A dashboard can be full of numbers and still leave a retailer asking the same question:
What should we do with this product now?
One retailer described the problem after building reports from sales and stock data. Purchasing staff could see the information, but working out when to order, what to order and how much remained difficult.
Another small-store operator put the consequences more plainly:
- guessing what to reorder
- reacting too late when products go out of stock
- sitting on products that will not sell
- not knowing what to discount and what to hold
This is not only a reporting problem. It is a decision-design problem.
Most inventory tools organise products, quantities and sales. Some add low-stock alerts, forecasts or recommended quantities. Those are useful inputs. But a retailer still needs a repeatable way to turn the inputs into one of four actions:
1. Reorder.
2. Hold.
3. Investigate.
4. Discount.
The missing third option matters. Without an investigate path, uncertain data often produces an expensive order or an unnecessary markdown.
Start with the decision, not the dashboard
The first question in a weekly stock review should not be, "Which report do we open?"
It should be, "Which products need a decision?"
Create a short exception list rather than reviewing every SKU with the same attention. Include products that meet one or more conditions:
- stock may run out before replenishment can arrive
- sales have slowed materially
- stock has been held longer than expected
- margin or cash exposure is becoming uncomfortable
- a supplier order or transfer is already open
- the count conflicts with recent sales, returns or movements
- a promotion, season or local event changes normal demand
This turns analytics into a focused action list. Staff can concentrate on the products where a decision has economic or customer consequences.
Build a minimum evidence card for each product
Before choosing an action, place the same minimum evidence beside each flagged product.
1. Sellable stock now
Do not treat every unit in the building as available.
Separate:
- sellable stock
- reserved or allocated stock
- damaged or quarantined stock
- stock in transit
- confirmed inbound stock
- inbound that has been discussed but not ordered
The decision should begin with what can actually satisfy demand.
2. Recent sales velocity
Calculate demand over a period that represents how the product normally sells.
Average daily sales = representative units sold / representative selling days
A seven-day window may suit a fast-moving product. A longer window may be more useful for an expensive or slower item. Compare more than one period when promotions, weather or seasonality can distort the result.
Do not simply repeat the last order. A retailer in one inventory discussion described how "Same As Last" failed as products moved in and out of popularity.
3. Stock cover
Stock cover estimates how long the current sellable quantity may last at the observed rate.
Stock cover in days = sellable stock / average daily sales
This is not a promise. It is a way to compare current cover with supplier lead time and business risk.
4. Lead time and variability
A supplier that normally delivers in three days creates a different decision from one that needs four weeks and sometimes slips.
Record:
- normal lead time
- recent late-delivery pattern
- order cut-off days
- pack size or minimum order
- whether a substitute supplier exists
The average is useful. The variability determines the buffer.
5. Margin and cash exposure
Two products with identical sales velocity may deserve different actions.
One has healthy margin, stable demand and a reliable supplier. The other ties up cash, occupies scarce space and can only be bought in a large case.
Check:
- gross margin or contribution margin used by the business
- purchase cost and required order quantity
- cash committed if the order is placed
- storage, expiry or obsolescence risk
- value of the customer demand that may be lost
The goal is not to maximise stock. It is to make a defensible trade-off between availability and exposure.
6. Stock age and last sold date
Slow stock needs context.
A coat held through summer may be sensible. A phone accessory for a discontinued model may not be. A product that has not sold because its barcode points to the wrong variant is not a pricing problem at all.
Use stock age and last sold date as prompts for review, not automatic markdown commands.
7. Known events
Add what the historical numbers cannot know by themselves:
- a promotion that just ended
- a planned campaign
- a seasonal peak
- a local event
- a supplier price change
- a display change
- a temporary stockout that suppressed sales
This is where retailer judgment remains essential.
Action 1: Reorder
Reorder when credible demand is likely to consume available supply before replenishment can arrive.
A practical starting point is:
Reorder point = expected demand during lead time + safety buffer
Consider a product selling about three units per day. The supplier normally takes seven days. The retailer uses a six-unit buffer because deliveries sometimes slip.
Expected lead-time demand: 3 x 7 = 21 units
Reorder point: 21 + 6 = 27 units
If only 24 sellable units remain and no confirmed inbound order exists, the product has crossed the decision point.
The order quantity should then reflect the retailer's target cover, current sellable stock, confirmed inbound, case sizes, minimum order and available cash.
Provisional order quantity = target stock - sellable stock - confirmed inbound
Do not place the order mechanically if the underlying count is doubtful or the recent demand came from a one-off event. Move the product to investigate instead.
Action 2: Hold
Hold is an active decision, not indecision.
It is appropriate when:
- current and confirmed inbound stock provide enough cover
- demand is uncertain and another review period will improve confidence
- supplier lead time is short enough to wait
- the minimum order would create too much exposure
- a seasonal or promotional event has just distorted sales
- another location can supply the product if necessary
Record what would change the decision. For example:
"Hold until Friday. Reorder if sellable stock falls below 18 units or the open transfer is delayed."
That is stronger than leaving the product on a list with no owner or review date.
