Picks per Hour and Lines per Hour: Measuring Picker Productivity

The weekly operations review shows zone B at 74 picks per hour and zone D at 131, and someone asks why zone B cannot "just pick faster". Nobody in the room can say whether the two numbers count the same thing. Zone B picks cartons of detergent, one line at a time; zone D picks phone accessories, several units per line. Zone B also lost forty minutes to a late replenishment that the report booked as picking time.

A pick rate is only useful when its numerator and its denominator are defined, sourced and cleaned the same way every week. Done properly, the figure separates three questions that the raw number mixes together: how much work arrived, how long the work should have taken, and how much of the shift was actually available for it. That separation is what lets a manager coach a person, fix a slot or reprice a contract with confidence.

• What this guide covers: choosing the counting unit, building the hour buckets, writing an engineered standard, coding indirect time, and the US and Indian rules that apply when a rate becomes a quota.
• What you can reuse: a formula block, a worked shift in dollars and rupees, an incident trace from a real-looking second shift, and a checklist for the first published report.

Units, lines, orders and totes: fix the counting unit first

Decide what one "pick" means in each zone before you calculate anything, and write it on the report. Most disagreements about picker productivity are not about effort; they are about two teams counting different things and calling both of them picks.

A warehouse management system usually records a pick task with a location, a SKU, a requested quantity and a confirmed quantity. From those four fields you can derive at least four output counts. Each count answers a different operational question and reacts differently when the order mix changes, so the choice should follow the work content of the zone rather than habit.

• Units or eaches per hour: the confirmed quantity summed across tasks. It suits single-unit e-commerce where nearly every line is one unit, and it flatters zones where one scan confirms twelve identical units.
• Lines per hour: one SKU, from one location, for one order or batch. It reflects the fixed effort of travelling to a face, finding the slot and scanning, which dominates most mixed case and each picking.
• Orders per hour: useful only where orders are small and similar, such as a dark store. It swings wildly when the lines per order shift between a weekday and a promotion weekend.
• Totes, cases or pallets per hour: the right unit for full-case and pallet picking, where the handling of each physical piece drives effort and the line count says little.

Treat short picks consistently. A picker who walks to a face and finds it empty has done most of the work of a line, so the line should count with zero units, and the shortage should be visible as its own exception. Excluding short lines makes a replenishment failure look like a slow picker; counting them at full quantity hides the shortage from inventory control.

Which hours belong in the denominator, and who owns each one

Use at least two denominators, paid hours and coded direct hours, and never mix them across zones. The hours come from a different log than the output, usually the time-and-attendance clock rather than the WMS, so their ownership has to be explicit before the rate is trusted.

A picker's shift breaks into layers. Paid hours include paid rest breaks; in the United States, 29 CFR 785.18 treats short rest periods of roughly five to twenty minutes as hours worked. Inside paid time sit coded direct hours (logged into a pick assignment), coded indirect hours (replenishment waiting, equipment checks, meetings), and the remainder that nobody coded. The table names who keeps each record and who approves changes.

Measure coverage before you compare anyone. Coverage is coded hours, direct plus indirect, divided by paid hours less scheduled paid breaks. When coverage falls below about 95%, the unexplained time is large enough to move a picker's rate by more than any coaching would, and the ranking says more about who forgot to scan into indirect codes than about who works well.

Engineered standards versus observed averages, and why you need both

An observed average tells you what happened; an engineered standard tells you what should have happened for that exact work. Report performance against the standard, and keep the observed average as a trend line for capacity planning.

Observed rates bake in last month's problems. If a zone waited twenty minutes a shift for replenishment, the average absorbs that wait and calls it normal. An engineered standard is built bottom-up from timed elements: walking or driving between faces, a fixed time per line for locating, scanning and confirming, a variable time per unit or case, and a percentage allowance for personal needs, fatigue and unavoidable delay. Because each assignment earns its own standard minutes, a picker given long-travel lines is not penalised for the slotting.

