Which should I implement first: APS or MES?

Shopfloor production control has always been central to industrial management.Productivity, repeatability, Taylorism... all of this is ingrained inindustrial management. The MES (Manufacturing Execution Systems) systems emerged to bridge the gap betweenERP data and shop floor reality, giving manufacturers a clear view ofproduction execution and material flow in real time. Though the category firstappeared in the 1990s, it gained much wider adoption in the following decade.At its best, MES tells you exactly what has happened and what is happening onyour factory floor.

Production tracking is one of the coreoutputs of any MES solution, and it's essential for scheduling. You need toknow what has already been produced to avoid scheduling it again — which wouldtie up capacity that should be available for something else. When thatinformation is wrong, the schedule starts from false premises and generates acascade of bad decisions. Simply put, an accurate, current picture of executionis the foundation of good scheduling.

Accuracy matters just as much astimeliness. This applies to inventory entries, stock positions, and materialmovements — an incorrectly recorded quantity can trigger a replenishment orderat exactly the wrong moment.

However,this premise of an updated executed scenario oftenraises doubts aboutthe frequency, detail, and accuracy required, from a scheduling perspective.In practice, though, this doesn't mean you need real-time data to scheduleeffectively. The right question isn't whether your data is perfect — it'swhether your current process meets the minimum standard your scheduling cadencerequires.

If your company schedules production once aday at 7:30 AM, all you need is for execution data to be current at thatmoment. If there's a shift change just before that time and your team has aconsistent process for entering production data before the handover, you canrun the schedule without any issues. The same applies if you reschedule at anafternoon shift change.

For a proactive scheduling process — onewith a defined cadence and a reasonable level of maturity — the absence ofreal-time execution data rarely causes meaningful problems.

The example below compares three scenarios:Scenario 1 is the initial schedule at the start of production; Scenario 2represents a subsequent scheduling cycle with entries recorded via MES; andScenario 3 uses entries recorded via ERP. The assumption is that MES enablesmore frequent updates and reduces the risk of human error from manual entry.Here’s what that looks like in practice.

[caption id="attachment_3695" align="aligncenter" width="725"]

                                                                                  Execution scenarios and notes

Lookingat the longest posting delay — operation A on resource R1 — Scenario 2 showsmore accurate data, with one additional posting compared to Scenario 3, whichcarries a larger remaining production balance. That said, this difference didnot change the sequencing for that resource

Foroperation G on resource R3, where an entry error occurred, Scenario 3 (via ERP)shows more units recorded than were actually produced, resulting in a lowerremaining balance than Scenario 2. This shifts the completion forecast by a fewhours. But again, in this specific example, it didn't change the underlyingschedule logic or sequence

The takeaway: the core scheduling logic andsequence hold up even with imperfect data entry. If running jobs in aparticular sequence reduces changeover time, the system will still find thatsequence — minor deviations in timing or quantity notwithstanding. That said,large data errors or significant delays can cause the sequence to change. Hereare the factors that raise that risk:

• Short cycleand processing times, which generate a high volume of scheduling events andsignificantly expand the range of sequencing options (for example, an operationin a bottleneck sector may have average processing times of 15–20 minutes,which means a very large number of orders scheduled within a single shift);
• A high number of operations per order with similar cycle times. When a producthas ten or more macro-operations of roughly equal duration, even smalldeviations can have a significant domino effect. In facilities with clearerbottlenecks, sequencing is generally more stable, since the system's behaviordepends on a smaller set of critical steps.
• Operationswith many eligible resources. The more machines or workstations that canexecute a given operation, the more the sequence can shift as executiondeviates from the plan. When an operation has only two or three machineoptions, this risk is much lower.
• Frequentcorrective maintenance and unplanned downtime, which can significantly reduceschedule adherence relative to actual output
  ‍

Inother words,proactive scheduling, the kind that has a set day andtime, can usually be done without an MES system, provided execution data isrecorded manually with the same frequency as your scheduling cadence —typically daily or by shift

Anotherkey decision when implementing MES is which processes to instrument with datacollection points. Monitoring every operation is often impractical, so it'sworth identifying which processes are most critical in terms ofcapacityandreliability.Those are the ones that justify the investment in dedicated data collection.For everything else, data entry can often follow a backflushlogic: if adownstream operation has been recorded, the preceding operation — without itsown collection point — is assumed complete. Ideally, collection points shouldbe positioned so that no production order goes more than one shift without atleast one data entry

Evenwith a solid proactive schedule in place, you still need the ability toreschedule when something unexpected comes up — what we call reactivescheduling. It’s a valuable tool for understanding the full ripple effect ofany change to a production order across the factory. But it should be usedsparingly, and particularly not become unpredictable — a condition we call a"nervous" schedule. When priorities are shifting every hour betweenthe scheduler and shop floor supervisors, the plan loses credibility and errorsmultiply. Reactive scheduling is a tool best used sparingly

Ascheduling process built primarily around reactive responses to short-termsurprises isn't a real process — it's firefighting. Companies that findthemselves in constant reactive mode usually have a deeper underlying problem:a weak maintenance plan causing frequent breakdowns, or poor coordinationbetween Sales and Production Planning and Control (Production Planning and Control), leadingto constant priority changes. Implementing  APS (APS generally helps bring stability to these cross-functional processes — which iswhy it typically starts at a lower scheduling frequency, giving all teams timeto adapt to the new way of working

Asyour scheduling process matures, the ability to make fast, confident decisionsin critical moments becomes increasingly valuable. But does reactive schedulingrequire MES? Not necessarily. Many manufacturers already record execution datain ERP at the end of each unit of movement (pallet, coil, box, and so on) or atthe end of each production order operation. If the cycle times you're trackingaren't too long — say, no more than two to three hours per operation —on-demand rescheduling is already quite feasible. You won't have perfect data,but in the vast majority of cases, you'll have more than enough to make gooddecisions (as the three-scenario example above illustrates)

AManufacturing Execution System adds speed, accuracy, and visibility to theentire scheduling process, making high-confidence rescheduling more practical —and more frequent, when that frequency is genuinely useful. It doesn't replacethe traditional daily or weekly proactive scheduling process; it enhances it.The bottom line: MES strengthens reactive scheduling, but it's not aprerequisite for proactive scheduling. The chart below illustrates howscheduling capability evolves as process maturity increases and MES isintroduced.

Where to begin?

Soknowing that neither system requires the other to be in place first — whereshould you start? It’s a question we hear often

Theanswer comes down to your priorities. In an OEE framework: if your productivityproblem is primarily about performance, MES will show you exactly where you’refalling short and why. If your biggest issue is availability — long wait times,excessive setup times, or imbalances — APS is where to focus. If maintenance isa critical concern, MES opens the door to predictive approaches. If demandvariability and scenario simulation are your pain points, APS handles thatwell. If you genuinely don’t know where your factory’s problems are and need athorough diagnosis, start with MES. But if you already know what the problemsare and want to do something about them proactively, start with APS. If you’reexperiencing productivity pain, can’t pinpoint the source, and have no sense of​​the scale of those issues, and need a completepicture — start with MES. But if you already know where the problems are andwant a proactive tool to help you act on them — start with APS

Mostfactories are run by engineers — and engineers tend to want completeinformation before acting. My advice: watch out for that instinct. A regularcheck-up with a clear, actionable diagnosis is worth far more than a perfectpost-mortem.

Want to learn more about APS and MES? Access our COMPLETE and FREE ebooks right now!