How to Validate a Product Before Buying Inventory

The first large inventory order is the moment an uncertain assumption turns into an expensive physical problem. Until then, a product idea is a sketch, a supplier quote and a feeling. Afterwards it is boxes in a spare room, cash out of your account, and a deadline you did not choose.

That is what makes this decision different from most early business decisions. A landing page can be rewritten. A launch date can move. A pallet of finished goods cannot be unmade, and a manufacturer's deposit is rarely refundable. The order is the least reversible thing most physical-product founders do in their first year.

This guide is about the evidence that should exist before that money moves — how to test price, demand, channel and margin while a wrong answer is still cheap, and how to decide whether the order you are about to place is justified, too large, or premature.

Almost every product that ends up unsold in a garage had encouraging feedback behind it. That is not because founders are gullible. It is because the feedback that arrives first is the cheapest kind to give.

Consider what each of these signals actually costs the person giving it:

None of this means feedback is worthless. Early reactions tell you whether the product is understood, whether the positioning lands, and which objections come up first. What they cannot tell you is whether enough people will pay your actual price through a channel you can afford. That is a different question, and it needs a different kind of test — one where the other person gives up something real.

A stock order looks like a purchasing decision. It is really a bet placed simultaneously on nine or ten separate assumptions, each of which can fail on its own. Writing them out individually is uncomfortable, which is precisely why it is useful.

Framed this way, the order stops being a single yes-or-no and becomes a portfolio of risks. Some of these assumptions are cheap to test this week. Some can only be tested with real buyers. And some — supplier consistency, for instance — can only be reduced, never eliminated. Knowing which is which is most of the work.

The practical exercise: write all nine on one page, and next to each one write the evidence you currently have. Most founders discover they have strong evidence for two, weak evidence for three, and nothing at all for the rest. The gaps are your test list, in priority order, starting with the assumption that would be most expensive to be wrong about.

Price is the assumption most often skipped, because testing it feels premature when the product does not exist yet. It is also the assumption that quietly determines whether every other number works. A product that sells beautifully at £24 and needs £40 to be viable is not a demand problem — it is an already-solved business that happens to lose money.

You can test price before a single unit is manufactured:

A pre-order at the intended selling price, with a clear and honest delivery window, is the strongest pre-manufacturing signal available. The buyer sees the actual number, actual delivery date and actual product description, and pays anyway.

Where a full pre-order is too big an ask — high-ticket items, long lead times — a refundable deposit still requires a card, a decision and a small amount of trust. It is weaker than a full payment, but far stronger than an email address.

For B2B or hospitality products, a paid reservation of the first production run, or a paid pilot with one venue, converts a friendly conversation into a commercial one. Ask for a purchase order, not enthusiasm.

A single product page with real photography or credible renders, honest copy, the real price, delivery cost and a working checkout tells you the conversion rate from qualified traffic. If you are not yet able to fulfil, say so on the page and set the delivery expectation explicitly — never take money while implying stock exists.

Where a supplier will produce fifty units at a worse unit price, that batch buys you real sales data, real returns data and real reviews. It is the most expensive test on this list and usually the most informative.

Not all evidence is equal, and the gap between adjacent rungs is often much larger than it looks. Ranked from weakest to strongest:

Two rungs down the ladder does not mean half as good. Moving from email signups to paid pre-orders typically removes the large majority of an audience, which is exactly the point: the people who remain are the only ones your inventory maths can rely on.

Product demand and channel viability are separate assumptions, and confusing them is one of the most common ways good products become bad inventory. People wanting the thing does not mean you can profitably reach them.

The test is not whether a channel can produce a sale. It is whether it can produce sales repeatedly, at a cost per order that leaves contribution behind. Run a small, genuinely cold test on the one channel your plan depends on most before the order, and record the cost per checkout, not just the click-through rate.

Gross margin is the number most founders quote and the least useful one at this stage. It flatters the decision because it ignores almost everything that actually consumes the money: fees, fulfilment, delivery subsidies, returns and acquisition. Contribution margin — what is left from one order after all the costs that vary with that order — is the number that determines whether the stock turns into cash or into shelving.

Here is an illustrative structure. Every number below is invented for the purpose of the example. They are not benchmarks, and your own figures will differ substantially by category, channel and country.

Total variable cost per order: £34.00. Contribution per order: £6.00 — fifteen per cent of the selling price, on a product whose gross margin looks like a comfortable seventy-one per cent before the other costs appear.

Now put the MOQ against it. Suppose the factory's minimum is 1,000 units at £9.20 ex-works, and landed cost brings that to £11.50 per unit:

The more immediate risk is not accounting profit; it is cash sitting in stock that has not sold. To see how quickly the £11,500 comes back, take the landed inventory cost out of the variable costs, because that money has already left the bank. The non-inventory variable costs are £22.50 per completed order (£34.00 − £11.50), so each sale returns £40.00 − £22.50 = £17.50 of cash toward the original outlay. On that basis, £11,500 ÷ £17.50 ≈ 657 sales are needed to recover the inventory cash committed.

That is the shape of the problem the gross margin hid. There are three levers: raise the price, reduce acquisition cost, or reduce the order. All three should be tested before the deposit is paid, because afterwards only the first two remain — and both take longer than your cash allows.

