Continuous vs annual inventory — which pays off more

The classic model: once a year the warehouse stops operations, the supervisor splits the racks into zones, pairs of counters get sheets (paper or terminal) and count everything from the first pallet to the last. A second count for items with discrepancies. A third one if differences persist. Final report, stock corrections, protocol approved by the inventory committee.

A medium warehouse with 3000-5000 SKUs and 12-15 people closes for 2-4 days. With preparation (freezing movements, splitting, briefing), reconciliation and posting corrections - the whole operation takes a week. During that time the warehouse doesn't work or works at half pace. Real cost: 30-80 thousand zlotys of lost margin on sales, plus 8-15 thousand zlotys of overtime, plus organisational stress for the entire logistics department.

The second cost is less obvious: discrepancies. After a year without systematic stock checks the gaps are huge - we've seen warehouses where 8-12% of items had a real stock different from the system one. Corrections at that scale are not just numbers in the balance sheet but weeks of nerves: who is responsible, how to settle the losses, whether to call the police, whether to fire the supervisor.

The idea is simple: instead of counting everything once, we count a small piece every day. Each day the operator gets a list of 10-30 locations to verify. In the morning or at the end of the shift, during a quieter moment, they walk to the rack, scan the location, count the contents and key the number in. The WMS compares this with the system stock and generates a discrepancy report.

Across the year, every location is checked at least once - usually 2-4 times. Discrepancies are detected days or weeks after they appear, not months. This radically changes the economics: a missing pallet detected after two days can usually be tied to a specific operation (pick task, correction, transfer) and explained. A missing pallet detected after nine months is detective work.

The warehouse never stops, not even for a day. Normal operations, plus a few dozen minutes of counting per day. The total cost is significantly lower than annual, while accuracy is higher.

A sensible continuous inventory does not treat all items the same. The ABC classification helps spread the effort where it's needed:

• Class A (typically 5-15% of the assortment, 60-80% of warehouse value) - counted weekly or every two weeks. Even small discrepancies here can distort the balance sheet, and theft hurts the most.
• Class B (15-25% of the assortment, 15-25% of value) - counted monthly or quarterly. Rotating items, but not as critical.
• Class C (60-80% of the assortment, 5-15% of value) - counted annually or twice a year. Small items, slow movers.

Plus a few extra rules: high-value items (above 10 thousand zlotys per piece) counted weekly regardless of ABC. FEFO items (close to expiry) counted at every change. Items with zero system stock checked quarterly - sometimes goods turn up that "should not be there anymore".

Every count can be done in two ways. Verified: the counter sees the system stock and confirms it (or corrects it). Blind: the counter has no access to the stock figure, types in what they counted and the system compares afterwards.

The difference is fundamental. With verified, the counter feels subconscious pressure - "the system shows 12, I see 11, I must have miscounted, I'll type 12". This creates false consistency that fixes nothing. With blind, the counter writes down their result independently - if they miscount, only the validation step finds it, not automatic alignment.

External auditors (KPMG, EY, BDO) require blind count at every audit - otherwise the inventory report has no credibility. In daily cycle counting you may start with verified as a softer onboarding, but after six months we move to blind. This is the foundation of the integrity of the process.

The Accounting Act in art. 26 says: entities are required to perform an inventory once a year, but section 3 allows the continuous variant. Specifically: "entities may carry out inventory using the method of continuous verification, consisting of checking the actual stock and comparing it with the books during the financial year, provided that every item is included in the inventory at least once a year".

From the auditor's standpoint that means continuous fully replaces annual - if it's documented. What is required: a written procedure, a schedule (e.g. ABC with cycles), counting sheets with signatures, discrepancy reports and their resolution, archiving for 5 years. The WMS does all of this automatically - it generates sheets, archives results, produces final reports for accounting.

Auditor practice: most reputable audit firms accept continuous inventory without reservation. They sometimes ask for an additional count of special items (fixed assets, advance payments, fuels) at year-end, but that is not "annual inventory" in the classic sense - it's a verification of a few dozen non-warehouse items.

Equipment choice depends on the scale of the warehouse and the type of goods. A short map:

• Android mobile terminals (Honeywell EDA, Zebra TC, Datalogic Memor) - 2500-6000 zlotys per unit. Standard for regular cycle counting. The operator gets a task list, scans the location and the goods, types the quantity, done.
• Ring scanners (Zebra RS5100, Honeywell 8680i) - 1500-3000 zlotys. Worn on a finger, paired with a smartphone or terminal. Great for counting in cold storage or where the operator needs both hands free (ladders, high pallets).
• Smartphones with a WMS app - for small warehouses or for ad-hoc counts by the supervisor. A 12 Mpix camera reads codes without trouble. Durability is worse than a terminal, but with 2-3 counters it is enough.
• Drones with a camera - for high-bay warehouses (above 8 metres). Eve Air, Skydio, Verity. They fly along the racks, take photos and AI recognises the SSCC. Investment of 30-100 thousand zlotys, but in a warehouse with 30 thousand high locations the payback is six months.
• RFID - reads an entire rack at once. Expensive (40-80 zlotys per active label), but in specific applications (clothing, tools, documents) it is the only sensible way.