Action 3: Investigate
Investigate when the evidence conflicts or the result looks implausible.
Common triggers include:
- the system says stock exists but staff cannot find it
- returns were processed without restoring the correct variant
- an inbound order is shown as received but remains unavailable
- a transfer is open between locations
- two product records split sales for the same item
- sales change sharply without a known cause
- stock age is high but the product has recently been unavailable online
Assign one specific check and one owner:
- count the shelf and back room
- reconcile open purchase orders
- inspect returns and adjustments
- check product identity and variants
- compare locations
- confirm supplier status
Only then return the product to reorder, hold or discount.
This pause protects cash and margin from bad data.
Action 4: Discount
Discount when weak demand and ageing stock are real, and a price intervention is commercially sensible.
Before changing price, ask:
- Has the product been available and visible for a fair selling period?
- Is the product record correct?
- Is the demand decline persistent rather than seasonal?
- Can the margin absorb the proposed reduction?
- Would a bundle, relocation, supplier return or store transfer be better?
- What result would make the intervention successful?
Choose a measured action. A small price test on selected units may be more informative than marking down the entire quantity immediately.
Set an end date and compare:
- units sold before and during the intervention
- margin retained
- cash released
- remaining stock age
- whether the product attracted useful basket sales
A discount without a review becomes a habit. A discount with a hypothesis becomes a controlled decision.
Why a low-stock alert is not the final answer
Lightspeed's current inventory documentation shows how quantity on hand, reorder triggers, recommended order levels, cost, last sold and period sales can be assembled in a reorder report.
Source:
https://shopkeep-support.lightspeedhq.com/hc/en-us/articles/47480050225947-Inventory-Tracking-Reorder-Report
Shopify's current retail sales documentation also shows product, variant, vendor and location views, with net quantity reflecting sold units minus returned units.
Source:
https://help.shopify.com/en/manual/reports-and-analytics/shopify-reports/report-types/default-reports/retail-sales-reports
These are useful building blocks. They also demonstrate why the retailer still needs a decision layer.
A trigger says a threshold was crossed. It does not know whether:
- another order is already inbound
- a promotion caused the demand spike
- the current count is wrong
- the supplier now has a different minimum
- cash is needed elsewhere
- a product should be transferred rather than purchased
The right system should shorten that investigation, not hide it.
Run a 30-minute weekly decision review
Keep the routine small enough to survive a busy trading week.
Before the meeting
Generate an exception list for products that may need one of the four actions.
During the meeting
For each product:
1. Confirm the evidence card.
2. Choose reorder, hold, investigate or discount.
3. Record the reason.
4. Assign an owner.
5. Set the next review date or trigger.
After the meeting
Complete purchase, transfer, count, supplier or pricing actions. Keep a short decision log.
At the next review, ask:
- Did the expected demand occur?
- Did the supplier arrive when expected?
- Did the discount release stock without destroying margin?
- Was the investigation caused by a recurring data problem?
- Which reorder rules or buffers need adjustment?
This feedback loop makes the decision system better without pretending every forecast will be correct.
Where analytics and AI should help
Analytics should bring the relevant evidence together, rank exceptions and make the reason visible.
AI can help a retailer ask questions such as:
- Which products may run out before their normal supplier can deliver?
- Which items have weak sales and rising stock age?
- Which low-stock alerts already have confirmed inbound quantities?
- Which products changed sharply from their normal pattern?
- Which decisions were repeatedly overridden by staff, and why?
The output should support judgment. It should not become an unexplained command.
EzyCarto's current public Analytics scope includes sales and inventory dashboards, alerts, product and location drill-downs, natural-language questions, predictive analytics and saved reports. Smart Inventory adds movement records, low-stock and ageing notifications. Supply Chain adds transfer visibility and movement validation.
Product pages:
https://ezycarto.com/analytics
https://ezycarto.com/smart-inventory
https://ezycarto.com/supply-chain
Together, those capabilities can reduce the work required to assemble a decision. The retailer still owns the thresholds, commercial constraints and final action.
The practical checklist
Before acting on a product, confirm:
- [ ] sellable stock is separated from reserved, damaged and unconfirmed stock
- [ ] recent demand uses a representative period
- [ ] returns and suppressed demand are considered
- [ ] confirmed inbound orders and transfers are visible
- [ ] supplier lead time and variability are current
- [ ] margin, order size and cash exposure are understood
- [ ] stock age, last sold and seasonality have been checked
- [ ] conflicting data moves the product to investigate
- [ ] the decision has an owner and a reason
- [ ] the outcome will be reviewed
The objective is not a perfect forecast.
It is a store that can make the next stock decision with less guessing, clearer accountability and better use of cash.
Related guides:
https://ezycarto.com/blog/stockouts-are-not-always-demand-problems
https://ezycarto.com/blog/barcode-inventory-starts-with-the-product-record-not-the-scanner
https://ezycarto.com/blog/stock-has-more-than-two-states
https://ezycarto.com/blog/before-you-buy-an-erp-supplier-purchase-order-control-map