Factories set minimums for rational reasons: tooling, machine setup, material purchase lots, changeover time and the administrative cost of a small customer. A run of 1,000 is not greed — it is often the smallest quantity at which their process is worth operating.

The important point is that the manufacturer's efficient order size and your safe order size are different numbers, calculated from different risks. Theirs is a production question. Yours is a survival question. When you accept their number without negotiation, you have let a stranger's cost structure set your cash exposure.

Ask for the reorder lead time before you ask for a price. A supplier who can reproduce your order in six weeks makes a small first batch sensible. One who needs sixteen weeks turns success into a stockout and hands your early demand to a competitor — that constraint should shape the size of the first order more than the unit price does.

Pre-orders are the most efficient way to convert an inventory bet into funded demand, and the easiest way to damage a new brand if handled carelessly. The difference is entirely in the disclosure.

A deliberately small first batch can be rational even when the unit economics are temporarily worse. You are not buying inventory at that point; you are buying information — real returns data, real reviews, real repeat rates, and proof the supplier can hold quality. Treat the unit-cost premium as the price of that information and compare it against the cost of being wrong at full scale. Under the illustrative numbers above, a 200-unit run at a 25% cost premium risks roughly £2,880 instead of £11,500.

The following example is illustrative. It is not a real company, a customer or a documented case study, and the numbers were chosen to demonstrate the method rather than to prove a conclusion.

A founder plans a premium insulated lunch container for commuters: stainless steel, leak-proof, six-hour heat retention, sold direct through a Shopify store at £40. The supplier MOQ is 1,000 units at a landed cost of £11.50, requiring £11,500 in total, £3,450 of it on deposit.

Under the surface, the order assumes commuters will pay £40 for a category where £18 alternatives are widely available; that the waitlist will convert; that direct-to-consumer acquisition is affordable at that price; that returns on a container with a leak claim stay low; and that the café interest converts into a purchase order rather than a polite conversation.

Before committing, the founder spends two weeks and roughly £600: a single product page with the real £40 price and honest 'ships in 8 weeks' pre-order copy, a small cold paid test on one channel, and an email to the 900-person waitlist showing the price for the first time.

Twenty-three paid pre-orders at full price is genuine evidence — the product sells at £40 to people who have seen the price. But £67 to acquire a cold customer against roughly £6 of contribution is not a business; it is a subsidy. The warm list converted at 1.6%, which on a 900-person list is a one-off, not a channel. And the original 1,000-unit order would tie £11,500 of cash into stock that needs roughly 657 sales just to return the inventory cash committed.

The supplier will run 200 units at £14.40 landed — a 25% unit premium, £2,880 committed instead of £11,500. Twenty-three units are already sold. The remaining 177 need a channel that works, and the founder now knows cold paid traffic is not it at this price.

Before spending the £600, the founder had written down: proceed with the full order only if the cold test produces an acquisition cost below £15, and the waitlist converts above 4%. Both were missed, and by a wide margin.

Not 'stop', and not 'go'. The honest reading is reduce and pivot the channel: place the 200-unit batch, fulfil the 23 pre-orders, and spend the next quarter testing marketplace and creator channels — where the category already has search demand — rather than buying cold traffic. The full 1,000-unit order is deferred until one channel produces orders below £15 acquisition cost. If no channel does within the quarter, the deposit was never paid and the loss is £3,480, not £11,500.

Any one of these is a reason to pause and test further. Two or more together, and the order is almost certainly too large or too early.

Kill criteria are thresholds you write down before you are emotionally or financially invested. Each one needs a number, a timebox and a decision attached — without all three, it is a hope, and hopes are infinitely revisable.

The following are illustrative examples, not universal thresholds. The right numbers depend on your price, category, cash position and how much of your capital the order represents.

That is the whole point of writing them early. Once £3,450 is with the factory, a 1.6% conversion rate stops being a red flag and starts being 'early days'. The evidence has not changed; your relationship with it has.

Framing this as launch or do-not-launch throws away the two answers that are usually correct. There are four:

Justified when full-price purchases exist at volume, acquisition cost sits comfortably below contribution, the supplier has demonstrated quality, and the order represents a share of your cash you could genuinely afford to lose.

The most common right answer. Real buyers exist but one assumption — usually channel economics or return rate — is untested. Pay the unit-cost premium, get real sales and real returns data, and place the larger order with evidence behind it.

When the product sells but the economics do not. Raise the price and test whether demand survives; change the channel to one where the category already has demand; bundle to increase order value; or reposition to a buyer with a larger budget.

When repeated tests at the real price produce no paid buyers, or the only viable economics require assumptions you have no evidence for. Stopping at this stage costs a test budget. Stopping after the order costs the order.

Failure Forecast cannot prove that inventory will sell. Nothing can, before it does. Real buyer behaviour — a deposit, a pre-order, a full-price purchase from a stranger — is the only thing that supplies that evidence, and this guide exists to help you collect it.

What a forecast does is expose the demand, price, channel, margin and execution assumptions that should be tested before the order is placed. Specifically, it helps you work out: