Building a Start Center That Actually Drives Work: A 2026 Practitioner Guide to KPI Configuration, Result Sets, and Role-Based Tiles

Most Maximo deployments have a Start Center problem. It loads with a default tile set, three out-of-date KPIs, a query that returns 47,000 work orders, and a "My Work" tile that nobody trusts. Users ignore it. Supervisors build their own ad-hoc queries in the Work Order Tracking application. The control desk becomes a Slack channel.

The Start Center is supposed to be the cockpit. Done well, it is the single highest-leverage UI surface in Maximo — the one place where a planner, supervisor, or reliability engineer sees exactly what they need to act on, in the order they need to act on it. Done poorly, it is wallpaper. This article walks through the configuration patterns that make the difference, with the 2026 twist of role-based tile scoping, modern chart types, and automation-script-driven personalization.

The five Start Center questions to answer first

Before you touch a single tile, sit down with the people who will use the Start Center and answer five questions:

1. What decisions does this person make in a day? A control desk supervisor decides "what gets worked on next." A planner decides "what work is ready to schedule." A reliability engineer decides "what assets are at risk." Each role needs a different Start Center.
2. What data answers that decision? A control desk supervisor needs an open-wo queue, a PM-due list, a crew-status board. A planner needs work-order readiness, materials availability, crew availability, and a priority view. A reliability engineer needs health scores, anomaly counts, and open-condition alerts.
3. How stale is acceptable? Real-time (seconds-old meter data) for a control desk. Hour-old for a planner. Day-old for a reliability dashboard. The Start Center is not the place for stale data pretending to be real-time.
4. What action do they take from here? If the only action is "click through to the WO record," the Start Center is just a launcher. If the action is "drag to reassign," "click to approve," or "right-click to send to a crew," the Start Center is a workflow surface.
5. What KPI tells them the system is working? A control desk supervisor should see "WO backlog age > 14 days = 12." A planner should see "% ready to schedule = 64." A reliability engineer should see "assets below health threshold = 23." A Start Center without an explicit KPI is decoration.

Write the answers down. They become the requirements doc for the tile set.

The architecture of a Start Center

A Start Center is composed of three layers:

1. Portlets (tiles) — the visual components. The 2026 default tile set includes: KPI, Result Set, Chart, Quick Insert, Bulletin, Map, Calendar, and custom (script-driven) tiles.
2. Result Sets — saved queries that feed the data tiles. A Result Set is bound to a Maximo object (WORKORDER, ASSET, SR, PO, etc.) and a WHERE clause. It is the only correct way to build a data tile.
3. KPI configuration — the calculation that drives a KPI tile or the inline KPI in a chart tile. A KPI is bound to a Result Set, an aggregation function (count, sum, average, min, max), a column, and a threshold map for color coding.

The mistake most admins make is treating all three as one. They are not. The Result Set is the most important of the three, and the one most often poorly built. A result set with a wide WHERE clause and no index hints is a performance problem waiting to happen, and a result set that returns 50,000 rows is useless as a tile data source.

Building a result set the right way

A result set is a saved query in the Result Set application. It binds an object structure (the FROM clause), a WHERE clause (the filter), a column list (the SELECT), and an optional order and limit. Here is a practical example — the "Open Work Orders, High Priority, My Crew" result set:

A few things to notice:

• Bind variables for personalization. :mycrew is a variable that the user's default insert site, owner group, or a script-driven personalization resolves at render time. Without it, you cannot build a "My Work" tile that respects crew structure.
• The Max Rows matters. A result set rendered to a chart or table tile will try to render all rows. 200 is a practical ceiling for browser performance; 500 is the absolute max.
• The WHERE clause needs to be index-friendly. status, priority, assignedownergroup, and istask are all indexed columns on a well-tuned WORKORDER table. A WHERE clause that uses functions (UPPER(status) = 'APPR') or computed columns kills the index and forces a full table scan.
• Filter on istask = 0 if you only want top-level work orders. If you want both WOs and tasks, you can do it, but build a separate result set.

Building a KPI that does something

A KPI tile is a number with a color. The number is calculated from a result set using an aggregation. The color is determined by a threshold map.

Practical example: "Critical WO backlog (priority 1, open > 7 days)":

The threshold map is the part that drives behavior. A KPI without thresholds is just a number on a screen. With thresholds, it is a behavior driver. The threshold values should be set in collaboration with the operations team — not by the admin alone — and they should be reviewed quarterly.

