AppPoints Strategy & the 13-Month Roadmap: How to Deploy MAS 9 Without Failing

Who this is for: IT leaders, Maximo administrators, project managers, and financial stakeholders who need to understand how AppPoints licensing works, how to allocate points by role without overspending, and how to phase a realistic MAS 9 suite deployment that does not collapse under its own ambition.

Estimated read time: 10 minutes

The Two Questions Nobody Answers Honestly

Every MAS 9 implementation starts with two questions that IBM sales materials gloss over and most consultants answer with hand-waving:

  1. How much is this going to cost per user, and how do we control it?
  2. How long does it actually take to deploy the full suite — not the demo, the real thing?

The answers are not complicated, but they are uncomfortable. AppPoints licensing is flexible and powerful, but only if you understand the math. The implementation timeline is at least 13 months post-Manage stabilization, and that is if your data quality is already good. If it is not — and for most organizations it is not — add months to Phase 0 before you even start.

We have seen organizations buy thousands of AppPoints, allocate them on day one across every application, and wonder six months later why Health scores are meaningless and Predict models will not train. The problem was never the technology. The problem was sequencing and data quality.

This post gives you the honest playbook.

AppPoints: One Pool, Infinite Flexibility

How AppPoints Actually Work

AppPoints are the licensing currency for the entire MAS suite. Instead of purchasing individual application licenses — one for Health, one for Monitor, one for Predict — you purchase a single pool of AppPoints that can be allocated across any combination of applications.

Here is the mental model: think of AppPoints like a corporate credit card with a spending limit. The limit is shared, and you choose what to buy with it. If you need more Health users and fewer Monitor users next quarter, you reallocate. No new purchase required.

Key concepts:

Concept — What It Means

AppPoint Pool — Total number of AppPoints your organization purchased — this is your budget

Application Cost — Each application consumes a specific number of points per authorized user

Authorized Usage — Most applications license by total users with access (not concurrent)

Concurrent Usage — Some applications license by users active at the same time — cheaper if utilization is low

Flex Allocation — Points can be reallocated between applications within your contract terms

The flex allocation is the part most organizations under-utilize. You are not locked in. If you allocated 500 points to Manage on day one but now want to deploy Health, you can reallocate points from users who only need Limited access to fund Health access for your reliability engineers.

What Each Application Costs

Exact AppPoint costs vary by contract and change over time. The following represents typical relative costs as of MAS 9. Confirm exact numbers with your IBM representative — but these ratios are consistent across most contracts.

Application — AppPoints Per User (Typical) — License Type — Notes

Manage - Limited — 5 — Authorized — View-only, limited transactions

Manage - Base — 10 — Authorized — Standard EAM functionality

Manage - Premium — 15 — Authorized — Full functionality including industry add-ons

Monitor — 5 — Authorized — IoT monitoring and dashboards

Health — 5 — Authorized — Health scoring and AIO

Predict — 10 — Authorized — ML failure predictions

Visual Inspection — 10 — Authorized — Computer vision inspection

Optimizer — 10 — Authorized — Scheduling optimization

AI Assist — Varies — Authorized — Often included with Premium in some contracts

Civil Infrastructure — 10 — Authorized — Infrastructure management

Mobile — Included — N/A — Included with Manage license

Notice the spread. Manage Limited at 5 points is one-third the cost of Manage Premium at 15. Health and Monitor are relatively affordable at 5 points each. The AI-heavy applications — Predict, Visual Inspection, Optimizer — run 10 points per user. These differences matter enormously when you are allocating across hundreds of users.

Base vs. Premium: Know the Difference

Category — Base AppPoints — Premium AppPoints

Manage Access — Standard EAM — Full EAM plus industry add-ons (Spatial, Transportation, Aviation, etc.)

AI Features — Basic — Full AI Assist, advanced recommendations

Integration — Standard APIs — Advanced connectors

Cost — Lower per point — Higher per point

The critical question: does the user actually need Premium features? If they are a planner who never touches Spatial or Aviation modules, Base saves you 5 points per user. Across 200 planners, that is 1,000 AppPoints — enough to fund Health access for your entire reliability team.

