How Ecommerce Teams Make Safer Inventory Decisions

Plan for forward cover, not months of stock.

I see months of stock calculated incorrectly on a regular basis. I recently reviewed a plan where the investor was told they had 20 months of stock because it was 220k of stock divided by only 11k of sales in January. Those 20 months were mathematically correct but strategically useless because January was an anomalous off month for that business.

I prefer to calculate forward cover as the amount of stock we need to ensure we don't run out before the stock we order today can be replenished. If it takes eight weeks to get stock, I want enough to cover December and January, not just December.

The formula I use is (stock to sell to the end of the cover period + backorders + target closing stock - current stock - incoming stock).

When calculating velocity, I also increase sales by the number of days out of stock. If a SKU sold 90 units over 15 days in stock, I calculate its velocity as 6 units in stock, not 3 units over the 31 days in the month. Otherwise, you end up in a self-perpetuating cycle of stockouts suppressing sales figures.

Spend your limited cash on the stock that makes the most money.

I don't allocate stock purchase budget based on number of units or total sales. Instead, I calculate the projected lost profit from a stockout and divide that by the cost to replenish that stock. Not surprisingly, you usually make more profit on more expensive units.

I maintain a backlog of out of stock SKUs by estimated stockout date and calculate the lost profit from being out of stock until I can get more. I then only allocate my limited purchasing cash to the SKUs where that can be converted to the most margin.

If spending 4k on a PO yields 10k of contribution margin but spending 4k on a different PO only yields 800 of contribution margin, I'm going to buy the 10k of margin first. I'm happy to run an out of stock on the second item to hold onto that cash for a better opportunity.

I've been buying stock for a physical-product business for seven years, and I've spent several of those years getting it wrong in both directions.

What changed my thinking is that forecast accuracy stopped being the goal. You can't predict demand for a new line, and no amount of modelling fixes that. What you can control is how wrong you can afford to be, which is a different question with an actual answer.

So I buy to a number I'd be comfortable writing off rather than a number I expect to sell. For anything unproven, that's usually smaller than the supplier's minimum, which means paying more per unit and being fine about it.

The habit is checking the reorder lead time before the quantity. If I can restock in three weeks, I can commit to almost nothing. If it's four months, I'm making a bet on whatever I do.

I learned this after ordering 200 units in August 2025 that did not sell. I stopped treating a short sales spike as proof that demand would continue. At FARUZO, I moved to made-to-order production and only pre-buy items after they have sold steadily for months. My planning habit is to review demand item by item before committing cash, rather than buying a broad batch because one period looked good. That keeps cash out of speculative inventory while repeated sales show which pieces are safer to buy ahead.

Ordering based on gut feeling about what might sell is probably the most expensive habit in ecommerce. Every unit sitting in a warehouse is cash you can't spend on ads, new product tests, or supplier negotiations. And when you guess wrong in the other direction, stockouts send your customers to someone else.

I run smaller initial orders and treat them like paid research. I buy enough to cover two to three weeks of projected sales, watch how fast those units move, then reorder based on real velocity data instead of a forecast. It costs a bit more per unit because I'm not buying in bulk, but I'm not parking thousands of dollars in product that might take months to clear.

The planning habit that keeps this from falling apart is a weekly reorder trigger. Every week I compare current stock against the rolling average sell-through rate from the previous few weeks. If inventory drops below a threshold tied to my supplier's lead time, the reorder goes in automatically. There's no waiting for a monthly review and no scrambling when something suddenly picks up.

I decide stock by using a short rolling forecast tied to small test production runs and clear safety-stock rules: forecast demand for the next 60-90 days, adjust weekly for actual sales velocity and upcoming marketing events, and only commit beyond that when tests prove sustained demand. For items with high uncertainty we use made-to-order or pre-order windows to avoid tying up cash while keeping popular pieces available.

A six-month forecasting project at MusaArtGallery cut unsold inventory by 28% and improved margins while keeping best-sellers in stock.