A few KPI patterns that work:

• Tiered counts. "WO backlog by priority" rendered as four KPI tiles, color-coded, with the red tile getting the largest font.
• Aging KPIs. "WO > 14 days open" using a date-difference calculation in a saved query. These are the ones that catch management attention.
• Ratio KPIs. "% PMs completed on time" as a single KPI tile, calculated from a result set with an aggregation on a comparison. These are the ones that drive process improvement.
• Cost KPIs. "Total WO cost MTD" using a result set that joins WORKORDER to WPLABOR and WPMATERIAL with a sum. These are the ones that finance cares about.

The role-based tile scoping pattern (2026)

The big change in 2026 is that Start Centers are role-scoped by default. The mechanism is the Start Center Template configuration in the Security Groups application.

The pattern is:

1. Build a base Start Center that has the tiles you want everyone to have (a "My Open WOs" tile, a "Bulletin" tile, a "KPI" tile).
2. Build role-specific Start Centers (e.g., STARTCTR_PLANNER, STARTCTR_SUPERVISOR, STARTCTR_RELIABILITY) with the additional tiles each role needs.
3. In Security Groups, assign the role-specific Start Center to the security group; users inherit the role-specific Start Center when they log in. The base Start Center is the fallback.
4. Use personalization (the user can add/remove tiles) on top of the role-scoped Start Center. The personalization is stored per-user, on top of the template.

This pattern avoids the classic mistake of trying to build one Start Center that works for everyone, which is the path to the default tile set nobody trusts.

The chart tile: what to use and when

The chart tile in 2026 supports a real chart library. The patterns that work:

• Bar chart for "WO count by status" or "PMs completed by week." This is the workhorse.
• Stacked bar for "WO count by priority and status." Good for showing backlog composition.
• Line chart for "WO backlog over time." Use this for trend KPIs, not point-in-time ones.
• Pie chart for "WO distribution by failure class." Use sparingly — humans read pies slowly and badly.
• Heat map (introduced in 2024, matured in 2026) for "WO density by location and priority." Excellent for control desks and reliability dashboards.

The chart tile is bound to a result set, and the chart's axes are bound to columns in the result set. The result set must return aggregated data, not raw rows, for the chart to render sensibly. A result set that returns 5,000 raw rows and a chart bound to count(wo) is a chart that takes 8 seconds to render and uses all the browser's memory.

Automation-script-driven personalization

The most powerful 2026 pattern is to drive Start Center tiles from an automation script. The use cases:

• Dynamic threshold maps. A script reads the user's site, role, and a system property to compute the KPI thresholds, instead of hard-coding them. The same KPI tile reads different thresholds for different users.
• Bind variable resolution. A script resolves :mycrew to the user's current crew at render time, so the "My Work" tile is always the right crew, even when a user is assigned to multiple.
• Conditional tile visibility. A script hides the "Cost KPIs" tile for users in a read-only role, or the "Approve WO" tile for users who have not been delegated approval rights this week.
• Cross-object tile. A script-driven custom tile that joins data from multiple objects and renders a single composite view (e.g., "My assets and their current health score"), which is not possible with the standard result set alone.

The pattern is a Jython script with a launch point on startcenter.render. The script runs at tile render time, has access to the user's session, the security group, the bind variables, and the result set API. The script modifies the tile configuration in memory before it renders. It does not persist back to the database unless you want it to (for example, to remember a user's last-selected site).

The minimum viable Start Center

If you are starting from scratch and need to ship something in a week, the minimum viable Start Center is:

1. One role-specific Start Center for the largest user group.
2. Three KPI tiles: WO backlog aging, PM compliance, and critical WO count.
3. One result set tile: "My open work orders" with bind variables and a 200-row limit.
4. One chart tile: WO count by status, bar chart.
5. One bulletin tile: company-wide or role-specific announcements.
6. Threshold maps on every KPI in collaboration with the operations team.
7. A review cadence: every 90 days, the Start Center is reviewed with the user group, and the result sets and KPIs are tuned.

That is the floor. The ceiling is a Start Center that the operations team actually opens first thing in the morning because it is the most useful screen in Maximo. The difference between the two is mostly the result sets and the threshold maps.