The second location is where inventory management stops being bookkeeping and starts being a discipline. One shop has one number for each product, and you can walk over and check it. Two locations plus a godown produce three numbers, a set of movements between them, and a permanent question: where is that stock right now, and who is responsible for it?
The cost of getting this wrong is asymmetric, and not in the direction most operators assume. IHL Group's 2026 research update attributes 65.6% of global inventory distortion to out-of-stocks and 34.4% to overstocks, with the combined figure at roughly 6.2% of worldwide retail sales (IHL Group, reported by Retail Insight Network, retrieved 2026-08-13). Having stock in the wrong branch produces both failures simultaneously: an empty shelf in one place and dead capital in another.
This guide covers the transfer workflow, the in-transit problem, how to set reorder points per location, and how to catch accuracy drift before it becomes a write-off.
Key Takeaways
- Out-of-stocks drive 65.6% of global inventory distortion against 34.4% for overstocks (IHL Group via Retail Insight Network, retrieved 2026-08-13). Misallocated stock across branches causes both at once.
- A stock transfer needs five states, not two. Skipping the in-transit state makes units vanish from your books for the length of the journey.
- Reorder points belong to the location, not the product. A single company-wide reorder level is the most common cause of one branch starving while another overflows.
- Count continuously, not annually. Rolling cycle counts weighted by value catch drift while corrections are still cheap.
- Measure per location. A company-wide accuracy figure of 96% can hide one branch running at 78%.
Why One Stock Number Stops Working
The moment goods can exist in more than one place, the single quantity-on-hand field becomes a lie of omission. It is arithmetically correct and operationally useless.
Consider a wholesaler with a Chattogram godown and two Dhaka outlets. The system says 240 units of a fast-moving SKU. A customer at the Uttara outlet wants 60. Can the outlet fulfil it? Nothing in that number answers the question. Perhaps all 240 are in Chattogram. Perhaps 200 are in Chattogram, 38 are at the other outlet, and Uttara has 2.
Location-level stock changes the question from "do we have it" to "can this location sell it today." That distinction drives four things a single number cannot support:
- Fulfilment promises. Telling a customer yes when the stock is 300km away is worse than telling them no.
- Replenishment decisions. You cannot reorder intelligently without knowing which location is short.
- Accountability. When stock goes missing, a location-level record narrows the investigation to one team and one building.
- Valuation by site. Landed cost differs by location when freight and duty differ. Site-level valuation keeps that honest.
The Transfer Workflow, in Five States
Most systems model a transfer as two events: stock leaves A, stock arrives at B. That works only if the journey is instantaneous and nothing is ever lost, which describes no real supply chain.
Why in-transit matters more than it sounds
When a truck leaves the godown on Sunday and reaches the outlet on Tuesday, those units exist. They are owned, they are insured, they are worth money, and on a two-state model they are invisible for 48 hours. Three problems follow.
Your valuation is understated. Stock in a truck is an asset. If it is deducted from the source before it is added to the destination, your balance sheet is wrong for the duration of every transfer, which for a busy operation means permanently.
Nobody is accountable for loss. If 100 units leave and 96 arrive, a two-state system offers no place to record the difference. The receiving branch enters 96, the source shows 100 dispatched, and the four units become an unexplained adjustment that everyone quietly absorbs. Repeat that fifty times a year and the shrinkage is material.
You cannot promise incoming stock. A salesperson at the receiving branch should be able to see that 100 units arrive Tuesday and commit them to a customer. In-transit visibility makes that possible.
Rules that make transfers work
- The receiving location requests. The source approves. Push-based distribution, where head office decides what each branch gets, systematically over-supplies slow branches and starves fast ones.
- Scan on dispatch and scan on receipt. Both ends. Typing quantities at either end defeats the purpose.
- Variances are posted, never absorbed. A short receipt creates a variance record with a reason code: damaged, short-shipped, or missing. Reason codes turn a pile of small losses into a pattern you can act on.
- In-transit stock is visible but not sellable. Show it in reports, exclude it from available-to-promise at both ends unless you have explicitly enabled forward commitment.
- Transfers cost money and should show it. If freight is material, attach it to the transfer so landed cost at the destination reflects reality.
Setting Reorder Points Per Location
A reorder point is the stock level at which you order more. Set it once for the company and you guarantee one of two failures at every branch.
The starting formula is straightforward:
Reorder point = (average daily demand at this location × lead time in days) + safety stock
Three inputs, each of which must be location-specific.
Average daily demand. Calculate per location, over a window that reflects your business. Thirty days is reasonable for stable retail. Use longer windows for lumpy demand and shorter ones for fast-changing categories.
Lead time. The gap between raising a replenishment and having sellable stock at that location. For a branch supplied from your own godown, this is transfer time. For a branch supplied directly by a vendor, it is supplier lead time. These are frequently very different numbers, which is exactly why one company-wide reorder point fails.
Safety stock. Cover for demand variability and lead-time variability. Branches with erratic demand or unreliable supply routes need more. A branch two hours from the godown needs less than one that receives a weekly truck.
Seasonality deserves explicit handling rather than a permanent buffer. Demand around Eid, Pahela Baishakh and the winter wedding season does not resemble a monthly average, and a static reorder point either overstocks for eleven months or understocks for one. Recalculate before known peaks rather than carrying the buffer year-round.
Keeping the Numbers Honest
Software records what people tell it. Accuracy comes from the counting discipline around it.
Cycle counting beats the annual count
The annual physical count has one virtue: it is thorough. Its defects outweigh that. It closes the business for a day or more, it is performed by tired people under time pressure, it produces one enormous adjustment nobody can attribute to a cause, and it tells you the size of a problem that has been developing invisibly for twelve months.
