Reorder point
Stock level at which an order must be placed to avoid running dry before it lands — velocity × (lead time + safety stock).
A replenishment system built for catalogues where the long tail is the business — hundreds of thousands of SKUs, stocked thin, turning over fast.
The main view answers one question for every product you stock: do I need to order this, how many, and how urgently? Each row carries the numbers you need to act — on-hand, incoming, velocity, days of cover left, the reorder point, and a suggested quantity that respects your supplier's minimum order.
Stock level at which an order must be placed to avoid running dry before it lands — velocity × (lead time + safety stock).
Sized to last until the next order arrives, not merely until this one does. Respects minimum order quantities.
Cover remaining at the current rate, counting stock already on its way — colour-coded by urgency.
Anything under seven days of cover is flagged separately, so a long list still tells you where to start.
Filter by supplier to build one purchase order at a time, search by SKU or product name, and narrow to items you've actually restocked recently — so a final unit of a three-year-old line doesn't pull you back into buying stock for an obsolete device.
Every replenishment tool divides units sold by days elapsed. That silently assumes the product was on the shelf the whole time — and any product that sold out is punished for the days it had nothing to sell.
Stockmind samples your warehouse's stock history and counts, for each product, how many days it genuinely had stock. That figure is its ShelfDays, and velocity is measured against it rather than the calendar.
Sampling is daily by default, which places availability within about half a day. Past snapshots never change, so each one is extracted exactly once and later runs reuse everything but the newest dates — the cost is paid once rather than on every calculation.
Estimation can only detect products that are out of stock right now. Measurement sees the ones that sold out in the middle and were restocked since — which are, by definition, your fastest movers.
The statistical name for the problem is censored demand: a stockout doesn't record zero demand, it records nothing. Correcting it sits upstream of any forecasting method — feed uncorrected data to Croston's method or a Poisson model and both inherit the same bias, because neither can tell an empty shelf from a quiet week.
Two products can average identically and behave nothing alike. Stockmind can account for both the direction demand is moving and how erratic it is — each optional, each adjustable.
Compares the recent rate against the full period so a rising product is ordered like one that's rising, and a fading model isn't reordered on the strength of sales it made months ago. Confidence scales with recent volume, so one good week can't inflate an order.
A metronomic seller and a spiky one shouldn't carry the same cushion. Buffers scale to each SKU's own demand variability — roughly half for predictable products, up to two and a half times for volatile ones.
Both ship switched off. Turn them on when you want them, at whatever strength suits your catalogue — changes apply immediately.
Open orders count against reorder calculations, so you never double-order something already bought. But when it lands matters as much as whether it exists.
Only stock arriving inside your coverage window counts as protection. A shipment landing in sixty days does nothing for a product that will be empty in twelve — so it's tracked and visible, but it never produces a false all-clear.
Replenishment isn't only about what to buy. It's about what to stop buying — and what's quietly holding your cash.
Dead stock runs on its own full-year window, separate from the window that drives reordering. That distinction matters: a product that sells hard every November would look dead on a summer view and get cleared right before its peak.
Physically sitting in the warehouse with no sales in a year. Sorted by quantity, so the biggest piles of tied-up capital surface first.
Zero-stock SKUs with no sales — pure catalogue cleanup you can delete straight from your systems, with no warehouse handling at all.
Filter by supplier and by how long ago stock was last received, then export a clean list for a return conversation or a clearance run.
Stockmind has individual accounts rather than a shared password, so access can be given and withdrawn per person and sessions expire on their own.
Reports tell you what is happening. They don't tell you what to do first. The briefing assembles every signal — products selling with nothing on order, ranges with no colour left in stock, device generations rising and fading, over-extended ranges, capital sitting in slow stock, measurement going stale — and ranks them by what each is costing.
An optional AI summary reads the week back as prose: what to restock, which ranges to widen with new products, and what to stop buying. The model receives only figures Stockmind has already calculated, is instructed not to compute anything, and every number it writes is verified against the source data — anything that doesn't match is flagged rather than presented as fact.
The replenishment engine itself is entirely deterministic. No AI touches a reorder quantity, and the summary can be switched off without affecting anything else.
