Questions to Ask About Warehouse Operations
Twenty questions for assessing a warehouse: throughput, inventory accuracy, picking method, safety record, labor turnover and cost per order. Written for managers, auditors and anyone touring a site they will have to depend on.
The questions
Open any question for the note
How many order lines does this site ship on an average day, and what was the busiest day last year?
Why ask it
Order lines are the honest unit of work; counting orders or pallets hides huge variation in size. The gap between average and peak tells you how much of the year the building runs stressed, and whether staffing is built for the average or the spike.
What is your inventory accuracy right now, and how do you measure it?
Why ask it
Ask for the number before the method. Sites quoting 99.9 percent are often measuring by location rather than by unit, which flatters the result. Anyone who cannot give a current figure is not counting often enough to have one.
How often do you count stock, and is that cycle counting or a full annual count?
Why ask it
Cycle counting surfaces errors within weeks. A once-a-year wall-to-wall count means the number everyone relied on all year was accurate on one day in January and drifting from then on.
Which warehouse management system are you running, and what version?
Why ask it
Version matters as much as vendor. A heavily customised or several-releases-behind install cannot absorb vendor updates, which quietly caps what the site will be able to do for years.
Which parts of the process still run on paper or spreadsheets alongside the system?
Why ask it
The spreadsheets are where the real process lives. Every hand-built workaround marks a gap in the system, and those files usually depend on one person who knows the formulas and has not documented them.
How is picking organised here: discrete, batch, zone, or wave?
Why ask it
Naming the method is the easy part. What you are listening for is whether it matches the order profile. Batch picking a site full of single-line orders wastes sortation effort; discrete picking a high-volume site wastes travel.
What do a new picker's first two weeks look like?
Why ask it
Onboarding length is a proxy for how much of the job lives in people's heads. Two days of shadowing followed by a live rate target usually explains the site's error rate without anyone having to say so.
What is your turnover rate on the floor, and how does it compare with two years ago?
Why ask it
High floor turnover resets the learning curve continuously and shows up later as mispicks and strains. The trend matters more than the level: turnover rising in a stable labor market points at supervision rather than pay.
How many recordable injuries have you had in the last twelve months, and what were they?
Why ask it
Ask for incidents, not for a safety philosophy. The pattern in what actually happened, forklift near misses at one aisle end, repeated strains at a badly placed pick face, points at a specific part of the layout.
Who is allowed to stop work if something looks unsafe, and has anyone actually done it?
Why ask it
Every site says workers can stop unsafe work. The second half of the question is the real one. If nobody can recall an example, or the person who did it had a difficult month afterwards, the policy is decoration.
What is your cost per order or cost per line, and what is included in that number?
Why ask it
The figure means nothing until you know its contents: whether it carries supervision, facility cost, packaging and returns handling, or only direct picking hours. Comparisons between sites usually founder here.
Where does work back up on a busy afternoon?
Why ask it
Bottlenecks move through the day, so a general question gets a general answer. Naming a specific busy afternoon gets you a place: packing benches, dock doors, one label printer, the outbound staging lane.
How full is the building, measured both by pallet positions and by cube?
Why ask it
The two numbers diverge sharply. A site 95 percent full by position and 60 percent by cube is storing air, and the answer is racking or reslotting rather than the extra square footage they are probably asking for.
What share of your SKUs has not moved in six months?
Why ask it
Dead stock occupies prime pick locations and lengthens every walk in the building. A large share also indicates that purchasing and the warehouse are working from different assumptions about demand.
How are returns handled, and how long does a return sit before it is back on the shelf?
Why ask it
Returns are where margin disappears quietly. Dwell time is the number that matters: goods sitting in a returns cage for three weeks are unsellable inventory that the system still counts as available.
What is your on-time ship rate, and how exactly is the clock defined?
Why ask it
On-time is measured against customer promise, carrier cutoff, or the order's own release date, and those give very different results. Unusually strong numbers usually mean the easiest of the three was chosen.
What happens when the system goes down mid-shift?
Why ask it
Contingency plans get written and never rehearsed. Listen for whether printed fallback pick lists exist, whether current staff have run the manual process, and when it was last used rather than last documented.
How do you staff peak season: overtime, temps, or a second shift?
Why ask it
Each option has a signature failure. Overtime exhausts the experienced crew, temps raise error rates for weeks, a second shift needs supervision the site may not have. The answer tells you which failure they have accepted.
Which supplier or carrier creates the most rework in this building?
Why ask it
Inbound quality decides how much labor goes on repalletising, relabelling and breaking down badly mixed loads. A site that can name the worst offender is measuring the cost; one that cannot is absorbing it.
If you had budget for one change next year, what would you spend it on?
Why ask it
Answers divide between space, equipment, and people. What they choose shows which constraint they feel daily, and holding it against their earlier answer about bottlenecks shows whether they have diagnosed their own site correctly.
How to read a warehouse
Practical guidance for the conversation itself
Before the walk
Get the volume numbers in writing first
Ask for daily order lines, units shipped and headcount for the last twelve months before you visit. Reading the seasonality yourself means you arrive knowing which weeks were hard, and you can ask what happened in those weeks rather than being told the site runs smoothly.
Decide whether you are assessing capacity or competence
They need different questions. Capacity is space, racking, dock doors and equipment, and is largely fixed. Competence is accuracy, training and how the site behaves on a bad day, and can change with management. Mixing the two produces a tour that answers neither.
Ask to see the site at its worst hour
Mid-morning on a quiet Tuesday tells you almost nothing. Late afternoon before a carrier cutoff, or the first day back after a holiday, shows you where the process actually strains.
What to look at while you are inside
- The floor at the end of aisles. Loose product, orphan pallets and stacked totes mark places where the process has no defined home for something.
- Pick face heights for the fastest movers. If the top sellers are at ankle or overhead height, nobody has reslotted in a long time.
- Handwritten signs. Every one of them is a rule the system does not enforce.
- The returns and damages area. Its size relative to throughput is the clearest single indicator of upstream problems.
- Whether anyone is waiting. Pickers queuing for a scanner, a printer or a forklift is capital sitting idle in plain view.
- Battery and charging areas for the fleet. Equipment availability constrains more sites than they admit.
- How supervisors are spending their time: on the floor solving problems, or in the office rekeying data.
Where the numbers get flattered
Accuracy measured the generous way
Location-level accuracy, or accuracy counted only on SKUs that were cycle counted, will always look better than unit-level accuracy across the whole building. Ask which denominator is being used, and whether adjustments are netted off against each other.
Productivity quoted as an average
Lines per hour across all pickers hides the spread. A site where the best picker is three times the slowest has a training problem that the average conceals entirely.
On-time measured after the fact
If ship dates can be revised once an order is late, the on-time figure measures the revision process rather than the shipping. Ask whether the promised date is frozen at order entry.
Safety reported as training hours
Hours of training delivered is an input, not a result. Recordable incidents, near-miss reports and days lost are the outputs, and sites with very low near-miss counts are usually under-reporting rather than unusually safe.
Follow-ups that get past the tour script
- 1Ask a picker, not a manager, what slows them down most. Do it away from the supervisor.
- 2Ask what changed after the last bad peak season, and whether it stuck.
- 3Ask for the last three inventory adjustment reports rather than the summary figure.
- 4Ask who decides a slotting change, and how long the last one took to get approved.
- 5Ask what the site is measured on by its own head office. That list, not the tour, is what the management team optimises for.