Real projects run in Taktly — DMAIC, Lean & 3PL examples

DMAIC · Pharmaceutical QC / Stability Lab

OOS investigation cycle time at QC Stability Lab

From 18.2 days to 5.8 — 92% of investigations within the 7-day SLA, ~$255K/yr recovered.

Problem

OOS (Out-of-Spec) investigation cycle time at the QC Stability Lab averaged 18.2 days over the prior quarter against a 7-day SLA. 60% of investigations breached the SLA, delaying batch release an average of 9 days per affected lot and driving ~$285K/yr in customer credit notes and delayed-release inventory carry. Off-site stability sample storage went live the previous quarter; the OOS Investigation SOP was never updated to reflect the new retrieval workflow.

Goal

Reduce mean OOS investigation cycle time from 18.2 days to ≤7.0 days with ≥90% of investigations within SLA, sustained for 8 consecutive weeks.

Mean OOS investigation cycle time

Baseline

18.2 days

Target

≤7.0 days

Result

5.8 days (8 wk sustained · 92% within SLA)

Key artifacts generated in Taktly

Project Charter

Problem, scope (QC Stability Lab · all instruments · all product lines), 18.2 → 7.0 days, $255K/yr savings model, sponsor-approved.

Process Map (Operator → Supervisor → Quality)

9 steps across 3 lanes. Steps 3–4 (sample retrieval) flagged as the cycle-time loss zone (4–6 days of every cycle).

5-Why Root Cause

No named owner for the OOS Investigation SOP since the QA Documentation Manager role was eliminated. A LIMS auto-retrieval feature ready for 14 months sat behind a missing signature.

Standardization Plan

SOP v3 published · LIMS auto-trigger enabled · QA SOP Owner role created · 12 high-volume SKUs pre-staged on-site · Tier-1 KPI on QA huddle board.

What we did

Charter + 5W2H + Define Observations across 5 sources (Lab Manager, Senior Analyst, Sample Custody, LIMS Admin, QA Director) — all five voices converged on the SOP-currency gap.
Pareto of 184 cycle-time overruns: sample retrieval delay = 49% of the loss; top 3 causes = 82%. Run chart confirmed the process was stable but operating above the 7-day SLA every week.
Hypothesis ledger: H1 (retrieval delay) + H2 (SOP not updated) confirmed; H3 (LIMS feature dormant) marked Contributing — both Yes and Contributing flowed into the 5-Why chain.
Improve register with CA/PA/Remedial classification: enable LIMS auto-trigger (Corrective), publish SOP v3 (Corrective), name a QA SOP Owner with quarterly cadence (Preventive), pre-stage SKUs (Remedial), 60-day daily huddle (Remedial). CAPA quality gate passed 9/9 on selected actions.
Welch's t-test on before vs after cycle times: p < 0.001, Cohen's d > 1.5 — improvement confirmed and practically large. Customer credit notes for QC-related delays trended to zero by week 6.

Outcome

Mean OOS investigation cycle time dropped 68% (18.2 → 5.8 days) and held there for 8 consecutive weeks. SLA achievement rose from 40% to 92%, exceeding the ≥90% success criterion. Annualized cost recovery: ~$255K from eliminated customer credits and faster batch release. The QA SOP Owner role is now active; the same SOP-currency review cadence is being rolled out across the other two stability lab sites.

Lean · Food & Beverage

Beverage filling line changeover

Cut SKU changeover from 78 minutes to 24 — four extra hours of output every shift.

Problem

A high-speed beverage filling line averaged 78 minutes of downtime per SKU changeover across 6 changeovers per 12-hour shift. That's 7.8 hours of lost output per shift — capping volume right as the seasonal demand spike hit and pushing customer fill rates below contract.

Goal

Cut average changeover time from 78 min to ≤25 min within one 5-day Kaizen event, sustained for 30 days.

Average changeover time

Baseline

78 min

Target

≤25 min

Result

24 min (30-day avg)

Key artifacts generated in Taktly

Value Stream Map (current state)

9 changeover steps mapped with cycle time, wait time, and operator count per step.

Waste Walk (DOWNTIME)

22 observations costed — top waste: operator waiting on QA chemistry release (47 min/cycle).

