting, monthly financials, quotations, invoicing, outreach in four countries, market intelligence, meeting follow-up. Every rule below is here because something broke first, and most of them cost me something before they became rules.

What to build, and in what order

1. Automate the report before you automate the action.

   A report that is wrong wastes your morning. An action that is wrong reaches a customer. Build the read-only things first, live with them for a month, and only then let anything write.

2. Automate what you do every week, not what would impress people.

   The value is frequency multiplied by minutes. A weekly report pays you back fifty times a year. A clever one-off pays you back once and then you maintain it forever.

3. Put the customer-facing thing last.

   Reporting, then internal actions, then anything a customer can see. By the time you get to the last one you will have learned enough to not embarrass yourself with it.

4. Never automate the judgement. Only the work around it.

   The decision is where your experience lives. Hand over the collecting, the formatting, the chasing and the reminding, and give the decision back a clean desk. I still record and edit my own videos, because the one hard cut is the part no tool knows how to make.

5. Know the monthly number and be able to say it out loud.

   Mine is around five hundred dollars. If you cannot say the figure, you have not built leverage. You have built a hobby with a subscription.

What breaks, and what to do about it

• Never retry a send unless you are certain the error arrived before delivery.

  Some services report a failure after the message has already landed. Mine did. The retry sent it twice to real customers and marked working numbers as unreachable. Those contacts do not come back.

• Anything that reports on itself needs a check that does not come from the same tool.

  If the only evidence that the work happened is the thing that did the work, you have no evidence at all.

• Every step you allow to continue past an error needs an alarm on it.

  Continue-on-error is not resilience. It is a silent stop with a green light on top. One password change revoked every mail token I had, and the flows kept reporting success for days.

• Make the queue advance even when a step fails.

  A process that errors before moving its own pointer will pick the same broken record every hour, forever. Classify the failure, mark it, move on. One dead phone number once jammed an entire campaign this way.

• Separate a temporary fault from a permanent one before you act on either.

  Treating a dead account as a slow connection burns your list. Treating a slow connection as a dead account stops your business. The classification is the whole job.

• Assume every credential dies without telling you.

  They expire, they get revoked by a password change you made for an unrelated reason, and they hit quota ceilings on a funded account. Build as if each one will fail silently, because one of them will.