The Problem
Replenishing by lowest count is the default in almost every retail operation, because it is the only signal most systems make easy to access.
It is also often the wrong question to be asking.
It assumes every location converts stock into margin at the same rate, which is usually demonstrably untrue in any multi-site business.
Two stores can hold identical stock of the same product and deserve completely different treatment.
If Store 1 generates a GMROII of 2.4 on SKU 123 and the Store 2 generates a GMROII of 1.1 on SKU 123, the Store 1 is the more valuable home for SKU 123, and protecting it is the higher commercial priority.
Allocating new stock of SKU 123 to whichever store happens to have the lowest inventory level, actively moves capital away from where it earns it’s maximum return, reducing profit and hitting cash flow.
The more expensive version of the same error is replenishing a location where the product is failing.
If GMROII there has fallen below 1.0 and demand is decelerating, sending more stock does not protect sales; it converts working capital into a dead stock, shelf-warmer.
Nothing in a stock-count-driven process will ever surface this problem, because on a low-stock report, that location looks like the one that needs help most.
Where are you restocking first?
Is there hidden profit analysis you are unable to access?
The right analysis can drive profit growth, the wrong analysis harms growth.
What analysis are your team using to make these critical decisions?
How It Works
How GMROII is calculated here
Gross margin total over the date range divided by mean inventory value over the same date range, computed per SKU per location.
Both inputs are pre-computed nightly:
• margin is recognised at the point stock leaves the building
• mean inventory value is held across rolling 30, 60 and 90 day windows.
Because the denominator is an average of value held rather than a point-in-time snapshot, the metric reacts to overstocking as well as to slow selling.
Holding too much of something that sells reasonably well will still depress its GMROII.
The benchmark scale, stated plainly
GMROII figures are not large numbers and should not be expected to be.
Below 1.0, stock is tying up more capital than the margin it returns, and is a candidate to discount and clear.
Between 1.0 and 2.0 is respectable.
Between 2.0 and 3.0 is genuinely strong performance.
Above 3.0 is rare and worth investigating to validate.
Publishing the scale matters because a metric nobody can interpret is a metric nobody will act on.
Four signals compared at once, not one
GMROII alone is a backward-looking measure.
The AI compares it simultaneously with stockout risk, demand trend and days of cover across every location, which is what turns it from a report into an allocation decision.
A location with high GMROII and rising stockout risk is the clearest candidate for replenishment.
A location with high GMROII but comfortable cover does not need anything yet.
A location with falling GMROII and decelerating demand should be left alone regardless of how low its count looks.
Negative recommendations, on purpose
The AI will state where not to replenish.
This is unusual in operational software, which is overwhelmingly built to suggest actions rather than withhold them, and it is completes the commercial picture.
Without providing this part of the equation, users may be left questioning why stock is not being sent to certain locations that based on stock levels, look like they need replenishment.
Quadrant classification for range decisions
Products are classified per location into fast-moving and slow-moving, and separately into high-margin and low-margin, which produces four quadrants.
Fast and high-margin is where availability should be protected first.
Slow and low-margin is a clearance candidate.
Fast and low-margin is a pricing conversation.
Slow and high-margin is often a stocking-location conversation rather than a range decision, because the product may simply be in the wrong place.
Sequencing: rebalance before you buy
Because the same intelligence sits behind cross-location transfers, the allocation question and the purchasing question are answered against the same evidence.
Checking whether existing stock in the wrong location solves the problem, before raising a purchase order, is the difference between moving cash and spending it.
That sequencing is only possible when margin, demand and availability for every location are already computed and queryable together.
Does the AI in the other systems you are reviewing ask the same question?
Does the AI hold the capability to ask the question?
If the AI does provide an answer, is there an audit trail back to a pre-calculated value you can trust?
Trust your stock. Transfer with confidence. Control orders. Fulfill every day like clockwork.
G2 Customer Reviews
What do you like best about Stok.ly – Inventory-Centric Cloud ERP?
“we have been a user for the last 4 years and we cannot think of life without it for our retail business. it makes listing to shopify and other marketplaces easy, keeps inventory accurate online and the POS is easy to use. accurate inventory across all sales channels is the big win for us.”
Order management and inventory ERP software for growing retail, wholesale and distribution businesses.