With modular retail shelving, a stockout isn't always about a complete shelving unit. Sometimes, one missing bracket or shelf can delay an entire store fit-out.

I recommend forecasting inventory around complete, usable systems rather than looking at individual products in isolation. Core components that customers regularly purchase together deserve priority over less common sizes or finishes.

A particularly useful planning habit is reviewing confirmed orders alongside outstanding quotations and upcoming store-opening dates. Quotations can signal future demand before it appears in sales figures.

The goal is to keep the components customers actually need available without tying up working capital in products that may sit untouched for months.

Commit to depth only after you've earned it. When I work with growing fashion and leather brands, I advise splitting the first order into a small "proof run" and a reserved second run. For example, instead of 600 jackets, order 150 to 200 across your core sizes and colors, and agree with the factory in advance that the hide, trims and a production slot are held for a repeat order within 4 to 6 weeks. You only pay for what you produce, but you've cut your reorder lead time roughly in half.

The planning habit that saves the most cash is reviewing sell-through every week by size, not just by style. Most stockouts aren't "the jacket sold out", they're "medium sold out while XXL sat untouched." When a size passes 60% sell-through in the first three weeks, we trigger the reserved run for that size only. That keeps bestsellers in stock without locking money into slow sizes.

We review exceptions each morning instead of checking the assortment on a schedule. The report flags demand changes, falling days of supply, rising search interest, and gaps between forecasted and actual performance. This helps us focus on areas where a decision is needed rather than spending time on items performing normally. It also gives us a view of where attention matters most.

The value is speed, not more meetings. A short daily review creates an early warning system before a strong item becomes a problem. It also shows when an increase is temporary and does not justify a commitment. In seasonal retail, these checks help us respond early and avoid turning changes into missed opportunities.

I size the first buy to the demand I can already see and let live signals decide the rest. Search volume, email click rates, add-to-cart counts and waitlist signups all show interest before a single unit ships. Soft numbers keep the first order small. Rising numbers pull the reorder forward.

The habit that matters most is sharing the marketing calendar with whoever places the purchase orders. The setup I keep running into is an ad budget that doubles on a Tuesday while the stock count was set a month earlier. Ads push hardest on the products that already convert, so those are the ones that sell out first. A stockout mid-campaign means paying for clicks that land on a sold-out page.

So I look weekly at days of stock left on the top sellers against planned ad spend for the next two weeks. When days of stock drop below the supplier's lead time, that triggers the reorder. Cash stays free for slower products because nothing gets bought on a hunch.

Check the last twelve months of sales for each product, not just the last three, since recent spikes get mistaken for a new normal constantly. Had a product sell out twice in one holiday season, lost close to $4,000 in missed sales both times, because the reorder number was based on the previous month's unusually slow stretch, not the full-year pattern. Started calculating a reorder point using the product's own average weekly sales over a full year, plus a buffer sized to its specific demand variability, not one flat buffer applied across every item. A steady seller gets a smaller buffer. An item that spikes unpredictably gets a bigger one, even if its average sales look similar on paper. Stockouts on sellers dropped from six incidents in one year to just one the next. Cash that was tied up in moving inventory also went down by about 15%. This happened because the steady sellers were no longer required to carry the large buffer stock that was once kept for unpredictable or uncertain items. What really made a difference was matching the buffer size to how volatile each product was, instead of applying the same rule to every item in the catalog.

