The end of the ticket

Shared services was the twentieth century’s smartest answer to back-office cost, and its founding move was an act of packaging: take the work nobody wanted to own, wrap it in a ticket, and route it to wherever an hour of labor cost least. Everything the modern GBS organization measures descends from that move — tickets opened, tickets closed, SLA attainment, cost per ticket. Twenty-five years on, the packaging has outlived its purpose. The ticket was never the work. The ticket was the transport mechanism for work that could not yet travel as software. Now it can.

Three generations, one queue

GBS 1.0 arbitraged the hour: same work, cheaper geography. GBS 2.0 — the platform generation most large enterprises now inhabit — standardized, tooled, and measured it: global process owners, workflow platforms, dashboards of green SLAs. Real progress, and look closely at what survived both generations intact: the queue itself. A request is still born somewhere, packaged, routed, prioritized, worked, closed, and surveyed. The platform generation made the conveyor belt faster and beautifully instrumented. It never asked why the item was on a conveyor belt at all.

Resolution at source

Trace any high-volume ticket class to its origin and you find the same anatomy: a person hit a wall — a blocked invoice, a master-data change, an access request, a report that looks wrong — and the wall was converted into a work item for a different person, days away, with less context. The autonomous pattern dissolves the conversion. The blocked invoice is investigated the moment it blocks — the mismatch diagnosed, the receipt traced, the fix applied within policy or escalated with the evidence attached (The System of Record Learns to Act’s transaction surface, operating). The master-data change is validated and executed under the owner’s envelope (The Foundation Eats the Roadmap’s standing patrol). The requester never files anything, because the wall repaired itself while they watched. Deflection, properly understood, is not self-service portals and chatbots deflecting humans from humans. It is the event resolving at the point of occurrence, so the ticket is never born.

What of the requests that genuinely must be requests — the judgment calls, the novel asks, the policy exceptions no patrol can pre-empt? They inherit a transformed intake. Today’s front door — the portal with its forty-field form, the category tree nobody navigates correctly, the triage queue re-routing the mis-filed — exists to compensate for a receiving system that could not understand a plainly-stated need. An autonomous front door can: the request arrives in ordinary language, is understood against the service catalog, enriched with the requester’s context (entity, role, history, entitlements), resolved on the spot when policy allows — and, when it truly needs a human, lands on the exception desk as a prepared case, not a raw ticket. The form dies with the queue it fed. Intake stops being a filing exercise the requester performs for the center’s convenience and becomes what it should have been: a conversation the center is finally equipped to hold.

Pricing the outcome

The ticket era’s greatest trick was making its unit of account feel natural: cost per ticket, benchmarked annually, arbitraged geographically. The successor unit is the outcome — the invoice matched, the record corrected, the access granted, the wall repaired — and the rate card above prices it honestly: platform, models, and the exception desk’s human judgment, divided by everything delivered, with no distinction between the deflected majority and the judged tail. Two properties make this card strategically different from its predecessor. It is comparable in a way ticket costs never were — an outcome is an outcome, whether the requester sat in Frankfurt or Manila. And it is deflating, riding the model-price curve of Paying for the Machines, which means the center that publishes it is the first shared-services organization in history whose unit price improves between benchmarks without a lift-and-shift.

And the transition has a sequencing gift the ticket era never offered: the center’s own data tells it exactly where to start. Twenty years of ticket history is a census of every wall the enterprise hits — categorized, time-stamped, volumed, costed. Mine it and the deflection roadmap writes itself: in most centers a Pareto pattern any GBS leader can confirm from their own data in an afternoon — the top ticket classes covering the bulk of volume — each traceable to a root event (a blocked invoice, a data change, an access request) that the autonomous pattern can meet at source. The queue’s final service to the enterprise is to serve as the map of its own retirement — a better handover than most operating models ever manage.

Follow-the-sun judgment

One asset of the arbitrage era survives the transition with its value enhanced: the footprint. A global network built to chase labor cost turns out to be exactly the topology an autonomous estate needs for its judgment layer — exception desks in three regions passing a follow-the-sun book, so the alert flagged in Singapore’s morning is judged before New York wakes, and the month-end close (The Continuous Close) is reviewed continuously around the clock rather than heroically in one time zone’s midnight. The skills transition is real — from processing to judging, per The Last Org Chart — but the geography already exists, staffed with people who know the processes end-to-end because they have been running them. GBS leaders sit, mostly without noticing, on the enterprise’s only ready-made global judgment network. That, not the ticket count, is the asset to defend in the next operating-model review.

The SLA inversion

A quieter casualty of the transition deserves its own paragraph: the service-level agreement. The SLA was the platform generation’s masterpiece — and it is a promise about the queue: how long an item will wait, how fast it will be worked, how often the promise is kept. Resolution at source makes the promise category-obsolete for the deflected majority; there is no wait to measure on a ticket that was never born. What replaces it is a promise about the operation: touchless rate by process, exceptions per thousand, time-to-resolution on the judgment tail, and — the one internal customers actually feel — walls repaired per period without anyone filing anything. Service management’s instincts survive; its unit of account changes. Centers that renegotiate their charters around the new metrics early will find budget conversations easier, because for the first time the numbers they report are the numbers the business experiences.

What the center becomes

This is existential for GBS leadership in exactly the way The Last Org Chart predicted, because the center’s political capital is denominated in the currency being retired: headcount and ticket volume. The GBS 3.0 center that thrives makes an early, deliberate identity trade — from processor of work to operator of the autonomous estate: the exception desks, the policy benches, the model stewardship, the control plane, run as a service to the enterprise. Its metrics migrate from cost-per-ticket to exceptions-per-thousand, touchless rate, and cost-per-outcome (Paying for the Machines’s rate card, institutionalized). Its geography advantage survives, transformed: the same global footprint that once arbitraged labor hours now provides follow-the-sun judgment coverage for exception desks that never close. The centers that make the trade become more strategic than they have ever been — the enterprise’s operators of autonomy. The centers that defend the ticket count will be measured by it, all the way down.

The Index below is, for shared-services leaders, effectively a self-portrait: six processes, six honest answers about where the queue still rules. The benchmark tells you how your center compares.