Sample inventory: salesman samples, PR pulls, and demo units

HARNESS-M-15: 12 units marked available in the system. Storefront sales in five months: zero. They're riding around in a sales rep's car.

HARNESS-L-02: 6 units available. Storefront sales: zero. Mailed to three pet-media PR contacts in June.

HARNESS-XS-01: 4 units available. Storefront sales: zero. Last confirmed sighting: a trade-show demo table in April.

Twenty-two units that will never sell, recorded as sellable stock — an illustrative ledger, and a normal one. Every one of those units inflates your on-hand counts, contaminates your reorder math, and turns into phantom shrink the next time someone counts a shelf. This guide covers the four kinds of never-sell units, where to park them so your numbers stay accurate, a ledger template to run them on, and what they actually cost per year.

Four kinds of units that never sell

• Salesman samples. Units your wholesale rep carries into boutiques to win accounts. You own them until they convert a store or vanish into the rep's trunk — usually both outcomes cost you the unit. (Scoring those samples factory-vs-factory before you commit is the other "sample" problem — the side-by-side evaluation lives in the sample comparison matrix.)
• PR pulls. Units mailed to magazines, newsletters, and creators hoping for coverage. The return on these is editorial, not revenue, and most units never come back.
• Event swag. Giveaways for charity walks, race goodie bags, logo pins handed out at a market stall. They leave as gifts, permanently, often in quantities nobody wrote down.
• Demo units. Trade-show table stock, the harness on the fit-model display, the size run kept in the car for in-person selling. They're working assets, not sellable ones — but the system can't tell the difference unless you make it. This guide covers the general, non-serialized case; a serialized electronics floor runs demos on a stricter clock — conversion logging, a depreciation price ladder, monthly in-place counts — in the electronics store inventory guide.

What they have in common: they leave your sellable pool permanently or near-permanently, without a sale to show for it. That's also what separates them from stock you're holding back deliberately — launch reservations and seasonal holds that will sell later are a different discipline, covered in the launch reserve stock guide. One vertical runs this at scale: sporting-goods stores' demo fleets — boards and kayaks that degrade with every weekend — get their own clock, count cadence, and price ladder in the sporting goods inventory guide.

What not-for-sale stock does to your numbers

• Counts turn them into fake shrink. The shelf count comes up 22 units short, the variance gets logged as missing stock, and someone spends an afternoon hunting for theft that's actually a rep's back seat.
• Reorder math treats them as demand-ready stock. Those 22 phantom units make the available count look deeper than it is, so the reorder trigger arrives later and smaller than it should — the real sellable pile runs dry while the system insists you're covered, and the revenue cost of those dry days is priced in the cost of out-of-stock days.
• Their cost stays invisible. At an average $12.50 cost per harness (illustrative), 22 units is $275 of inventory sitting outside any "what did samples cost us this year" number — money that was spent, isn't sellable, and isn't attributed to anything.

Three ways to park them out of the sellable pool

1. A dedicated samples location. Create a location named "Samples — not for sale," move the units there with a transfer order, and set that location so it doesn't fulfill online orders. Units stay traceable, and storefront availability stops seeing them, as of November 2026 (Shopify Help Center, accessed November 2026).
2. Draft or archived products. For samples of SKUs that don't exist on your storefront at all — a wholesale-only line, a discontinued colorway — a draft product holds the record without ever appearing to sell (Shopify Help Center, accessed November 2026).
3. Tracked out with a reason code. If the SKU must stay live and sellable at your main location, adjust the units out with a cause note naming the purpose and the person: "rep sample — Dana, March." The adjustment history keeps the trail; the ledger below keeps the detail a note can't.

Pick one approach and apply it to everything. Three systems for the same problem is how units end up in none of them.

The sample ledger

One row per hand-off, whether it goes to a rep, a magazine, or a raffle barrel:

Three rules make the ledger hold:

• Nothing leaves without a row. The unit and the record move together; a sample without a row is future fake shrink.
• Quarterly review, written decision. Every "Out" row older than its review date gets decided: returned, re-issued, or written off. Undecided rows are how ledgers die.
• Write-offs get their own cause code. Sample cost should never read as shrink in your variance history, or you'll lose both numbers at once.

PR pulls: run them like a loan, not a gift

PR pulls deserve their own workflow because they're the stream most likely to be tracked as "sent, whatever happened next." Run them like a loan library:

• Pitch first, pull second. Units go out after an editor or creator says they're interested. An unsolicited box is a gift with extra steps — it burns stock at $12.50 a unit (illustrative) with no coverage date attached, and most unsolicited product never gets opened.
• Photograph each unit at pack-out. Condition proof at the moment it left. If a unit comes back scuffed, the photo settles which leg of shipping did it; if it never comes back, the photo is what the write-off row references.
• Ship tracked, log the row the same day. SKU, units, outlet, sent date — the ledger columns above — plus a deadline: the outlet's stated editorial lead time plus two weeks. Tracking number goes in the row, not in your inbox search.
• Follow up once, at the deadline. One note: did it land, is coverage scheduled? A feature that runs gets linked in the row's status — that's the return the PR pull was actually buying.
• Decide at deadline plus two weeks. Returned and re-issued, kept for a re-pitch, or written off. An "Out" row with no decision past that point is a unit you've mentally spent but never recorded — the ledger equivalent of the rep's trunk.

None of this changes how often media keeps product; it changes what a non-return means. Instead of a unit that quietly vanishes from both the shelf and the books, every dead pull becomes a decided row with a date — and your annual samples number stops hiding write-offs inside shrink.

What samples actually cost per year

Worked example for the illustrative dog-harness brand above:

Of that $1,070 (illustrative), assume 30% of rep samples and PR pulls never return — set the rate from your own history — and roughly $240 a year (estimated) becomes write-off rather than working sample. The point isn't that $1,070 is a crisis. It's that most owners can't currently produce this number at all, because the units sit scattered across "available" stock, a rep's car, and three magazine mailrooms — and unattributed cost is cost you'll repeat every year.

One tax note worth ten seconds: the IRS caps deductible business gifts at $25 per recipient per year (IRS Publication 535, Business Expenses, accessed November 2026), a cap that gifts to press and creators brush against more often than people expect. Samples sent for review or demonstration may be treated differently, so ask your accountant before assuming the whole PR pull is an ordinary expense.

Next step

Today, run one filter: SKUs with no sales in 90+ days, checked against reality. Some of what you find will be samples and demos that get ledger rows and a transfer out of the sellable pool this week — and some will be genuinely dead stock, where the disposal options are covered in the dead stock clearance guide. Units you actually want to hold back for a future launch are the different pool with different rules in the launch reserve stock guide.