Three Allocation Strategies (Pick One)

Strategy 1: Start Narrow, Expand Later

Allocate most points to Manage initially. As you deploy Health, Monitor, and Predict in subsequent phases, reallocate points from over-provisioned users. This is the safest strategy for phased implementations.

Best for: Organizations that are still migrating to Manage and want to keep licensing costs contained during stabilization.

Strategy 2: Power User Model

A small group of power users gets access to all applications (high points per user). The larger user base gets Manage-only access (lower points per user). This is effective for organizations exploring the suite with a dedicated evaluation team.

Best for: Organizations with a dedicated reliability or innovation team that wants to pilot suite applications while the broader user base stays on Manage.

Strategy 3: Role-Based Allocation

This is the strategy we recommend for most organizations at scale. You map application access to job roles and calculate the exact AppPoints per role.

Role — Applications — AppPoints Per User

Maintenance Manager — Manage Premium + Health + Predict — 30

Reliability Engineer — Manage Base + Health + Predict + Monitor — 30

Planner/Scheduler — Manage Base + Optimizer — 20

Technician — Manage Base + Mobile — 10

Inspector (Visual) — Manage Limited + Visual Inspection — 15

IoT Analyst — Manage Limited + Monitor — 10

Executive — Manage Limited + Health (dashboards) — 10

Look at the spread. A Technician at 10 points costs one-third of a Maintenance Manager at 30 points. If your organization has 500 technicians and 20 maintenance managers, the allocation math is dramatically different than treating everyone the same.

The role-based model forces the right conversation: What does each person actually need? Not what would be nice to have — what do they use daily?

Six Ways to Optimize AppPoints Spending

  1. Audit actual usage. Not every licensed user actively uses every application. MAS provides usage reports. If 40% of your Health-licensed users have never opened Health, reallocate those points.
  2. Use Limited licenses where possible. Many users only need view access to work orders or dashboards. Limited at 5 points is half the cost of Base at 10. Across hundreds of users, this adds up fast.
  3. Phase your application rollout. Do not allocate points to applications you have not deployed yet. If Predict goes live in Month 7, those points are wasted from Month 1 through Month 6.
  4. Monitor utilization continuously. MAS provides usage reports to identify over-allocation. Review quarterly, minimum.
  5. Consider concurrent licensing. If only 20% of your authorized users are active at any given time, concurrent licensing may be significantly cheaper than authorized licensing. Run the math.
  6. Right-size Premium vs. Base. Only use Premium where the features justify the cost. If a user never touches Spatial, Transportation, or Aviation modules, Base is the correct choice.

The Implementation Prioritization Matrix

Before we get to the roadmap timeline, you need to understand why the applications deploy in this specific order. It is not arbitrary. Each phase has technical dependencies on the previous one.

Application — Business Value — Complexity — Prerequisites — Phase — Priority

Health — HIGH — Immediate asset condition visibility — MEDIUM — Requires good data quality — Manage deployed, quality asset data — Phase 1 — 1

Monitor — HIGH — Real-time visibility, enables Predict and Health IoT — HIGH — Requires IoT infrastructure and connectivity — Manage deployed, IoT sensors available — Phase 2 — 2

Predict — VERY HIGH — Prevents failures, reduces unplanned downtime — HIGH — Requires data science, sufficient failure history — Manage + Health + Monitor, failure data — Phase 3 — 3

Visual Inspection — HIGH — Automates inspections, improves consistency — MEDIUM — Requires GPU and camera infrastructure — GPU nodes, camera access, labeled images — Phase 3 — 4

AI Assist — MEDIUM-HIGH — Improves productivity, reduces training burden — MEDIUM — Requires watsonx.ai and training data — watsonx.ai access, 2+ years WO data — Phase 4 — 5

