First-acknowledgment response in a well-configured digital Andon deployment with mobile routing is under 90 seconds, while traditional stack-light-only installations average roughly 22 minutes to first response, a structural gap confirmed in May 2026 industry data [S2].
Plants that completed the cutover from manual cord-pull or single-channel stack lights to a four-layer digital architecture (stack light, mobile push, Andon board display, and MES-integrated event log) report mean response-to-resolution times reduced by 58% and unplanned downtime cut by 34% within the first operating quarter [S2]. At a mid-size U.S. automotive assembly operation, an unaddressed line stoppage costs roughly $18K per hour, which is why the 4-minute versus 22-minute delta is now treated as a controllable engineering variable rather than a culture problem [S2].
Defining the 2026 Response-Time Stack: Three Tiers, Three Numbers
A digital Andon system in 2026 turns machine conditions, operator requests, and production delays into immediate, actionable signals routed through stack lights, mobile push, and shop-floor displays [S3]. The benchmark that procurement and continuous-improvement teams now track is split into three distinct tiers: first-acknowledgment latency (target under 90 seconds for a well-configured digital Andon with mobile escalation [S2]), average response-to-resolution time (4 minutes at well-configured U.S. plants versus 22 minutes for traditional stack-light-only systems [S2]), and root-cause identification speed (3x faster when Andon event logs are integrated with MES production records than with manual incident logging [S2]).
First-acknowledgment is the latency between the event (a machine stop, defect detection, or operator pull-cord trigger) and a human pressing "ack" on a mobile device or Andon board; it is the metric that determines whether escalation logic fires. Response-to-resolution is the full clock from trigger to the line running again, and it is the metric that determines the dollar cost on the P&L. Root-cause identification is a different clock entirely, measured in days rather than minutes, and it is the one that feeds back into OEE calculations on the plant floor dashboard.
Where the Time Actually Goes: Four-Layer Architecture Breakdown
A production-grade Andon deployment in 2026 is structured as four concentric layers: sensing/trigger, signaling (stack light + board), routing/escalation, and integration (MES, ERP, OEE) [S2]. The physical stack light, red/yellow/green, now represents less than 15% of the functional value of a modern Andon architecture; the remaining 85% is the signal routing and escalation engine, which is the part that compresses the 22-minute traditional response into 4 minutes [S2].
When a stoppage fires in a digital deployment, the same event simultaneously illuminates the stack light, sends a push notification to the assigned team leader's mobile device, posts an alert to the Andon board display visible from the aisle, logs the stoppage with timestamp and station ID in the MES, and starts a response-time clock that escalates to the supervisor if acknowledgment is not received within the configured window [S2]. For continuous-improvement teams already wiring pressure transmitters and flow meters into the same MES for process visibility, the Andon event log becomes a natural second data stream on the same historian.
Manual, Automatic, and Digital: Comparison Across Decision Criteria
Andon systems range from simple manual setups to integrated digital networks, and the choice is now driven by four criteria: first-response latency, data integration depth, escalation logic, and root-cause traceability [S1]. Manual Andon (rope-pull or button-trigger illuminating a single light) sits at the slow end; automatic Andon (sensor-triggered, still single-channel) is faster but still blind to MES; digital Andon (multi-channel, mobile, MES-integrated) is the only tier that hits the under-90-second first-acknowledgment target.
The four options line up against the four criteria as follows. Manual Andon delivers first-response latency measured in minutes (no mobile push, no escalation), zero MES integration, no automatic escalation, and no event log beyond a paper ticket. Automatic Andon cuts the trigger latency to sensor-detection speed (sub-second) but still relies on a human noticing the light, offers no native MES integration, has no configurable escalation window, and produces no searchable event log. Digital Andon with mobile push hits under 90 seconds first-acknowledgment, integrates with MES through OPC-UA or REST connectors, runs configurable escalation timers (commonly 60-180 seconds to supervisor, 300 seconds to plant manager), and produces timestamped event logs that feed OEE dashboards. Digital Andon with mobile push and MES + ERP integration is the tier that hits the 3x root-cause identification improvement and the 34% unplanned downtime reduction, because the event log is the precondition for analytics, not an output of them [S2].