Observed lines per direct hour   LPH_obs  = L / H_direct
Observed units per direct hour   UPH_obs  = U / H_direct
Paid-hour rate                   LPH_paid = L / H_paid
Coverage                         C        = (H_direct + H_indirect) / (H_paid - H_breaks)   target >= 0.95

Engineered minutes for one assignment a:
  T_std(a) = [ D(a) / v  +  n_lines(a) * t_line  +  n_units(a) * t_unit ] * (1 + PFD)

Performance                      P        = SUM_a T_std(a) / (60 * H_direct)

where
  L, U        confirmed lines and units (short picks count as lines with 0 units)
  H_paid      paid hours from time and attendance, including paid rest breaks
  H_breaks    scheduled paid rest breaks, removed only for the coverage test
  H_direct    hours coded to pick assignments in the labour log
  H_indirect  hours coded to approved indirect activities
  D(a)        travel distance for the assignment in metres, from the location master
  v           travel speed in metres per minute, measured per equipment type
  t_line      minutes per line (locate, scan, confirm), from time study
  t_unit      minutes per unit or case handled, from time study
  PFD         personal, fatigue and delay allowance as a fraction, for example 0.12

Keep the standard honest by tying each element to evidence. Travel speed comes from timed walks with the actual cart or truck; per-line time comes from filmed or stopwatch studies of a sample of pickers across shifts; the allowance is a documented policy decision, not a tuning knob. When the standard is revised, version it, because last quarter's performance should be judged against last quarter's standard.

Picker performance reporting compares standard minutes earned from WMS pick tasks with direct hours coded in the labour log; a site should publish rankings only when coded hours cover at least 95% of paid hours, and US paid rest breaks remain hours worked under 29 CFR 785.18.

Indirect time: the hours that quietly drag the headline rate down

Indirect time is real work that produces no pick line, and it belongs in its own codes so the pick rate measures picking. Hide it inside direct time and you will coach pickers for problems that the replenishment team or the equipment pool caused.

Start with a short list of codes that a picker or supervisor can select in seconds on the scanner, and audit how often each is used. Too many codes produce guesswork; too few push everything into an "other" bucket. The codes below cover most each and case picking operations, and each one points to a different owner when it grows.

• Waiting for replenishment: the face is empty and a replenishment task is open. The owner is the replenishment lead, and the trigger quantity or the release timing needs review.
• Equipment check and swap: pre-use truck inspection, battery change or scanner swap. Measure it per shift and compare with the planned allowance.
• Start-up and shutdown: walking to the zone, collecting carts, the pre-shift briefing and returning equipment at the end.
• Housekeeping and returns-to-stock: clearing cardboard, re-shelving items from cancelled orders or putting back overpicks.
• Training and coaching: time with a trainer or supervisor, coded so new starters are not ranked against experienced pickers.
• System down: WMS, network or printer outage, logged with the incident number so the minutes can be tied to an IT ticket.

A worked shift for a US each-pick zone and an Indian case-pick aisle

The same arithmetic works in both markets; only the unit of work and the cost per hour change. The figures below are illustrative, chosen to show how the direct-hour rate, the paid-hour rate and the cost per line diverge for one picker on one shift.

The US example is an each-pick mezzanine for an e-commerce client, with two paid fifteen-minute breaks and an unpaid thirty-minute meal outside the eight paid hours. The Indian example is a case-pick aisle at a consumer goods distributor, with an eight-hour working day and intervals set by notification. Labour cost per hour is a fully loaded illustrative figure, including benefits and statutory contributions.

Three readings follow from the table. First, the US picker's 100 lines per direct hour and the Indian picker's 60 are not comparable as effort; the standards earned show both performing at about 92%. Second, the gap between the direct-hour and paid-hour rates, roughly a quarter of the shift in both cases, is where cost leaks out. Third, both sites sit just under 95% coverage, so neither should rank individuals this week.

How order profile, travel and slotting set the achievable rate

Most of the variation in lines per hour between zones is explained by travel per line and units per line, not by picker skill. Measure those two drivers before you set any target, and recalculate the target when they move.

When lines per order fall, a picker visits more faces for the same output, and travel per line rises. Batching several orders into one trip reverses that, which is why the release method matters; our comparison of batch and wave picking works through that choice using real order files. Slotting sets the other half: fast movers placed far from the induction point add walking to every line that touches them.

Lines-per-hour targets after a re-slot are recalculated from travel metres per line in the WMS location master and the engineered standard, not from last month's average; where the target is a California quota, Labor Code section 2101 requires a written description before any missed rate justifies adverse action.

The earlier article on warehouse slotting decisions decides which SKUs deserve a forward pick face. This article starts after that decision: the layout is given, and the task is to measure whether people are picking well within it. Use the two together when a zone's rate drops, because a re-slot shows up first as a change in travel distance per line, which the engineered standard captures and the observed average does not.

• Travel per line: metres walked or driven per confirmed line, from the location master and the pick sequence. A rise of 10% usually costs more than any coaching can recover.
• Units per line: confirmed units divided by lines. Report it beside units per hour so that a promotion of multipacks is not read as a productivity gain.
• Pick density: lines per aisle visit. Low density means many trips for little work and points to batching or zone redesign.
• Equipment mix: carts, order pickers and pallet jacks move at different speeds, so the standard needs a travel speed per equipment type.