Optimizer — MEDIUM-HIGH — Reduces travel, improves productivity — MEDIUM — Requires clean craft/skill data — Manage Scheduler, accurate labor data — Phase 4 — 6

Civil Infrastructure — HIGH (if applicable) — Regulatory compliance — MEDIUM — Domain-specific configuration — Manage deployed, infrastructure asset data — Phase 5 — 7

Parts Identifier — LOW-MEDIUM — Niche use case — LOW — Simple deployment — MAS deployed — Phase 5 — 8

The decision framework for each application boils down to five questions:

  1. Do we have the prerequisite data? If Health requires quality data and ours is poor, fix data first.
  2. Do we have the infrastructure? If Monitor requires IoT sensors and we have none, plan procurement first.
  3. Do we have the skills? If Predict requires data science and we have no data scientists, hire or train first.
  4. Does the business case justify the investment? Run an ROI analysis for each application.
  5. Are our users ready? Each application requires change management and training.

If you cannot answer "yes" to all five, you are not ready for that phase. Move it back and focus on prerequisites.

Phase 0: Foundation — Where Most Projects Fail

Timeline: Concurrent with Manage migration
Objective: Establish the foundation that every add-on application depends on

This is the phase that nobody wants to do and everybody needs. Phase 0 is not exciting. There are no dashboards, no AI models, no demos for leadership. It is data cleanup, infrastructure planning, and skills assessment. It is the phase that determines whether everything after it succeeds or fails.

Activity — Description — Owner — Duration

Data quality assessment — Audit asset data, work order history, failure codes, meter readings — Data team — 2-4 weeks

Data remediation — Fix install dates, standardize failure codes, clean meter data — Data team + SMEs — 4-8 weeks

Infrastructure planning — Assess GPU needs (for MVI), IoT connectivity, storage requirements — Infrastructure team — 2-4 weeks

Skills assessment — Identify skill gaps for Health, Monitor, Predict, Visual Inspection — HR + Technical leads — 1-2 weeks

Training plan — Develop training plan for identified gaps — Training team — 1-2 weeks

Pilot asset selection — Identify pilot asset classes for each application — Reliability team — 1-2 weeks

Why Data Quality Is Non-Negotiable

We cannot overstate this. Every AI application in the MAS suite depends on the data in Manage. Health reads asset install dates, work order history, and meter readings. Predict reads failure history and sensor data. Visual Inspection needs labeled images. AI Assist needs years of work order descriptions.

If that data is garbage, every application built on top of it produces garbage.

Specific data quality thresholds:

  • Health needs 90%+ install date population on asset records. If only 40% of your assets have install dates, age-based health scoring is meaningless for 60% of your fleet.
  • Predict needs a minimum of 20 failures per failure class to train a statistically meaningful model. If your failure codes are inconsistent — the same failure recorded as "BEARING FAIL," "BRG FAILURE," "BEARING - FAILED," and "MECHANICAL" — Predict cannot aggregate enough training data.
  • Monitor needs consistent asset classification so you can map device types to asset classes. If your compressors are classified five different ways, you cannot create a unified device type.

The uncomfortable truth: Most organizations that compress the MAS implementation to 6 months do so by skipping Phase 0. They deploy Health, get meaningless scores, deploy Predict, get models that will not train, and blame the technology. The technology works. The data did not.

If your data quality is poor, fix it first. Do not deploy applications and hope they will surface the data problems for you. They will — by producing results nobody trusts, which poisons user adoption for years.

Phase 1: Health — Months 1 Through 3

Timeline: Months 1-3 after Manage goes live and stabilizes
Objective: Deploy Health, configure scoring, validate with pilot assets

Month 1: Deploy Health, select pilot asset classes, verify data quality for those assets, set up integration with Manage.

Month 2: Configure scoring contributors (age, meters, work history, inspections), define weights for each contributor, generate initial health scores for the pilot fleet.

Month 3: Validate scores against known conditions with subject matter experts, create degradation curves, run an AIO (Asset Investment Optimization) scenario with realistic budget data, present results to maintenance leadership.

