Vendor in the first ninety days. What I am actually looking for.
A new vendor walks into the room with a swatch book in one hand and a costed line sheet in the other. Three months later one of three things has happened. They have become a vendor my team buys from on autopilot. They have become a vendor we are slowly winding down. Or they have become a meeting we keep rescheduling.
The third outcome is the most common one and the most expensive. A vendor we are not buying from and not winding down is a vendor occupying the meeting calendar without paying for the seat. The first ninety days exist to make sure we get to the first outcome or the second one fast, and not the third.
This essay is about what those ninety days look like from the buyer's side of the table. It is for vendors who want a shorter feedback loop, and for buyers who haven't yet codified what they are actually looking for.
What the ninety days are for
A new vendor on a new line is roughly six bets stacked on top of each other. Bet on the costing. Bet on the fabric base. Bet on the dye route. Bet on the cut and fall. Bet on the delivery confidence. Bet on the team's professional behaviour over time. Each bet has its own failure modes. Each one breaks in a way that is hard to predict from the line sheet.
The job of the ninety days is to convert six bets into six observations. Not to validate the vendor. Not to "give them a chance." Convert bets to observations. After ninety days, every one of those six should have moved from we think to we have seen.
The buyer who treats the ninety days as a relationship-building window is the buyer who is still flying blind on day ninety-one. The buyer who treats it as a forced-march evidence collection is the buyer who can write the open-to-buy for the next season with confidence.
The first conversation
I have a small set of questions I ask in the first meeting. None of them are about the vendor's history. The history is on the deck. The questions I want answered are about how the vendor operates today, when nobody is watching.
How long does it take you to produce a fit sample once we send a tech pack? The vendor who answers in days is being honest. The vendor who answers in business days is hedging. The vendor who answers with a range is the most honest of all. The vendor who answers with a single number under three days is either lying or has a sample-room I want to visit.
How do you handle a vendor who is late? This is the most important question I ask. The vendor who has good answers is a vendor who has been late and has learned. The vendor who says they are never late has either not been in business long enough or is not telling me the truth. The vendor who says "we don't take orders we can't deliver" is the vendor I want to work with. That vendor has thought about their own constraints, and that thought will save us a season.
Tell me about a recent quality call you made that cost you money. This question gets to whether the vendor has a quality function or just a quality team. A vendor without a quality function will tell me a story about a thread count or a fit issue that got resolved at the buyer's expense. A vendor with a quality function will tell me a story about a delivery they refused to ship, an order they recut at their own cost, a swatch they sent back even though the dyehouse signed off.
What is your worst category for us, and what do you do about it? The vendor who can name their worst category in the first meeting is the vendor who knows their own range. The vendor who claims to be good at everything is either new or wrong.
Four questions. Twenty minutes. By the end of the conversation I know more than I would know from a six-month commercial relationship with most vendors.
The first order
The first order is a small one. Anyone who has been in retail knows this, and most of us run it anyway. The first order is the diagnostic. It is not the relationship. It is not the bet. It is the test.
I write the first order with three things in mind.
One: the order has to be large enough that the vendor takes it seriously. A token order produces a token outcome. We learn nothing.
Two: the order has to be small enough that an absolute failure does not break my open-to-buy. I want to find out a vendor cannot deliver. I do not want to find out by losing the season.
Three: the order has to include at least one piece of complexity. A buyer who orders only the simplest possible SKU from a new vendor has bought the vendor's best-day performance. I want the average-day performance. So I add a fabric they have not used, a wash they have not done, a finishing they have to outsource. The complexity is deliberate.
The vendor who delivers a clean first order is fine. The vendor who flags an issue in the first week and proposes a fix is better. The vendor who delivers on time and badly is the one I learn the most from, because they have shown me what they will do under pressure. The vendor who delivers late and well is teaching me that their pipeline runs hot and their schedule cannot be trusted.
The first order is a diagnostic. It is not a referendum.
The first quality call
Somewhere between day forty and day sixty, a quality issue surfaces. The first one always does. It is built into the system. The interesting question is what the vendor does about it.
There are five categories of vendor response, in roughly decreasing order of how much I want to work with them.
The fixer. Hears about the issue, calls back within the hour, sends a person to the warehouse, takes the goods back, reworks them, ships within the original window. This vendor has just told me everything I needed to know.