Future State VSM

Parallel QA chemistry release, pre-staged tooling cribs, SMED internal → external.

Standard Work

New 14-step changeover SOP with time, key point, quality check, and safety watch-out per step.

What we did

5-day Kaizen run in Taktly's Lean workspace — VSM, waste walk, future state, standard work.
DOWNTIME analysis identified $420K/yr in wait-time cost, mostly QA chemistry release delays.
Pre-staged tooling cribs at the line, parallel QA chemistry release, operator SMED training.
Standard Work for the new changeover rebuilt with quality and safety watch-outs at each step.
30-day sustainment audit: 94% compliance to the new SOP.

Outcome

Changeover time dropped 69% (78 → 24 min). Capacity unlocked: 4 additional hours per shift. Fill rate moved from 89% to 97%. Annualized benefit: $720K in recovered labor + new revenue.

DMAIC · Warehouse / 3PL distribution

Zone 3 pick error rate at Building A

From 6.8% to 1.4% — look-alike SKU confusion eliminated through slotting + RF gun warning.

Problem

Pick error rate on wave picking in Zone 3 of Building A climbed from a 1.8% historical baseline to 6.8% over 8 weeks of Q4 holiday inventory load-in. ~410 errors per month escaped to shipping, driving a 3× increase in customer complaint volume and ~$42K/yr in re-pick labor + return freight cost. No equipment, system, or staffing change on the picking side; only the Q4 SKU slotting had changed.

Goal

Reduce Zone 3 daily pick error rate from 6.8% to ≤2.0%, sustained for 8 consecutive weeks.

Zone 3 daily pick error rate

Baseline

6.8%

Target

≤2.0%

Result

1.4% (8 wk sustained · 8/8 weeks under target)

Key artifacts generated in Taktly

Project Charter

Problem, scope (Zone 3 wave picking only · all 3 shifts), 6.8% → 2% baseline-vs-target, $42K/yr savings, sponsor + process owner named.

Process Map (Operator → Supervisor → Quality)

10 steps across 3 lanes. Cause-type column flagged 2 steps (slot scan + tote scan) where the workflow silently accepts the wrong SKU.

5-Why Root Cause

No look-alike-segregation rule in the Slotting Playbook + no named owner. Q4 SKUs were slotted by bin capacity only, placing visually-similar pairs adjacent.

Standardization Plan

Slotting Playbook v2.1 with look-alike rule · 12 high-risk pairs relocated · RF gun warning on look-alike SKU codes · Slotting Standard Owner named with quarterly cadence.

What we did

Charter + 5W2H + Define Observations across operator, supervisor, and inventory-control voices — all three lanes converged on look-alike SKU confusion at the pick face.
Pareto of 842 pick errors: 'Wrong SKU — look-alike confusion' = 49% of the loss. Run chart showed a sustained climb starting at the Q4 load-in date. Stratified by zone confirmed Zone 3 was the only outlier.
5 hypotheses tested in the Prioritized Cause Analysis ledger: H1 (look-alike SKU adjacency) confirmed with evidence; picker fatigue, RF gun mis-scan, training gap, and bin-label fade were ruled out with reasons.
Improve register with CA/PA/Remedial: enable RF gun look-alike warning (Corrective), publish Slotting Playbook v2.1 (Corrective), name a Slotting Standard Owner (Preventive), relocate 12 look-alike pairs (Remedial), 60-day daily look-alike audit at 2nd-shift handoff (Remedial). CAPA quality gate 9/9 on selected actions.
Welch's t-test on before vs after Zone 3 daily error rate: p < 0.001, Cohen's d > 2.0 — both statistically significant and practically very large. Customer complaint volume back to pre-Q4 baseline within 3 weeks.

Outcome

Zone 3 daily pick error rate dropped from 6.8% to 1.4% and stayed below the 2% target every week for 8 consecutive weeks. Annualized recovery: ~$42K in re-pick labor + return freight + customer credits. The Slotting Standard Owner role is now permanent; the look-alike-segregation rule is propagating to Zones 1, 2, and 4 and to Building B operations next quarter.

See what actual Taktly projects look like.

OOS investigation cycle time at QC Stability Lab

Beverage filling line changeover

Zone 3 pick error rate at Building A

Your project is next.