A product with a high return rate looks like it sells better than it does. Demand here is measured net of returns, because a returned unit goes back on the shelf — counting it as a sale overstates the rate and double-counts a unit already in your stock figure.
A high return rate usually points at something fixable — a photograph, a size chart, a compatibility claim — rather than at the product itself.
A Shopping campaign bids the same on a product with three days of stock as on one with three months, and the same on a range that is climbing as on one that is fading. It has no way to know the difference.
Availability-adjusted velocity is the point: a product that sold twenty units while in stock for fourteen days of a quarter is not a slow seller, and a campaign bidding on the calendar figure is bidding wrong on exactly the products that were selling well enough to run out.
Every replenishment tool tells you what to do today. Far fewer will show you whether any of it helped — which is the question your finance director, your board or your own scepticism will eventually ask.
Stockmind records the state of the catalogue at each calculation and reports how ten measures have moved: best sellers in stock, products selling while empty, demand being missed, dead stock lines and units, surplus stock, products at zero.
Each measure carries the direction that counts as an improvement, and the verdict follows that rather than the direction of travel. The report can tell you a measure got worse — which is precisely what makes the improvements worth showing anyone.
Figures come from the calculation that ran on each date. Nothing is back-filled and nothing is restated later.
Alongside availability and stock health, the report tracks how sales are spread across the range: the share of the catalogue that sells at all, and how much of the volume the fastest tenth carries. If holding a long tail is working, both should move — more products transacting, less concentration.
Every measure carries a sentence explaining how to read a movement in it. A change is reported as flat when it is smaller than the ordinary variation between runs, however far apart the endpoints sit.
Your products aren't independent. A strap exists as one design across six colours; a case exists across four device models. Treating those as unrelated SKUs hides decisions that are obvious to a human buyer.
A colour sitting empty while its siblings have stock is a soft loss — some demand shifts across. A range where every variant is empty has no substitute at all. Identical per-SKU numbers; completely different urgency.
Proven ranges with an empty variant, ranked by demand missed — the fastest wins, because the range already sells.
The share of sales each colour takes historically. Splitting a new launch across variants stops being guesswork.
One dead colour is a variant to drop. Six dead colours is a product line to exit. The distinction is now visible.
Every stockout says whether anything is left to absorb the demand, so the genuinely costly ones stand out.
Your top performers ranked by true sell-through rather than raw units — so a product that sold twenty units while only being in stock a fortnight ranks where it should.
The maths is only right if it reflects how you actually buy. Four settings shape every suggestion, and all of them apply instantly — no waiting for an overnight recalculation to see the effect.
How long from placing an order to stock being on the shelf — set globally, or per supplier so an overseas source isn't held to a local one's schedule.
The gap until your next order. An order has to last until the next delivery arrives, not just until this one does — order monthly with a two-week lead and today's stock must cover a month.
Your buffer against demand spikes and late shipments, flat or scaled to each SKU's variability.
How much the recent rate influences velocity, from off to fully recency-weighted.
Because changes apply live, you can see what a longer lead time or a leaner buffer does to your order size before committing to it.
Analysis is only useful if it becomes a purchase order. Every view exports in CSV or Excel, carrying whatever filters are active — so "supplier 7057, needs reorder, restocked within two years" comes out as exactly that list, with numbers as real numbers.
A long tail of hundreds of thousands of SKUs is where the quiet losses hide, so it can't be something the tool samples or truncates.
Search, filter, sort and export run across the entire catalogue and are delivered on demand — the browser never holds it all.
Reports read only the rows they need rather than loading the catalogue into memory, so a report over 200,000 SKUs answers in about a second and stays that way as you grow.
Stock levels, supplier codes, bin locations and order history sync directly. No weekly export ritual, no spreadsheets already out of date when saved.
If a data fetch can't complete, the run aborts rather than publishing a partial catalogue. Your last complete dataset stays intact.
Token-authenticated API, protected application directory, and cached data kept out of public reach.
Curious how the numbers are actually derived? The methodology page sets out every formula, including where the model is approximate.
Tell us about your operation and we'll show you what measured availability does to your numbers.
Request Access →