On a 12-month total-cost basis, the inflection point is the second line in a plant: a single-line deployment cannot amortize the MES integration work, while a four-line plant typically breaks even on the Andon-plus-MES bundle within 9-12 months once the $18K/hour stoppage cost is applied to even two or three avoided 22-minute incidents per week [S2].
Escalation Timers and Acknowledgment: Where Plants Misconfigure the Stack
The most common deployment error in 2026 is setting the first escalation timer too long, which collapses the under-90-second benchmark back toward the 4-minute average and erases the 58% resolution-time improvement [S2]. Best-practice configurations, drawn from the same benchmark dataset, use a 60-second window for team-leader acknowledgment, a 180-second escalation to shift supervisor, and a 300-second escalation to plant manager, with parallel push to the on-call maintenance technician for any event classified as a downtime stop rather than a quality hold [S2].
Digital Andon software also integrates mobile alerts and advanced analytics to track downtime patterns, classify events by reason code at the moment of acknowledgment, and drive long-term continuous improvement, which is the layer that turns a real-time alert system into a closed-loop engineering tool [S7]. On the data-quality side, Andon event logs share the same flatline and gap rules as any other sensor stream feeding the MES: a station that reports zero events for an entire shift is not a 100% OEE line, it is a dead Andon node, and the same gap-detection logic that flags a stalled flow-meter signal will flag a stalled Andon trigger [S2].
What Digital Andon Is Not For: Three Failure Modes
Digital Andon is not a substitute for root-cause engineering, and the benchmark numbers above do not survive a plant that treats Andon data as the endpoint rather than the input. Roughly 68% of recurring line stoppage causes are identifiable from Andon event logs within the first 30 days of deployment, but only if engineering actually reviews the reason-code distribution weekly; plants that install the system and walk away see the 58% improvement decay back toward 20% within two quarters [S2].
Digital Andon is also not a safety alarm system and should not be specified to SIL-rated time relay logic for E-stop functions; the latency profile (under 90 seconds to acknowledgment, 4 minutes to resolution) is incompatible with the sub-second response required for machine-guarding, and conflating the two systems creates audit risk under ISO 13849 and IEC 62061. Finally, digital Andon is not a single-vendor stack: the four-layer architecture presupposes that the stack light, mobile client, Andon board display, MES, and OEE dashboard can exchange event data through standard interfaces, and a procurement spec that locks all four layers to one vendor typically pays a 20-30% license premium for the same first-acknowledgment benchmark [S2].
Sourcing, Standards, and Trackable Signals
The May 2026 benchmark figures (under 90 seconds first-acknowledgment, 4-minute response-to-resolution, 22-minute legacy baseline, 58% resolution-time reduction, 34% unplanned downtime reduction, $18K/hour unaddressed stoppage cost, 68% recurring-cause identification in 30 days, 3x root-cause speedup with MES integration) are reported by a single vendor dataset and should be treated as directional rather than industry-wide until independent multi-vendor surveys replicate them [S2]. The general architectural claims (four-layer design, mobile push as the dominant compression mechanism, MES integration as the precondition for analytics) are corroborated across multiple 2024-2026 sources [S2][S3][S4][S5][S7].
Trackable signals to watch through the rest of 2026: independent OEE-and-Andon survey data from at least one non-vendor analyst firm; published response-time benchmarks from automotive-tier-1 and discrete-electronics plants outside the United States; and any revision to the IEC 62264 (ISA-95) batch-and-event reporting model that touches reason-code classification for Andon events. Plants that want to validate the benchmark on their own floor should instrument one line with the four-layer stack, run it for 30 days, and compare their first-acknowledgment distribution against the under-90-second target before scaling to additional cells.
This topic is covered further in Sizing a Pneumatic Cylinder for 500 N Push: Bore, Stroke, and Safety Margin.