Quota rules that limit how a pick rate may be applied to people

A pick rate becomes a legal matter once it is used to discipline someone. In California the rules are specific, US federal law decides which minutes are paid, and in India the working-hour limits now sit in a single code.

California's warehouse quota law, Labor Code sections 2100 to 2112, applies to employers with 100 or more employees at a single warehouse distribution centre, or 1,000 or more across the state's centres, including staffing-agency workers the employer controls. The Labor Commissioner's guidance defines a quota as a required productivity speed or quantity of tasks within a defined time period, and gives "200 packages per hour" as an example. Each quota must be described to the employee in writing.

California warehouse quota rules under Labor Code sections 2100 to 2112 require covered employers with 100 or more workers at one centre to describe each pick quota in writing, bar quotas that prevent meal, rest or bathroom breaks, and supply 90 days of work speed data within 21 days of request.

The same guidance says a quota is unlawful if it prevents meal or rest periods, bathroom use including reasonable travel time to and from the bathroom, or compliance with safety standards. An employee may request the written quota and the most recent 90 days of personal work speed data, and the employer must respond within 21 days. Adverse action within 90 days of such a request or complaint triggers a rebuttable presumption of retaliation under section 2105.

• US federal paid time: under 29 CFR 785.18, short rest periods of about five to twenty minutes count as hours worked, so they sit in the paid-hour denominator even though no picking happens.
• India working hours: the Occupational Safety, Health and Working Conditions Code, 2020, effective from 21 November 2025, limits work to eight hours a day under Section 25(1)(a), with intervals and spread-over fixed by notification under Section 25(1)(b).
• Design consequence: a standard built with a realistic allowance and breaks outside direct time keeps a target achievable without skipping breaks, which is what the California test looks at.

Second-shift trace: a 31% rate drop that was a scanner and replen gap

Most sudden drops in picks per hour come from missing time codes or blocked faces, not from people. The trace below follows one evening at an illustrative regional DC and shows where the uncontrolled report went wrong and what the corrected process recorded instead.

1. 16:05, zone B dashboard: the WMS shows 69 lines per direct hour against a usual 100. Four pickers have open sessions but no task-close records since 15:40.
2. 16:12, equipment cage: the four RF guns were swapped at the battery station; the replacements were never logged into the labour module, so the LMS kept booking time to the last pick assignment as direct time.
3. 17:40, aisle B-14: face B-14-03 is empty. Replenishment task RPL-448172 sits in status "released, not started" because the reach truck operator was redeployed to receiving. Three pickers wait 22 minutes, again booked as direct time.
4. 22:30, shift close: the uncontrolled report shows the zone 31% below target and flags three pickers for a performance conversation, with coverage never checked.
5. Next morning, corrected run: the scanner now forces a login after a battery swap, and an empty-face scan opens the "waiting for replenishment" indirect code until RPL-448172 is confirmed. Re-running the shift with corrected codes gives 96 lines per direct hour and 94% performance, and the flags are withdrawn.

Where software can estimate and where the rate must stay rule-based

Keep the arithmetic and the eligibility rules fixed and auditable; let software help people find and explain gaps. An algorithm may propose why a rate dropped, but a supervisor decides what is recorded and a person decides any consequence for an employee.

Labour modules and analytics tools increasingly classify idle time, suggest standards from historical data or summarise scanner logs. Those are useful aids, provided their output is a suggestion that a named person accepts or rejects. The table separates tasks by whether they should run as fixed rules or as assistance.

Think of related systems the same way. A voice or wearable RF set-up changes per-line time because confirmation is faster; our evaluation of voice picking and RF scanner ergonomics covers check-digit design. When incentive pay and engineered standards are on the table, the labour management system throughput and ROI analysis sets out the payback model.

A first-report checklist for a picks-per-hour figure people trust

Publish the first report only after the definitions, sources and thresholds are written down and tested against two weeks of history. The checklist below is the minimum a DC manager and a finance controller should agree before the figure appears on a wall screen or in a client review.

Run the report in parallel with the old figure for two weeks and explain every difference larger than 5% before switching over. Most differences will trace back to uncoded time or a counting rule, and resolving them in the open is how the new number earns trust. For more measurement guides by topic, browse the logistics and supply chain library.

If you are specifying a labour or WMS reporting change and want a second opinion on standards, coverage rules or data extracts, you can send us your requirements and we will reply with a scoped approach for your site.