The Spreadsheet Replenishment Ceiling
Manual replenishment planning has a ceiling, and most growing brands hit it somewhere between 80 and 120 active SKUs. Below that threshold, a well-maintained spreadsheet with manually-set reorder points and weekly review cadence can work reasonably well. Above it, the cracks start to show: reorder points that haven't been updated since the product launched, velocity assumptions that no longer reflect actual selling rates, and planning meetings that run long because someone has to re-check the numbers before any decisions can be made.
The failure isn't laziness or poor process. It's that manual replenishment fundamentally doesn't scale with catalog complexity. A planning team of two or three people can effectively monitor reorder points for 80 SKUs on a weekly review cycle. They cannot do it reliably for 200 SKUs across multiple channels, multiple warehouse locations, and variable lead times — not without significant errors and omissions.
Replenishment automation isn't about removing human judgment from the process. It's about removing the low-value manual work that prevents humans from applying judgment where it actually matters.
What Most Brands Get Wrong When They First Automate
The most common mistake in early replenishment automation is automating the wrong logic. Specifically: taking the manual replenishment rules that existed in the spreadsheet — which were often rough approximations built under time pressure — and encoding them directly into a system as if they were correct.
Manual reorder points are typically set with one of two methods: a buyer's intuition based on experience with the product, or a simple average demand calculation multiplied by lead time with a flat safety stock buffer added. Neither of these accounts for demand variability in any meaningful way. They work adequately when conditions are stable and fail when demand spikes or supply lead times shift.
When you automate these flawed rules, you get fast wrong answers instead of slow wrong answers. The system dutifully fires purchase orders at the old reorder points, and six months later the brand has the same overstock and stockout problems they had before automation — plus a false sense of security because "the system is handling it."
The right sequence is: fix the underlying replenishment logic first, then automate it. Not the other way around.
Building Replenishment Rules That Actually Work
A well-designed reorder point has three components: cycle stock demand coverage (demand over your replenishment lead time), safety stock (buffer for demand and supply variability), and lead time itself.
The piece most brands get wrong is safety stock, which should scale with variability — not be a flat buffer applied uniformly across all SKUs. A SKU with consistent weekly demand needs less safety stock than one with high variance. A SKU sourced from a supplier with reliable 14-day lead times needs less safety stock than one with a 60-day international lead time and frequent delays.
Consider a footwear brand carrying roughly 160 SKUs across sizes and colorways that had set a flat safety stock of 30 units for all styles. Their core classic sneaker in black sold 85 units/week with low variance — the 30-unit buffer represented less than 2.5 days of demand coverage. Their seasonal canvas shoe in a fashion colorway sold 15 units/week with high variance (anywhere from 5 to 35 units/week based on weather and social media activity) — and needed proportionally much more buffer to handle the volatility. The flat safety stock left the core sneaker dangerously exposed and over-buffered the fashion colorway with unnecessary inventory.
Building variability-adjusted safety stock requires knowing the standard deviation of weekly demand for each SKU over a rolling window. This is calculable in a spreadsheet — it's just not something most teams bother to do until they've experienced a painful stockout on a top-selling SKU.
Lead Time as a Variable, Not a Constant
Most replenishment systems treat supplier lead time as a fixed constant: "this supplier takes 45 days." In practice, lead times vary, and that variability is one of the most underappreciated sources of stockout risk.
Lead time variability means that sometimes your order arrives in 38 days and sometimes in 57 days — and if your replenishment logic assumes 45 every time, the 57-day delivery will catch you short. The solution is to factor lead time variability into your safety stock calculation, which requires tracking actual delivery dates against expected delivery dates on a per-supplier basis.
For brands working with overseas suppliers, international shipping lead times during high-demand periods can run 20–30% longer than during off-peak periods. A replenishment system that doesn't account for seasonal lead time variation will be systematically under-stocked going into peak periods — precisely when you can least afford it.
Automating Reorder Triggers Without Automating PO Issuance
There's a useful middle step between fully manual replenishment and fully automated PO issuance that most growing brands should start with: automated reorder triggers that feed into a human review queue.
The system calculates daily when each SKU hits its reorder point, generates a suggested purchase order with quantity calculated from the replenishment logic, and surfaces it to the planning team for review and approval. The human reviews the suggested PO, checks for context the system doesn't have — a planned promotion that will spike demand, a supplier relationship issue, a product launch that's about to cannibalize this SKU — approves or adjusts, and releases.
We're not saying fully automated PO issuance is wrong — for very high-velocity, well-understood SKUs with reliable suppliers, automating the full cycle can make sense. But starting with the human review step gives planning teams the confidence to trust the system's calculations while retaining the ability to catch edge cases. Once the team finds they're approving 90%+ of suggestions without changes over a 6–12 month period, the case for removing the review step on stable SKUs becomes much stronger.
Measuring Whether Your Replenishment System Is Working
Replenishment automation success has specific, measurable indicators. After implementation, you should be tracking:
- Stockout frequency by SKU: Are you experiencing fewer stockout events per week/month? On which SKUs are stockouts still occurring — are they systematic (safety stock is wrong) or one-off (unusual demand spikes)?
- Average weeks-of-stock across the catalog: Is average WOS trending down while stockout rates are also falling? Reducing both simultaneously is the signal that replenishment logic is well-calibrated.
- PO accuracy rate: What percentage of system-generated suggested POs are approved without modification? High modification rates suggest demand assumptions are outdated or reorder logic needs adjustment.
- Planning team time allocation: How many hours per week is the team spending on reactive replenishment firefighting vs. forward-looking planning? The goal is to shift that ratio over time.
The transition from manual to automated replenishment is rarely a single event. It's a series of incremental improvements — better data, better logic, better tooling — that compound over time. The brands that get there don't do it all at once. They start with their highest-velocity SKUs, get the logic right for those first, then expand the scope once the model is validated.