Key deliverables:

  • Health deployed and integrated with Manage
  • Health scores generated for at least one pilot asset class
  • Degradation curves showing meaningful trends
  • AIO scenario completed with realistic budget data
  • Business case validated with maintenance leadership

Success criteria:

  • Health scores align with known asset conditions (validated by SMEs who know which assets are in bad shape)
  • Degradation curves show meaningful trends — not flat lines or random noise
  • AIO recommendations are actionable — not "replace everything" or "do nothing"

Do not advance to Phase 2 until Health scores are credible and leadership trusts the data.

Phase 2: Monitor + IoT Connectivity — Months 4 Through 6

Timeline: Months 4-6
Objective: Deploy Monitor, connect first devices, establish real-time IoT data flow

Month 4: Deploy Monitor, define device types for pilot asset classes, register pilot devices (sensors, PLCs, edge devices), establish MQTT or REST connectivity to the IoT data layer.

Month 5: Ingest data from pilot sensors, create summary and entity dashboards, validate metric accuracy, build custom Python functions for analytics specific to your operational context.

Month 6: Configure anomaly detection algorithms, define alert rules, establish the Monitor-to-Health pipeline (sensor anomaly counts feed Health scoring contributors), establish the Monitor-to-Manage pipeline (alerts generate work orders automatically).

Key deliverables:

  • Monitor deployed with pilot device types
  • Real-time data flowing from pilot sensors
  • Summary and entity dashboards operational
  • Anomaly detection configured and validated
  • Alert-to-work-order flow operational
  • Monitor data feeding Health scoring contributors

Success criteria:

  • Data ingestion is reliable — less than 1% data loss from sensors to dashboards
  • Dashboards provide meaningful operational visibility (not just raw numbers)
  • Anomaly detection has an acceptable false positive rate — less than 20%
  • At least one alert has successfully generated a Manage work order automatically

Risk mitigation: If IoT connectivity proves challenging (and it often does — industrial environments are not data centers), start with CSV file upload to validate the Monitor configuration. Then migrate to live MQTT connectivity incrementally. Do not let connectivity blockers stall the entire phase.

Phase 3: Predict + Visual Inspection — Months 7 Through 9

Timeline: Months 7-9
Objective: Train first prediction models, deploy first visual inspection model

This is where the suite becomes genuinely transformative — and where the data quality investments from Phase 0 either pay off or expose every shortcut you took.

Month 7: Extract training data from Manage (failure history, meter readings, work order patterns) for Predict. Perform feature engineering. Simultaneously, collect inspection images for Visual Inspection, label them, and set up GPU infrastructure.

Month 8: Train Predict models using Watson Studio or watsonx.ai. Evaluate accuracy against historical holdout data. Iterate on features and hyperparameters. In parallel, train Visual Inspection models and validate classification or detection accuracy.

Month 9: Deploy Predict models so failure predictions are visible in Manage. Deploy Visual Inspection models to the API and test the mobile inspection app. Link MVI results to Manage work orders. Validate both systems against known outcomes.

Key deliverables:

  • Predict model deployed for at least one asset class with predictions visible in Manage
  • Visual Inspection model trained and deployed for at least one inspection use case
  • Both models validated against known historical outcomes
  • At least one preventive action taken based on a Predict prediction

Success criteria:

  • Predict model accuracy exceeds 70% (measured against historical holdout data)
  • MVI model accuracy exceeds 85% for classification or 80% for object detection
  • Maintenance teams find predictions actionable — not just technically accurate
  • At least one preventive action taken based on a prediction

Risk mitigation for Predict: If you do not have enough failure history (fewer than 20 failures per class), supplement with industry-provided models or focus on asset classes with the best data. Do not try to force a model on sparse data.

Risk mitigation for MVI: If you do not have enough training images, use data augmentation techniques. Start with classification (which needs fewer images) rather than object detection. You can graduate to detection after the initial model proves value.

