On September 11, 2024, our marketing director walked into my office, dropped a folder on the desk, and said six words: "Spring capsule. One hero SKU. Mirror."

I'm the procurement manager at a 40-person beauty retailer. Six years in this seat, roughly $1.1 million in annual sourcing spend, and a standing rule that anything over $2,000 gets three quotes minimum plus a TCO spreadsheet. I've negotiated with more than 60 vendors and every order since 2018 lives in our cost tracking system. The folder contained a spec sheet for a lighted makeup mirror — 9-inch vanity style, dimmable LED, USB-C and wall plug — plus a note that if the test sold, we'd add a wall-mounted version for the second drop.

Four hundred units. Small. That was the entire point — a trial run to find out whether the category moved before we committed real money to it.

What followed over the next seven weeks was the most useful lighted makeup mirror sourcing exercise I've done. Not because it went smoothly. Because it didn't.

Nine emails. Four replies. Three walls.

I sent the RFQ on September 13 to nine suppliers — three I'd used before, six pulled from a sourcing platform and two trade directories. Nothing exotic in the ask: 400 units, ETL listing for US retail, samples by November 1, delivery to our New Jersey 3PL by January 15.

Four came back inside 48 hours. Three said some version of the same thing: 1,000-unit MOQ, take it or leave it. One of those three was polite about it and offered to hold pricing for 90 days. The other two just sent a price list and stopped replying.

Not ideal. Workable, I told myself. I could buy 1,000 and carry 600 in inventory. I ran the carrying cost — about $0.85 per unit per month in our 3PL, six months average before sell-through — and quietly added $3,060 to the "cheap" option that was starting to look not-so-cheap.

The remaining two quotes were where it got interesting.

The quote that looked best on the first line

Supplier A — a mid-size factory rep I'd never used — quoted $18.20 per unit for 400 pieces. Third line down, in 9-point font, was an ETL certification fee of $1,200 that "may apply." Plus $65 per testing sample (two required). Plus a $300 artwork setup charge for the retail box. None of it mentioned in the cover email.

Supplier C — a smaller trading firm, twelve employees, a website that last got a design refresh in 2019 — quoted $17.40 per unit at 400 pieces, all-in. ETL handled by them, lab fee passed through at cost ($1,100), one pre-production sample free, retail box artwork included.

On the sticker price, A was 4.6% more expensive than C. On total cost of ownership, A ran $8,910 versus C's $8,060. That's 10.5% — for a 400-unit order, $850. Small in absolute terms. Larger as a percentage than anything else in my spreadsheet that quarter.

The surprise wasn't the size of the gap. It was that A's rep had told me on the phone that the quote was "everything included." The lesson I keep relearning: the number in the first line of a quote is a price, not a cost (i.e., it's a starting point, not the finish line).

Ten days, one decision, three suppliers

Here's where the timeline became a problem. To hit a November 1 sample date and a January 15 landing, I needed to place the order by October 4. Normally I'd run a second round of quotes, request two more samples, and take four weeks to decide. There was no four weeks.

The marketing director needed the sample for a catalog shoot booked on November 6. Our photographer's cancellation fee alone was $1,800 — more than the entire price difference between my two finalists.

Looking back, I should have pushed back on the shoot date when the capsule was first discussed in August. At the time, the November 1 sample deadline looked generous. It wasn't, once two of my original vendors dropped out of contention in early September.

So I made the call with incomplete information. Went with Supplier C — the small firm — based on two signals: they were the only one who asked what our returns data looked like on a comparable SKU, and they were the only one who sent a spec sheet back with margin notes in a different color. Someone had actually read the RFQ.

I won't pretend that's a rigorous sourcing methodology. It isn't. But with ten days and a hard deadline, the quality of the questions a supplier asks is often the only signal you have time to evaluate.

What the small supplier caught that the big ones didn't

Three things, and all three ended up mattering.

First: magnification. The brief called for a lighted makeup mirror with magnification — 10x, marketing said, because that's what the category leader used. C's rep came back with a note: in their experience, 10x on a 9-inch vanity mirror generated adjustment-level returns because customers found it distorted at close distance. They proposed 5x on the vanity and kept 10x only on the wall-mounted SKU, where customers tend to stand farther back. We pulled a Fascinate lighted makeup mirror as a comparison sample during that same week, and their rep confirmed the same thing — fixed 10x is a returns driver, and their landed cost ran about 40% above our target anyway.

We went with 5x on the vanity. Return rate on that SKU through the first 60 days: 1.4%, against a 4% category average in our own historical data.

Second: the wall mount bracket. A wall mounted lighted makeup mirror is a different animal from a vanity version. The four suppliers who quoted the wall mount all specified hardwired installation — a junction box, an electrician, $150–$250 of additional install cost for the customer. C asked whether we'd consider a plug-in adapter with a cord cover, so the lighted makeup mirror wall mount could hang on drywall without an electrician. That wasn't in our original brief. It ended up being the feature that sold the SKU.

Third: packaging. C's $0 artwork-included offer wasn't a sales tactic. Their account manager sent back a retail-ready box design in six days, with two rounds of revisions, for a 400-unit order. (Note to self: put the ETL requirement and box artwork scope into the RFQ template itself, not into the follow-up email that only half the vendors read.)

Where the numbers landed

Samples arrived October 28. Shoot happened November 6. Order landed at our 3PL on January 9 — six days early. Total landed cost on the 400-unit trial: $8,060, or $20.15 per unit including the lab fee.

We sold through 312 units in the first 11 weeks. Not a breakout. Solid.

The wall mount version was the one that surprised us. We ordered 1,200 units in the second drop as a test, and the wall mounted lighted makeup mirror outsold the vanity version almost three to one in the first six weeks — mostly to customers who lived in rentals and couldn't install a junction box. That's a durable insight about the category that none of the four big suppliers would have surfaced on a 400-unit order.

Total spend with Supplier C across six months: $8,060 on the trial, then $71,000 on the February reorder of 4,000 units across two SKUs. They're now one of three suppliers on our standing list for the beauty-tools category.

What I'd do differently, and what I'd keep

If I could redo the September decision, I'd write the ETL listing requirement into the RFQ's first paragraph instead of the second. Two of the nine suppliers who responded didn't know what "ETL listed" meant, and one of them took a week to find out — a week I didn't have. That's a template problem, not a vendor problem.

I'd also keep a sixth vendor in reserve. Nine emails sounds like enough coverage until five of them turn out to be MOQ walls or non-responsive.

What I'd keep is the small-supplier shortlist. Our procurement policy has a paragraph about diversifying the vendor base that I mostly skimmed for years. Now I don't. Here's the argument I'd make to any buyer who's been told to skip the little guys: I'm not saying a 400-unit order should cost the same per unit as a 4,000-unit order. It shouldn't — setup costs are real, tooling is real, and every factory has to amortize them. What I am saying is that the quality of service shouldn't scale down with the order size. A trial order is how a relationship gets tested. Suppliers who treat 400 units like a real order are the ones who get called back for 4,000.

The vendors who wouldn't take our 400 units? Two of them followed up in February, when the reorder went out to bid. I sent both a polite note. The 1,000-unit MOQ on the original trial wasn't the problem — that's a legitimate business constraint, and I respect it. The problem was that nobody picked up the phone to explain why, or to ask what a smaller starting point might look like.

Small doesn't mean unimportant. It means unproven. And everyone is unproven at some point.