The owner. Hears about the issue, acknowledges fault, proposes a credit + a rework on the next order, asks what we need today. This is also fine. Less heroic than the fixer, but more sustainable.
The negotiator. Hears about the issue, asks for photos, asks if we are sure, suggests we accept a partial discount and move on. This vendor has just told me that quality is a number in a spreadsheet to them, not a craft. Not the worst, but not the future.
The lawyer. Hears about the issue, points to the spec, points to the sample, suggests our QC is being too strict. This vendor I will finish the open order with and then quietly stop calling.
The ghost. Hears about the issue, doesn't reply, lets it sit until the goods are at the store and the markdown ledger is the only conversation. This vendor I am done with.
The lesson is not that I never use a negotiator or a lawyer. They have their place, especially on low-margin commodity lines where the brand can absorb the friction. The lesson is that the vendor's response to the first quality call is a strong predictor of their behaviour over the next three years, and most buyers spend longer than they should figuring out which category the vendor falls into.
The first re-order
Day seventy or so, I am writing the second order. By now I have data. The fabric did or did not arrive on the day they said. The cut hit the rate or it didn't. The fits passed at first round or required a second round. The first store delivery happened cleanly or required a re-run. None of these are subjective.
The re-order is the moment I treat the vendor as a vendor, not as a candidate. The first order taught me what they could do under one set of conditions. The re-order tests whether the conditions were the variable.
I usually structure the re-order in one of two ways.
Same brief, larger volume. If the first order went well, I want to know if the vendor can do it under load. A vendor that runs clean on three thousand units may or may not run clean on twelve. The next order tells me which.
Different brief, similar volume. If the first order went well and I do not need more units of the same SKU yet, I write a brief in a category they did not show me on the first round. A vendor who is excellent at structured drape may or may not be excellent at jersey. The next order tells me which.
The re-order is also where I find out whether the vendor has institutional memory. Did they save the fit pattern? Did they remember the trim spec? Did they keep the QC manager on our account? A vendor with institutional memory pays compounding returns. A vendor without it costs me a buyer's-team week of re-onboarding on every order.
Day ninety
By day ninety the file should be thick. I have a real picture, not a hopeful one.
I sit down and write three lines.
Are they someone I can buy from on autopilot this season? Yes / no / not yet.
What is the risk I am carrying with this vendor that I would tell my successor about? One sentence.
What is the single thing they do better than any other vendor on this line? One sentence.
If I can answer the first question yes, the vendor has joined the autopilot list and the next conversation we have is about the next season. If I can answer it no with the second and third lines clear, the vendor has either moved into the "specific use cases" bucket or onto the wind-down list, and I know which.
If I find I cannot answer at all, that is a signal too. It usually means I did not write the first order with enough diagnostic intent. Sometimes the next ninety days fixes it. Sometimes I have to acknowledge that I gave a vendor a meeting calendar without a job to do.
Why this matters more than people think
A retail buyer evaluates vendors all the time. There is nothing original about the framework above. What I think is original is the discipline of doing it on a clock.
Most buying organisations let a new vendor coast for six months. Some let them coast for a year. The result is a vendor list that gets longer every season, with three or four real producers in the middle and a long tail of vendors who attend the line meetings, take up planning bandwidth, and contribute nothing to the open-to-buy. The buyer who is willing to call the question on day ninety has a tighter list, a deeper relationship with the vendors who survive, and significantly less wasted calendar.
This is also where the systems I have written feed back into the human work. The vendor decision engine I described in the previous letter weighs fabric, delivery, dye route, and a private quality history. It does not replace the four questions above. It compresses what I would otherwise be reconstructing in my head from spreadsheets. The first conversation still has to happen. The first order still has to teach. The first quality call still has to be made. The system just makes sure I am not surprised by what the data already knew.
Closing
The first ninety days are the cheapest evidence a buyer ever buys. Six bets become six observations. The vendor either earns the autopilot or doesn't.
Vendors who read this and want to know what they should do: be the fixer, not the lawyer. Tell the truth about your worst category. Send a person to the warehouse the day a quality call lands. Save the fit pattern.
Buyers who read this and want to start: pick the next new vendor you onboard and write the three lines on day ninety. The discipline compounds.
Always up for a conversation about a vendor who is doing it right, or a vendor who is doing it wrong, or any other thing on the floor.