Phase 4: AI Assist + Optimizer — Months 10 Through 12

Timeline: Months 10-12
Objective: Deploy AI-powered assistance and scheduling optimization

Month 10: Deploy the AI Service and configure watsonx.ai for AI Assist. Deploy Optimizer and configure the optimization model. Clean labor data — craft assignments, skill matrices, and location coordinates must be accurate.

Month 11: Train AI Assist models on your organization's work order data. Test field value recommendations (failure codes, job plans, priority suggestions). Run Optimizer schedules against real work order backlogs. Tune Optimizer parameters and test the Dispatching Dashboard.

Month 12: Conduct user testing for both applications. Collect feedback on AI recommendation quality and optimized schedule practicality. Compare Optimizer schedules against manual scheduling outcomes.

Key deliverables:

  • AI Assist deployed with field value recommendations operational
  • Natural language query capability available for technicians
  • Optimizer producing feasible schedules that reduce travel
  • Dispatching Dashboard operational for supervisors
  • User feedback collected on both systems

Success criteria:

  • AI field recommendations accepted by users more than 60% of the time
  • Optimizer reduces travel time by at least 15% compared to manual scheduling
  • Users report a positive experience — not just tolerance, but genuine value

Risk mitigation for AI Assist: User resistance to AI recommendations is common. Emphasize that AI Assist is an assistant, not a replacement. Involve field technicians in the evaluation process so they feel ownership rather than imposition.

Risk mitigation for Optimizer: If optimized schedules are not practical, tune constraints with input from field supervisors. Keep manual override capability. Optimization that ignores field reality will be rejected no matter how mathematically optimal it is.

Phase 5: Civil Infrastructure + Full Integration — Month 13 Onward

Timeline: Month 13 and beyond
Objective: Deploy remaining applications, achieve full suite integration, enterprise-wide rollout

This is where the MAS suite vision becomes reality. The full data pipeline flows: sensors feed Monitor, Monitor feeds Health, Health feeds Predict, Predict drives Manage work orders, Manage schedules through Optimizer, Visual Inspection augments inspections at every stage, and AI Assist empowers every user along the way.

Key activities:

  • Deploy Civil Infrastructure (if applicable to your organization — DOT agencies, utilities, municipalities)
  • Deploy Parts Identifier (evaluate for field maintenance scenarios)
  • Enterprise-wide rollout beyond pilot asset classes to the full fleet
  • Train advanced custom models using watsonx.ai on organization-specific data
  • Build cross-application dashboards and executive reporting
  • Establish continuous model retraining and improvement cycles

Key deliverables:

  • All applicable MAS applications deployed and operational
  • Full data pipeline: Monitor to Health to Predict to Manage to Optimizer
  • AI models trained on organization-specific data (not just IBM defaults)
  • Enterprise-wide rollout plan with change management and training
  • Continuous improvement cycle established for all AI models

The Team You Need

You cannot explore all applications simultaneously. Follow the phased roadmap and focus on one to two applications at a time.

Role — Responsibility — Applications

Reliability Engineer (Lead) — Champion Health and Predict evaluation — Health, Predict, Monitor

IoT/OT Engineer — Lead Monitor pilot, device connectivity — Monitor, Edge Data Collector

Data Scientist — Lead Predict and AI model development — Predict, AI Assist, Visual Inspection

Inspection Specialist — Lead Visual Inspection pilot — Visual Inspection, Civil Infrastructure

Scheduler/Planner — Lead Optimizer evaluation — Optimizer, Manage Scheduler

Manage Administrator — Support integration, data quality, configuration — All applications (supporting role)

Estimated effort per application:

Application — Team Size — Estimated Hours

Health — 2-3 people — 42-84 hours (1-2 weeks)

Monitor — 2-4 people — 56-104 hours (2-3 weeks)

Predict — 2-3 people — 60-112 hours (2-3 weeks)

Visual Inspection — 2-3 people — 55-138 hours (2-4 weeks)

AI Assist — 2-3 people — 68-128 hours (2-3 weeks)

Optimizer — 2-3 people — 60-120 hours (2-3 weeks)

Civil Infrastructure — 2-3 people — 60-112 hours (2-3 weeks)

Parts Identifier — 1-2 people — 22-44 hours (1 week)

Risk Mitigation by Phase

Every phase has a primary risk and a mitigation strategy. Knowing these in advance is the difference between a timeline slip and a project failure.

