How Small Businesses Can Optimise Inventory Without Enterprise Software
- 19 hours ago
- 9 min read
A small business can improve inventory without buying enterprise software. The three changes that matter most are classifying SKUs by how their demand actually behaves, replacing fixed reorder points with service-level-driven stock targets, and treating slow-moving lines with a stocking policy rather than a forecast. Applied properly, this typically releases working capital within months rather than years. In one Australian distribution business we worked with, backtesting the new approach lifted fill rate from 69% to 83% and cut stockouts by 45% — on around 250 SKUs, delivered in Excel, with no platform licence.
Why small business inventory gets stuck
The problem is well documented and it is not a small one. Netstock's 2025 Supply Chain Planning Benchmark found that 55% of surveyed small and mid-sized businesses were holding at least 20% excess stock, with 46% reporting that at least 5% of inventory had become dead stock and 17% saying more than 10% had (reported by Xorosoft).
That stock is not free to hold. APQC benchmarking puts inventory carrying cost at 20% to 30% of inventory value per year once you add cost of capital, storage, insurance, shrinkage and obsolescence (Impact Analytics). On $500,000 of stock, that is $100,000 to $150,000 a year of cost that never appears as a single line on an invoice, which is precisely why it gets ignored.
And the pain runs in both directions. IHL Group's inventory distortion research puts the global cost of overstocks and out-of-stocks combined at over US$1.7 trillion, with smaller businesses losing an average of around 4% of annual revenue to stockouts. Most owners feel this as a contradiction they cannot explain: the warehouse is full, and the thing the customer wanted is not in it.
That contradiction is the signature of a business that has outgrown its inventory method rather than one that is badly run.
Traditional inventory optimisation methods, and where they break
These methods are not wrong. Most small businesses run on some combination of them and they work perfectly well up to a point. It is worth knowing exactly where that point is.
Min/max reorder points. Set a minimum, reorder up to a maximum. Simple and transparent. But the numbers are static, usually set once from judgement, and they do not move when demand or lead times move. They also contain no notion of a service level, so you cannot say “hold enough to hit 98% availability” — you can only say “hold 40”.
Economic order quantity (EOQ). Balances ordering cost against holding cost to find an optimal batch size. Genuinely useful, but it assumes steady, known demand — an assumption that rarely survives contact with a real order book.
ABC analysis. Rank SKUs by value and manage the top ones closely. Still worth doing. The limitation is that it sorts by value only, and value tells you nothing about how predictable an item is. Two A-class items can behave completely differently and need completely different treatment.
Periodic review and gut feel. A buyer looks at the list every Monday and orders what looks low. This scales to roughly the number of SKUs one experienced person can hold in their head. Past that, the tail gets ignored — and the tail is usually where the dead stock accumulates.
Averaging the last few months. The most common forecasting method in small business. It works on smooth demand and fails badly on lumpy demand, because an average of mostly-zeros produces a number that is wrong in every single period.
The common thread is that all of these treat every SKU the same way and none of them respond to variability. That is fine when you hold 200 predictable lines. It stops being fine somewhere around the point where you have more SKUs than a person can personally know.
What newer inventory optimisation methods actually do differently
The modern approach is less about clever algorithms than about matching the method to the demand pattern. Four ideas do most of the work.
Classify demand before forecasting it. Split the catalogue by how often an item sells and how variable that demand is — commonly using average demand interval and coefficient of variation. This tells you which SKUs are smooth, erratic, intermittent or lumpy, and each of those needs a different method. Applying one forecasting model across a whole catalogue is the single most common cause of poor accuracy.
Use purpose-built methods for intermittent demand. Croston's method and its variants exist specifically for items that sell in occasional bursts. They forecast the size of a demand event and the interval between events separately, which is a far better description of reality than an average that assumes something sells every month.
Make safety stock an output, not a guess. Calculate it from demand variability, lead time and lead-time variability, against a service level you choose per class. Set 98% on the lines that matter and 85% on the long tail, and let the numbers fall where they fall. This is usually where the working capital release comes from, because it takes stock off the stable lines that never needed the buffer.
Stop forecasting the sparsest tail entirely. This is the counter-intuitive one. For items selling under about one unit per period, percentage-error metrics break down mathematically — one unit of error against an actual of one unit is a 100% miss, and no model fixes that. The better approach is to describe the distribution of demand over the lead time and set a stocking policy against a service level, rather than chasing a point forecast that cannot exist.
The real caveat: process and tools, not method
None of the above is secret. It is taught, published and built into every planning platform on the market. So why do most small businesses still run on min/max?
Because the tools that implement it were built for someone else. Enterprise planning platforms are priced by SKU count and stocking location, sold as multi-year subscriptions, and quoted rather than listed. Implementations run from around 30 to 45 days at the fastest mid-market end to 12 months or more at enterprise tier. For a business turning over a few million dollars, that is both more capability than the problem requires and more commitment than the balance sheet justifies.
The second constraint is time, and it is the one vendors underestimate. In a small business the person doing the purchasing is usually also doing three other jobs. A sophisticated system that needs configuring, tuning and babysitting will lose to the spreadsheet every time — not because the spreadsheet is better, but because it is already there and nobody has to learn it. Software that goes unused is the most expensive kind.
So the practical question is not “what is the best inventory optimisation method”. It is “what is the best method that this specific business will actually keep using in six months”. Those are different questions and they often have different answers.
