The Cheapest Unit Price on RC and Toy Inventory Is Costing You More Than You Think

The number that looked fine

In Q3 2024 I ran our standard purchasing review. Unit cost across our RC and youth categories was down 6.2% year over year. Gross margin was down 2.1 points.

Same warehouse. Same three pickers. Same freight broker since 2021. Nothing had visibly changed — which is exactly why it took me six weeks to find the leak.

If you'd asked me in September what the problem was, I'd have said freight. Freight is always the answer when you don't want to open the spreadsheet.

Context: procurement manager at a 42-person hobby and toy distributor. Roughly $1.9M in annual purchasing across RC, ride-on, and youth craft, seven years, 60-plus vendors, every order logged in the same cost tracking system since 2018. I've made this mistake twice. The second one is the one worth writing down.

It's tempting to think a lower unit price is a lower cost. It isn't. Unit price is one line in a stack of lines, and the stack is where the money goes.

Four costs that never appear on the quote sheet

1. The MOQ “discount” that isn't

A vendor offered 15% off at 24 units and free freight at 12. I took 24. Twelve of them are still on the shelf (as of January 2025, at least). The 15% saved us about $310. The eleven units that didn't move until we marked them down at 40% off cost us $640 in margin, plus a shelf bay we could have used for a truck that actually turns.

Free freight isn't free. It's a discount on volume you didn't need.

2. Parts support is a line item, even when it isn't invoiced

Nobody buys an ARRMA Gorgon RC car and disappears. They come back. For a spur gear. A hex driver. A gearbox housing. Those are $6–$14 tickets with the same handling cost as a $180 ticket, and if you don't have them, the customer doesn't wait — they order online and stop walking in.

We now stock parts at a deliberate loss on handling and count the whole thing as retention spend. Sounds soft. It isn't: parts buyers convert to vehicle buyers at roughly three times the rate of walk-in traffic in our own numbers.

3. Claims that need verification

“In stock.” “Ships today.” “Fastest.” Suppliers say these things in email as though they were weather.

Per FTC advertising guidance (ftc.gov), claims must be truthful, not misleading, and substantiated with evidence before they run.

Two of our vendors' “in stock” claims failed a spot check in 2024 — the SKU existed, the container didn't.

4. Shipping assumptions with an expiration date

USPS adjusts rates in January and July (usps.com). Our 2024 landed-cost model was still running 2022 figures because nobody reopened the file. That's not fraud. That's just a spreadsheet nobody owned.

What I actually missed

Here's the part I'd rather not put in writing.

We track margin by category. RC is one line. Youth craft is another. Ride-on is a third. For three years those lines looked fine, because the good SKUs were quietly subsidizing the bad ones inside the same category. A hot entry-level truck was carrying the losses on two slow movers, and the category average hid all three. We were making reorder decisions on a number that had been averaged into meaninglessness.

The fix was embarrassingly simple: SKU-level, quarterly, always. Not category. Not annual. SKU.

Second thing I missed — and this one cost more — is that search volume and demand are different animals.

Half our web traffic isn't buyers. It's parents typing “how to make slime without an activator” into Google at 9pm. We saw the traffic, built the landing pages, and bought slime inventory accordingly. Traffic went up. Sales didn't. That searcher wants a free solution, not a $19 kit — she converts only when the kit is physically in the aisle and the kid is already asking.

Same story twice more. A “make market dollhouse” kit looked like a winner because the search data was beautiful; it moved 14 units in nine months and ate a shelf bay. A drifter bike for kids is a spring SKU in our region and dead weight by August — we kept reordering it in September because the spring numbers were good.

The “stock whatever people search for” thinking comes from an era when shelf space was free and returns were rare. Neither of those is true now.

What it cost to keep doing it the old way

• $34,000 in dead stock written down in 2024 — roughly 9% of our average inventory value
• 11% of SKUs sitting past 180 days
• One lost club account because we couldn't produce a $9 gearbox on a Saturday morning. That account was worth about $6,800 a year. Best trade we ever lost.
• Two quarters of margin I can't explain line by line, because we weren't tracking at the level where the answer lived

There's also the part that doesn't show up anywhere. When your fastest-moving SKU is funding three slow ones, you stop trusting your own reorder decisions. You start guessing. Guessing is expensive and it's exhausting, and after about the fourth month of it I was ready to hand the whole category to someone else.

Hindsight is cheap, but here's mine: I should have built the SKU-level model in 2022, when we had eleven active vendors and it would have taken a weekend. At the time, the category numbers looked fine. That's the trap. Everything looks fine at the level where you're not looking.

The fix, briefly

This part is short on purpose. The problem was never the model. It was the level of detail.

• Quarterly TCO per SKU, four lines only: unit cost, landed freight, handling, markdown risk. Anything more and nobody updates it.
• First orders under $500, no minimum. Not charity. A $400 test order is cheaper than a $4,000 mistake, and the small vendors who took our $380 first order in 2019 are the ones we now spend $20,000 a year with. Small doesn't mean unimportant — it means potential.
• Parts availability is part of vendor evaluation, not an afterthought. If a supplier can't support the spur gear and the hex driver, the vehicle doesn't get the shelf space.
• Claims get a spot check before they get a purchase order.
• Reopen the freight model in January and July, every year, on a calendar reminder with a name attached to it.

None of this is clever. It's just the unglamorous work of counting the things that don't appear on the invoice.

The unit price was never the problem. It was just the number I could see.