Cycle counting spreads the work across the year, counting a subset of items continuously. The advantages compound:
- Discrepancies surface within days of occurring, when the cause is still traceable.
- Counts happen during normal operation, so no trading day is lost.
- Counters are fresh and counting small quantities, so the counts are better.
- Corrections are small and frequent rather than large and shocking.
Weight the frequency by value
Not every item deserves the same attention. Classify your catalogue by annual consumption value, the standard ABC approach:
| Class | Share of items, typical | Share of value, typical | Suggested count frequency |
|---|---|---|---|
| A | Around 20% | Around 80% | Monthly |
| B | Around 30% | Around 15% | Quarterly |
| C | Around 50% | Around 5% | Twice yearly |
Treat these bands as a starting configuration and tune from your own data. Add two overrides regardless of class: count anything with a history of discrepancies more often, and count anything with a high theft risk more often, independent of its value.
Count blind
The counter should not see the expected quantity. If the sheet says 47 and the shelf looks like roughly 47, the count becomes confirmation rather than measurement. Blind counts, where the system quantity is revealed only after entry, produce meaningfully more useful results. Where a variance exceeds a threshold, recount before adjusting.
Five Metrics Worth Tracking Per Location
Company-level averages hide the branch that is failing. Track each of these by location.
Inventory record accuracy. The percentage of counted items where the counted quantity matches the system quantity exactly. Measure it by item, not by total units, since offsetting errors flatter a unit-based figure.
Stockout rate. The percentage of items unavailable at a location when a customer wanted them. Harder to capture than it sounds, because unfulfilled demand often leaves no record. A practical proxy is the count of days each A-class item spent at zero stock.
Inventory turnover. Cost of goods sold divided by average inventory value, calculated per location. A branch with visibly lower turnover than its peers on similar products is holding the wrong stock.
Days of stock on hand. More intuitive than turnover for branch managers. Ninety days of cover on a fast-moving item is a signal you can act on immediately.
Transfer variance rate. The percentage of transfers arriving with a discrepancy, tracked by route and reason code. This is the metric that finds problem routes and, occasionally, problem people.
Set a review cadence and hold to it. Weekly for stockouts on A-class items, monthly for accuracy and variance, quarterly for turnover. Metrics reviewed inconsistently stop being metrics.
Common Failure Patterns
The unofficial branch stash. A branch manager holds back stock outside the system to cover for shortages. It is well intentioned and it destroys accuracy for everyone. The fix is not a rule, it is faster replenishment so the workaround becomes unnecessary.
Adjustments without reason codes. If staff can adjust stock with a free-text note or no note, adjustments become the drain that all errors flow into. Require a reason from a fixed list, and review the reason mix monthly.
Same SKU, different codes per branch. Usually a legacy of branches that ran independently before consolidation. Until the codes are unified, no consolidated report means anything. Fix this before you go live on a shared system.
No cut-off discipline at period end. Goods received on the 31st and invoiced on the 1st, or dispatched in one period and received in the next, will misstate both periods unless there is a rule everyone follows.
Treating the storefront as a separate universe. If you sell online as well as in-store, the online channel must draw from the same location-level stock, with an explicit rule about which location fulfils online orders. Otherwise you will oversell, which is more expensive than a lost sale because it costs a refund, a delivery and a reputation.
Frequently Asked Questions
How many locations before I need dedicated software?
Two. The transition point is not a number of warehouses but the existence of transfers between them. Once stock moves between locations under your ownership, a spreadsheet has to model the in-transit state manually, and it will not survive contact with a busy week.
Should each branch keep its own stock file?
No. Separate files per branch guarantee divergent product codes, no consolidated view, and no way to see whether the stock a customer wants exists elsewhere in your business. One shared item master, with quantities held per location, is the structure that works.
How do I handle stock that belongs to a different legal entity?
If you operate several registered businesses, their stock and books should be genuinely separated, not merely filtered. Each entity needs its own ledger, its own VAT position and its own reporting. That is an architectural question, covered in multi-tenant versus single-tenant inventory software.
What inventory accuracy should I target?
Aim for consistent measurement before you aim for a number. Most operations that begin measuring honestly discover their true accuracy is well below what they assumed. Establish the baseline per location, then set improvement targets against your own starting point. A branch improving from 82% to 91% is doing better work than a branch holding a comfortable 95% that was never verified.
Do I need barcode scanning for multi-location stock to work?
It is not strictly required, but manual entry at both ends of every transfer is where accuracy erodes fastest. Scanning on dispatch and receipt removes an entire class of error. Setting up a barcode inventory system covers what the hardware and label workflow actually involve.
How do I decide which location fulfils an online order?
Set an explicit rule rather than leaving it to whoever sees the order. Common rules are nearest location with stock, cheapest to ship, or a designated fulfilment location with others as fallback. The requirement is that the rule is in the system, so stock is reserved automatically and two channels cannot sell the same unit.
Where to Start
If you are running more than one location today without location-level stock, do these three things in order. Unify your product codes across branches. Perform a physical count at every location on the same day and use those figures as opening balances. Then implement the five-state transfer workflow before anything else, because transfers are where the accuracy you just established will leak away.
After that, set reorder points per location, start cycle counting your A-class items monthly, and review accuracy and transfer variance by branch every month.
For the wider system decision, the Bangladesh inventory software buyer's guide covers evaluation and pricing. If your locations issue VAT invoices under different registrations, Mushak 6.3 invoicing from your POS covers the compliance side.
PicoStore tracks stock per warehouse and per outlet, with transfer documents, in-transit visibility and per-location reporting on every plan. Start free and set up your second location in an afternoon.
Sources
- IHL Group inventory distortion research, reported by Retail Insight Network
- National Retail Federation returns data, reported by Retail Insight Network