Case study: a specialist distribution business
An Australian distribution business importing specialist technical product into Australia and New Zealand came to us with the classic symptom: cash tied up in stock, and still missing sales on the lines customers asked for.
The catalogue was around 250 SKUs. Analysis showed a very particular shape: roughly 33 regular movers carried about 81% of unit volume, while more than 200 lines sold under one unit per SKU per quarter. Averaging across that catalogue — which is what was happening — could never work, because the two halves behave nothing alike.
What we did:
Classified every SKU by demand interval and variability, then selected a forecasting method per segment rather than one method for all
Validated with rolling-origin backtesting on held-out history, not in-sample fit, so the numbers reflected what the model would have done at the time
Replaced point forecasting on the intermittent tail with an empirical lead-time demand distribution and service-level stocking
Delivered the whole thing as a planning workbook in Excel, inside their own environment, because that is what the team already used
Backtested against actual history, the new approach lifted fill rate from 69% to 83% and reduced stockouts by 45% — better availability on the lines that mattered, with less capital spread across the ones that did not.
One honest note, because it matters for anyone benchmarking their own numbers. Blended forecast accuracy across the whole catalogue still looked poor on paper. That is not a modelling failure — it is a mathematical ceiling on demand that sparse, and published academic benchmarks using methods designed specifically for intermittent demand land in similar territory. The metric worth reporting to a business owner is service and stock, not a percentage error on items that sell three times a year. Any consultant who promises you a low forecast error on a long tail is either measuring it wrong or not measuring it.
What a low-cost setup looks like
Supply Logis builds inventory and replenishment capability for small businesses inside the client's own environment, shaped around how they already work. The methods are the same ones the enterprise platforms use. What differs is the fit and the commitment.
Built to your process, not the other way round. If your team plans in Excel, the tool is Excel. The calculation underneath changes; the screen they use does not have to. This is the single biggest determinant of whether anything survives past month three.
Inside your own data environment. Sandboxed, with your sales history and cost data staying where it already sits. No migration into a vendor tenancy and no security review to survive first.
Implemented in 1 to 2 months. From commencement to a working planning cycle, rather than a multi-quarter programme.
Payback typically under 6 months. Driven mostly by working capital released from over-stocked stable lines, which is cash that comes straight back into the business rather than showing up as an accounting adjustment.
No subscription and no per-SKU fee. A one-off build you own, so adding SKUs or users does not change what you pay. Depending on complexity we can also run a proof of concept on your own history first.
For a fuller comparison of the licensed platforms and where each fits, see our guide to the best demand planning and inventory software in Australia, and our published engagement pricing.
Five things to do this month, whoever you work with
Calculate your own carrying cost rate. Cost of capital, storage, insurance, shrinkage, obsolescence. Once you know the real number, every purchase order looks different.
Run a dead stock report. Anything with no movement in 180 days. Most owners are surprised by the total, and it is the fastest cash you will find.
Split your catalogue by demand pattern, not just by value. Count how many SKUs sold in fewer than half the periods last year. That group needs different treatment entirely.
Check your lead time variability, not just the average. Stockout risk is driven by how much lead times move, and planning off a mean understates it badly.
Pick a service level per class deliberately. Even writing down “98% on A lines, 90% on B, 85% on C” puts you ahead of most businesses your size, because it makes the buffer a decision rather than an accident.
Frequently asked questions
How can a small business optimise inventory without expensive software? Start by classifying SKUs by demand pattern rather than value alone, set service levels per class, and calculate safety stock from demand and lead-time variability instead of using fixed min/max numbers. For the slowest-moving lines, use a stocking policy based on lead-time demand distribution rather than trying to forecast them. All of this can be implemented in a spreadsheet inside your own environment; it does not require a licensed planning platform.
How much does excess inventory actually cost? Carrying cost is typically 20% to 30% of inventory value per year according to APQC benchmarking, covering cost of capital, storage, insurance, shrinkage and obsolescence. On $500,000 of stock that is $100,000 to $150,000 annually. Netstock's 2025 benchmark found 55% of small and mid-sized businesses were holding at least 20% excess stock.
Why are we overstocked and out of stock at the same time? Because a single fixed reorder point cannot express both a demand forecast and the uncertainty around it. Min/max holds too much of the stable, predictable lines where variability is low and too little of the volatile ones where it is high. It is the signature symptom of a business that has outgrown static reorder points.
How long does it take to improve inventory in a small business? A Supply Logis build is typically implemented within 1 to 2 months of commencement, with payback usually under 6 months, driven mainly by working capital released from over-stocked stable lines. Licensed platforms range from around 30 to 45 days at the fastest mid-market end up to 12 months or more at enterprise tier.
Can you forecast products that only sell a few times a year? Not accurately, and you should be suspicious of anyone who says otherwise. Where an item sells under about one unit per period, percentage-error metrics break down mathematically. The right approach is to describe the distribution of demand over the lead time and set a stocking policy against a chosen service level, rather than chasing a point forecast. Judge the result on fill rate and stock value, not forecast error.
Do we have to replace our current system? No. The planning layer sits on top of whatever you already run, using sales history and item data your ERP or accounting system already holds. Keeping the interface your team already knows is usually what determines whether the change sticks.
If your stock is tying up cash you would rather put into growth, Supply Logis offers a free diagnostic that sizes the opportunity in your own numbers before you commit to anything.
Current as at 31 August 2026. Case study figures are backtested results on the client's own historical data and are anonymised at the client's preference; outcomes vary with catalogue size, demand pattern and data quality. External benchmarks are attributed and linked to their published sources.