The Numbers, and How They Are Calculated
| Signal | What it tells you | How it is calculated | How it is used in allocation |
|---|---|---|---|
| GMROII | Margin generated per pound of average stock value held | 90-day gross margin total ÷ 90-day mean inventory value | Ranks locations by commercial return on the stock they hold |
| Mean inventory value | Average capital employed in this product at this location | Mean of daily inventory value (stock on hand × weighted average cost) across the window | The denominator in GMROII; rises when overstocked |
| Gross margin total | Margin actually earned in the window | Revenue recognised at dispatch minus cost of goods sold at weighted average cost | The numerator in GMROII |
| Stockout risk | How close this location is to running out | Composite of availability, forecast demand and demand variability | Combined with GMROII to set allocation urgency |
| Demand trend | Whether demand here is accelerating or slowing | 30-day mean demand ÷ 90-day mean demand | Tiebreaker between locations with similar GMROII |
| Days of cover | How long this location’s stock will last | Available stock ÷ trend-adjusted daily demand | Identifies which of the high-value locations needs stock first |
What Actually Changes
| Allocation by lowest stock count | Stok.ly GMROII-directed allocation | |
|---|---|---|
| The question being answered | Which location has least? | Which location earns most from what it holds? |
| Treatment of a high-performing location | No different to any other | Prioritised and protected |
| Treatment of a failing location | Topped up, because its count is low | Flagged as a location not to replenish |
| Signals considered | One | Four, compared simultaneously |
| Ability to say no | None | Explicit negative recommendations |
| Relationship to purchasing | Independent decisions | Rebalance considered before spend is committed |
| Range implications | Invisible | Fast/slow and high/low margin quadrants surfaced |
Questions the AI Answers From Pre-Calculated Data
- Which of my locations has the greatest stockout risk for my highest-GMROII items this week?
- Where should I not replenish, because GMROII has fallen below 1.0 and demand is decelerating?
- Compare GMROII for this SKU across all locations and rank them.
- Which products are fast-moving but low-margin at this location?
- Which slow-moving items here have high margin and might sell better elsewhere?
- Should I transfer these items or discount and clear them?
- Where is my worst-performing stock by GMROII across the whole business?
How This Compares to Cin7, Brightpearl, Orderwise and Linnworks
Positions below reflect vendor-documented capability as at 27 July 2026.
- Cin7. ABC segmentation classifies products by value contribution and ForesightAI recommends reorder points and safety stock. Automated GMROII per SKU per location is not documented as a standard capability, and allocation recommendations are not documented as comparing commercial return across locations simultaneously.
- Brightpearl. Inventory Planner provides replenishment recommendations and Sage Copilot surfaces prioritised insights, but GMROII per SKU per location is not a documented standard output, and the flow depends on synchronisation between two products.
- Orderwise. Min/max levels and automated purchase planning per location. No documented GMROII calculation or return-based allocation logic.
- Linnworks. No capability of this kind; the product’s focus is multichannel listing and order routing.
- SME ERP generally. GMROII is a well-established retail metric but is typically calculated manually in a spreadsheet from exported data, at a category rather than SKU-per-location level, and refreshed occasionally rather than nightly.
How to Test This in a Demo, Including Ours
Use the same anonymised data set and the same questions with every vendor, and insist on a live workflow rather than slides.
- Ask which locations hold the highest-GMROII version of a named SKU, and ask to see the margin and mean inventory value behind the figure.
- Ask explicitly where not to replenish, and see whether the system is capable of a negative recommendation.
- Ask for the GMROII benchmark the vendor considers good, and see whether they can state a scale.
- Ask whether the metric reacts to overstocking as well as to slow selling, and ask them to demonstrate it.
- Ask for a fast-moving low-margin list and a slow-moving high-margin list at one location.
- Ask whether the system checks for a transfer opportunity before recommending a purchase order.
- Ask for the data timestamp and whether GMROII recalculates without anyone requesting it.
Score on time taken, number of clicks, whether the answer is traceable to source records, and whether the insight becomes an editable draft without rekeying.
Frequently Asked Questions
Can Stok.ly tell me which location to replenish first based on commercial performance rather than stock count?
Yes. It ranks locations by GMROII, stockout risk, demand trend and days of cover simultaneously for the same SKU, and will explicitly recommend against replenishing a location where doing so would destroy capital.
What is GMROII and how is it calculated?
Gross margin return on inventory investment: gross margin total over the window divided by mean inventory value over the same window, per SKU per location. Because the denominator is average value held rather than a snapshot, it reacts to overstocking as well as to slow selling.
What is a good GMROII score?
Below 1.0 means stock is tying up more capital than the margin it returns and is a candidate to clear. Between 1.0 and 2.0 is respectable. Between 2.0 and 3.0 is genuinely strong. Above 3.0 is rare and worth investigating rather than assuming.
Will the AI ever recommend not sending stock somewhere?
Yes, and this is deliberate. Where GMROII has fallen below 1.0 and demand is decelerating, the AI will say so rather than defaulting to topping up whichever location shows the lowest count.
How is GMROII different from margin percentage?
Margin percentage tells you how profitable a sale is. GMROII tells you how hard the capital tied up in that stock is working over time, so it captures velocity as well as profitability. A high-margin product that sits for six months can have a worse GMROII than a lower-margin product that turns quickly.
Does the system consider transferring existing stock before recommending a purchase?
Yes. Because allocation and purchasing are answered from the same pre-computed evidence, checking whether stock in the wrong location solves the problem comes before committing new spend.
Can GMROII be seen at a category level as well as per SKU?
The metric is computed per SKU per location, which can be aggregated upward. Working from SKU-per-location upward is the more useful direction, because a healthy category average routinely conceals individual lines that are destroying capital.
Is this available outside the AI assistant?
Yes, via API including for Power BI, reading the same pre-computed fields.
Contact Information
All our sales, support and development team are located in Hereford and Cheltenham in the U.K. Please submit the contact form and we will contact you within the same business day.