Most dropshipping stock problems do not start with a bad supplier. They start with a forecast that nobody owns. Your 3PL holds the physical inventory, your store holds the sellable quantity, and the gap between those two numbers is where overselling, dead stock and emergency air freight come from. This guide walks through a forecasting workflow you can run with a 3PL without hiring a data team, and shows exactly which data to exchange so the numbers stay honest on both sides.
The goal is not a perfect forecast. It is a forecast that is wrong in ways you can afford: enough buffer to absorb surprise demand, few enough units that slow movers do not eat your cash. Storage and SLA terms determine how expensive being wrong is, so read your pricing sheet before you set buffer levels.
Start With Demand Signals You Actually Have
You do not need machine learning to forecast a dropshipping catalogue. You need three series: units sold per SKU over the last 90 days, units sold per SKU over the same 90 days last year if you have them, and open orders. For most catalogues under a few hundred SKUs, a spreadsheet handles this. The 90-day average gives you the baseline; the year-over-year view catches seasonality; open orders show committed demand that has not shipped yet.
Split the baseline by channel if you sell on more than one. Marketplace orders often arrive in bursts tied to promotions you do not control, while your own store demand is smoother. When channels disagree, forecast against the blended rate but keep the split visible, because a promotion on one channel can drain stock the other channels were counting on.
- Fast movers: forecast weekly, review weekly. These SKUs cause most overselling events.
- Mid movers: forecast monthly, review monthly. Reorder when lead time demand crosses your reorder point.
- Long tail: do not forecast at all. Cap the depth, sell through, and let the SKU retire rather than tying cash to slow units.
Safety Stock, Reorder Points and What Your 3PL Sees
Safety stock exists to cover the two delays you cannot schedule: supplier lead time variance and inbound receiving time at the warehouse. A workable starting formula for each SKU is the square root of average lead time in days multiplied by daily demand, multiplied by a service factor you tune by season. It is not academic precision, but it beats the gut-feeling “one more pallet” that quietly doubles your storage bill.
The reorder point is daily demand during lead time plus safety stock. When your supplier-side inventory crosses that line, you order. The catch is that your supplier-side inventory and your 3PL-side inventory are different numbers, and both matter. Your 3PL sees sellable stock on the shelf; your supplier sees production or stock ready to ship. Syncing those two views is what prevents the classic failure where the store shows 40 units, the shelf holds 12, and the supplier needs three weeks to fill the gap.
Technical background: the arithmetic behind buffer stock comes from safety stock theory, and the trade-off between ordering often versus ordering deep is formalised in economic order quantity. Inventory frameworks such as ABC analysis also explain why it is rational to forecast fast movers weekly while letting the long tail coast on simple caps.
How Often Stock Counts Should Sync
Sync cadence should match how fast the number changes. For most dropshipping operations, a three-layer rhythm works:
- Real-time decrements from your store or order management system for every paid order. This is non-negotiable; batched decrements are how overselling happens during a promo.
- Daily file or API pull from the 3PL covering on-hand, allocated and in-transit units per SKU. Reconcile against your store quantity daily for fast movers.
- Cycle counts scheduled by the 3PL, usually weekly or monthly by zone. A 3PL that never counts is guessing, and their guess becomes your oversell.
Ask prospective warehouses how they handle count discrepancies before you sign. A good partner reports variance the same day with a photo or a recount note, and their SLA states who absorbs the cost of mis-picks caused by bad counts. This is one of the questions worth asking during a 3PL migration, because fixing count discipline after go-live is painful.
Channel Quantity Rules That Prevent Overselling
Once the warehouse number is trustworthy, protect it with channel rules. The simplest rule set that works for most brands:
- Reserve a buffer of 2 to 5 units per fast-mover SKU that never lists on marketplaces, so your own store can keep selling when a marketplace promotion spikes.
- Push separate quantities per channel rather than one shared pool, if your order system supports it. Shared pools mean one channel’s burst starves the others.
- Cap marketplace listings at 80 percent of true available for mid movers. The visibility cost of a slightly lower listed quantity is smaller than an oversell cancellation.
During Q4 peak season, tighten these rules: cut the marketplace cap to 60 to 70 percent for gift-range SKUs and shorten your reconcile cycle from daily to twice daily. Carrier cutoffs compress the recovery window, so an oversell in December costs more than the same mistake in July.
A Weekly Routine That Holds It Together
Forecasting fails when it is a project instead of a routine. Fifteen minutes per week is enough for a small catalogue:
- Pull the 3PL on-hand report and compare against store quantities for every SKU that sold in the last 7 days.
- Review reorder points for SKUs where supplier lead time changed, where weekly demand moved more than 30 percent, or where a cycle count found variance.
- Flag SKUs with more than 90 days of cover as candidates for a promotion or a reorder freeze, using the same logic you would apply to packaging MOQ decisions: cash tied in depth has a carrying cost.
Keep a one-line log of every manual override you make. After a quarter you will see whether your overrides beat the formula or hurt it, and that evidence is what tells you when to trust the model and when to trust yourself.
How much safety stock should a dropshipping store hold at a 3PL?
Start with about one week of average daily demand for fast movers, tuned up during promotions and Q4 and down for SKUs with stable demand. Treat it as a running number: review monthly and adjust from observed stockout and overstock events rather than setting it once.
Who is responsible for stock accuracy, the store or the 3PL?
Both, with a split: the 3PL is responsible for the physical count and reporting variance the same day; you are responsible for sync cadence, channel rules and reconciliation. Your contract SLA should state this split explicitly, including who absorbs the cost of mis-picks caused by stale counts.
How often should stock sync between my store and the 3PL?
Real-time decrements per paid order, a daily on-hand and in-transit reconciliation for all SKUs, and scheduled cycle counts at the warehouse. During peak season, move the reconciliation to twice daily for fast movers.
Can I forecast dropshipping inventory in a spreadsheet?
Yes, up to a few hundred SKUs. A 90-day unit-sales baseline, a year-over-year comparison for seasonality, open orders, and a safety-stock line per SKU covers most needs. Move to dedicated tools when updates start taking more than an hour per week or when multi-warehouse allocation becomes routine.
What data should I ask my 3PL for every day?
On-hand, allocated and in-transit units per SKU, plus cycle count results and any variance notes. If your 3PL cannot produce a daily per-SKU file or API feed, treat that as a serious operational risk before you scale ad spend.
How do promotions change my forecast?
Promotions concentrate demand into a short window and pull demand forward from later weeks. Forecast the uplift from comparable past promotions, raise the marketplace cap temporarily rather than listing full depth, and pre-position safety stock for the promoted SKUs at least one lead time before the campaign starts.
When should I stop restocking a slow SKU instead of forecasting it?
When projected sell-through exceeds 90 days of cover or the SKU has needed repeated manual overrides with no payoff. Retire it with a clearance offer instead of tying storage and cash to units that ship once a month.