Phase — Primary Risk — Mitigation

Phase 0 — Data quality worse than expected — Start data remediation early, set realistic expectations with leadership

Phase 1 — Health scores do not match reality — Iterate on contributors and weights with SME input — do not ship scores nobody trusts

Phase 2 — IoT connectivity challenges — Start with CSV upload, migrate to MQTT incrementally

Phase 3 — Insufficient failure history for Predict — Supplement with industry models, focus on assets with the best data

Phase 3 — Not enough training images for MVI — Use data augmentation, start with classification (needs fewer images)

Phase 4 — User resistance to AI recommendations — Emphasize AI as assistant not replacement, involve users in evaluation

Phase 4 — Optimizer schedules not practical — Tune constraints with field input, keep manual override capability

Phase 5 — Integration complexity across all apps — Maintain strong integration testing, monitor data pipeline health continuously

The Honest Timeline

Here is the complete roadmap laid out on a timeline:

Phase 0     Phase 1      Phase 2      Phase 3       Phase 4      Phase 5
Foundation  Health       Monitor      Predict+MVI   AI+Optimizer Civil+Full
            Mo 1-3       Mo 4-6       Mo 7-9        Mo 10-12     Mo 13+
   |-----------|-----------|-----------|------------|-----------|-------->
   Data        Scoring     IoT         ML Models     AI Assist   Enterprise
   Quality     AIO         Dashboards  MVI Models    Optimizer   Rollout
   Infra       Degradation Anomaly     Predictions   Scheduling  Custom AI
   Skills      Validation  Alerts      Validation    Feedback    Retraining

Phase 0 runs in parallel with your Manage migration. Phases 1 through 5 are sequential, each building on the previous phase's outputs. The full suite takes 13+ months after Manage stabilization.

Is 13 months slow? No. It is realistic. We have seen organizations attempt 6-month implementations by skipping Phase 0 and compressing Phases 1 through 3. The result is always the same: Health scores nobody trusts, Predict models that will not converge, and a project team that has lost credibility with leadership.

The organizations that succeed are the ones that respect the sequence, enforce the success criteria at each phase gate, and refuse to advance until the current phase is genuinely complete.

Key Takeaways

  • AppPoints are a shared pool, not individual licenses. You can reallocate between applications as your deployment matures. Use this flexibility deliberately, not accidentally.
  • Role-based allocation is the right model for most organizations. A Technician at 10 points and a Maintenance Manager at 30 points is a 3x difference that matters across hundreds of users.
  • The 13-month roadmap is realistic, not conservative. Every phase has technical dependencies on the previous one. Compressing the timeline means skipping prerequisites.
  • Phase 0 is where most projects fail. Data quality, infrastructure planning, and skills assessment are unglamorous but non-negotiable. Health needs 90%+ install dates. Predict needs 20+ failures per class.
  • Success criteria per phase prevent premature advancement. Do not move to Phase 2 until Phase 1 Health scores are credible. Do not train Predict models until you have sufficient failure history.

References

Series Navigation:

Previous: Part 14 — AI Assist & Optimizer
Next: Part 16 — Licensing: Free vs Paid in MAS 9

View the full MAS FEATURES series index

Part 15 of the "MAS FEATURES" series | Published by TheMaximoGuys

AppPoints licensing is one of the most misunderstood aspects of MAS 9. Organizations that treat it as a cost center miss the strategic flexibility it offers. And the 13-month roadmap is not a suggestion — it is the minimum viable timeline for an implementation that produces results your maintenance team actually trusts. Phase 0 is not optional. Fix the data first.