Hi, I'm Charles Liu, founder and Marketing Director of Cubic Promote, a Sydney-based business employing around 30 people across Australia, the Philippines, Vietnam, and India. For us, the safest approach is to separate predictable demand from the products that are harder to forecast. We already know that best-sellers such as pens, umbrellas, tote bags, drink bottles and T-shirts move consistently, so we are more comfortable securing stock for those. We also prepare more heavily around predictable periods such as EOFY and the Christmas season in Australia, when demand usually increases. The harder part is unexpected product enquiries. If a client shows genuine interest in an item and we think the order is likely to proceed, we try to secure or reserve the stock as early as possible rather than waiting for the purchase order. That habit has helped us avoid losing an order simply because the client needed a few days to approve quantities, artwork or budget and the stock disappeared in the meantime. The balance is not stocking everything. We commit more confidently to proven best-sellers and seasonal demand, then use live client enquiries as an early signal for less predictable items. That helps us protect availability without tying up too much cash in products that may not move.

We size the commitment off cash recovery speed, not off the forecast. The question we ask per SKU is how many weeks until this purchase order turns back into cash, counting production, freight, Amazon receive time, the sell through period and then the disbursement cycle. A product sitting on a fourteen week cash cycle has to earn a much smaller commitment than one at six weeks, even when the two forecast identically.

The habit that saved us the most is ordering in tranches with a real trigger instead of a calendar. We split a buy into an initial quantity plus a pre agreed top up with the supplier, and the top up fires on observed sell through in the first two or three weeks rather than on a date. That costs a little per unit in lost volume pricing. It is cheap next to parking cash in slow units you then pay monthly storage on.

The second habit is ranking SKUs by cash at risk rather than by revenue. The top revenue line is usually not the one that hurts you. The dangerous SKU is the mid tier product with a long lead time and a bulky footprint, because it absorbs storage fees and drifts toward the aged inventory surcharge quietly while everyone watches the bestseller.

In practice we hold deeper cover only where the organic rank is hard to rebuild. Everywhere else I would rather be slightly short and reorder than be right about the forecast and out of cash.

A demand forecast should include the cost of waiting. I ask whether delaying a purchase by one replenishment cycle would create a loss that exceeds the carrying cost of buying now. That framing avoids the false choice between aggressive availability and excessive caution. It turns inventory into a capital-allocation question.

The habit is maintaining a list of substitute actions for constrained items, such as bundling, shifting marketing, or later delivery promises. These options reduce the pressure to solve every demand surprise with more stock. Flexibility has financial value, particularly when a business cannot see ahead to justify an irreversible bet.

Inventory planning improves when we focus on dependencies, not just items. A cabinet may sell steadily, yet availability depends on the top, sink setup, hardware, packaging, and freight arriving together. We have seen strong demand turn into a stockout because one supporting component was overlooked. That risk is easy to miss when items are tracked separately.

We map the complete sellable combination instead of counting components alone. We set reorder points around the first constraint that could stop fulfillment. This gives us a clearer view of what is available to customers. It prevents false confidence when the spreadsheet shows enough inventory, but the complete configuration is not ready to ship.

I start by measuring the replenishment rate of each SKU and by grouping items into hero/core, seasonal, and those that have not sold through in four months. Each group receives a different inventory commitment: higher for core heroes, cautious for seasonals, and minimal for slow movers after accounting for seasonality. My rule of thumb is that if a nonseasonal product has not needed replenishment in four months, we change its operational positioning or marketing rather than simply increasing stock. The planning habit that helped most was maintaining a recurring review of replenishment rates and category status so orders are adjusted to current demand signals rather than forecasts alone. That practice keeps us from running out of popular items while avoiding excess cash tied up in slow sellers.

I run my store FAZVIDA by myself, and I don't buy inventory before I know it sells. I sell through CJ Dropshipping's US warehouses, so my cash and space never get stuck in products nobody buys.

My n8n automation only lists a product if it has at least 20 units for each size and color in one US warehouse, is priced at or below market, and isn't another brand's product.

My biggest lesson: my supplier's own sync mixed China and US stock in Shopify, which could oversell. So I built my own workflow that updates US stock every 4 hours.

My tip: don't trust one stock number. Let your supplier hold the inventory, check it often, and set clear rules before anything goes live.

Basit Ahmad Fazulyar, Founder, FAZVIDA (fazvida.com)