Every discrepancy has to be settled, you cannot just "correct and forget". The standard process in Weaver WMS looks like this:

• First count shows a discrepancy. The WMS records: location, item, system stock, counted stock, difference.
• Second count by another person for items above a threshold (e.g. discrepancy value over 100 zlotys, or quantity difference over 5). The second person does not know the system stock or the result of the first count.
• Supervisor decision - if both counts agree, correct. If they disagree, third count or investigation (who recently operated on this location, are there pallets put away in unmarked places).
• In-plus / in-minus correction entered with a reason code: counting error, operational error, theft, damage, no other explanation.
• Manager approval for corrections above a value threshold. Without this approval the correction does not enter the final stock.

Every correction stays in the archive with signatures. After a year, you can generate a "who had the most discrepancies" and "for what reasons" report - this data is invaluable for evaluating the process and the people.

A specific case from our practice: a distributor of electrical materials, 15 people in the warehouse, 3000 SKUs, around 4500 locations. Previously they did an annual count - two days of downtime, 60 hours of work, corrections of 4-7% of stock value.

After the switch to continuous (ABC classification, blind count, Zebra TC22 terminal):

• Daily, 2 people spend about 30 minutes counting (15-25 locations each)
• Class A counted weekly, B monthly, C every six months
• Yearly full coverage, with surplus for A and B
• The warehouse has zero days of downtime due to inventory
• Annual corrections after the first full year: 1.2% of stock value (down from 4-7%)
• Reaction time on a discrepancy: 5-10 days on average (instead of 6-12 months)

After two years the client cancelled the full annual inventory - the auditor accepted it based on the continuous documentation. Annual savings (downtime plus overtime) estimated at 60-90 thousand zlotys, plus a reduction of corrections from 250 thousand zlotys per year to 70 thousand zlotys.

A real comparison for a medium warehouse (3000-5000 SKUs, 15 people, 5000 locations):

• Annual: 2-4 days of downtime (30-80 thousand zlotys of lost margin), 60-100 hours of overtime (8-15 thousand zlotys), corrections at 4-7% of stock value (with 5 million turnover that's 200-350 thousand zlotys per year). Total 240-450 thousand zlotys per year.
• Continuous: 2 people x 30 minutes daily x 250 working days = 250 hours, about 25 thousand zlotys of labour. Corrections at 0.8-1.5% (40-75 thousand zlotys). Investment in terminal and WMS deployment (one-off 15-40 thousand zlotys). Total 80-140 thousand zlotys per year.

A 100-300 thousand zloty annual difference in favour of continuous. Plus data quality - with continuous, management has up-to-date stock every day; with annual, they know the truth once every 12 months. When making purchasing decisions or telling a customer "yes, I have this in stock", the data quality is measurable.

Some companies don't want to go 100% continuous - they run cycle counts during the year but still stop the warehouse for a full count at year-end. Arguments: "the auditor accepts it better", "the supervisor has learned it that way", "we have to do it for tax authorities anyway".

This model works in practice but it is the worst of the three options - it combines the cost of both methods without their benefits. You have to stop the warehouse at year-end and you also have to spend daily time on the cycle count. The only justification for hybrid: the first year of deployment, when the team is still learning the process and the supervisor wants a "safety net" in the form of a full year-end count.

After six months of continuous discipline, hybrid has no justification. An additional annual count is 30-80 thousand zlotys thrown away and unnecessary stress for the team. Better to spend that money on improving the process or buying tools.

• Counting by the person responsible for the zone. The picker knows their zone by heart and subconsciously "corrects" the result. The counter should always be from outside the zone.
• No blind count. The operator sees the system stock and types the same thing - "because the system is probably right". The cycle becomes fiction and everyone knows it.
• Class A counted too rarely. Class A counted quarterly means a 50 thousand zloty shortage is detected after 90 days. That's too late to identify the cause or the responsible person.
• No discrepancy resolution. The system shows a list of differences, the supervisor clicks "accept all" and the topic is closed. Without root-cause analysis nothing changes - the same errors keep coming back.
• Inventory as punishment. Some warehouses give counting to operators who "messed up" - as a disciplinary measure. The effect is the opposite: counting becomes a burden instead of a standard quality procedure.
• No archiving of count sheets. The auditor asks "show me the March count" but only the final stock is in the system. The archive of count sheets should be available for 5 years - the WMS does this automatically if configured.

Continuous wins almost every time. Lower costs, smaller discrepancies, faster reaction to anomalies, no warehouse downtime. The only reason companies keep running annual counts is organisational inertia and the fear of "what will the auditor say". In practice, auditors accept continuous without issues as long as they see a documented procedure and a trail in the system.

Weaver WMS supports continuous natively - automatic ABC schedules, generation of inventory tasks, blind count from the mobile app, archive of count sheets, discrepancy reports and effectiveness metrics. Deployment in a medium warehouse takes 2-4 weeks, full ROI within 6-12 months. Annual inventory is a relic - it's time to treat it like a black-and-white TV: someone still has one, but everyone is wondering why.