How AI Ticket Deflection Works in Workday AMS: A Modeled Example

Note: this is an illustrative model using round numbers, not a specific client result. The only substantiated figure here is 68% ticket deflection, which Assistly® has achieved in production. Everything else — ticket volumes, hours, costs, backlog — is a clearly-hypothetical scenario built to show how the mechanics work. Your actual results will differ by tenant, scope, and ticket mix.

The short answer

AI ticket deflection works by resolving routine, repetitive Workday questions automatically, before a human agent ever touches them. In a typical AMS desk, a large share of tickets are password and access requests, "how do I…" questions, and policy lookups. Those are exactly the tickets automation handles well. In the illustrative model below, applying a 68% deflection rate (Assistly's production figure) to a hypothetical 500-ticket month drops the human-handled queue to 160 tickets, cuts agent effort by roughly the same proportion, and lets a growing backlog start to fall instead of climb.

Again: the numbers below are made up for teaching purposes. Only the 68% is real.

What gets deflected (and what does not)

Deflection is not magic, and it is not "the AI answers everything." It works on tickets that are routine, well-defined, and answerable from policy or self-service. In a Workday support context, that usually means:

Typically deflectable:

• Password and access: resets, locked accounts, "I can't log in," role and security-group questions answerable by pointing the user to the right place.
• "How do I…" questions: how do I submit time off, change my direct deposit, find my payslip, run a report, delegate an inbox task.
• Policy and lookup questions: accrual rules, pay-period dates, benefits-enrollment windows, "what's our remote-work policy."

Still needs a human:

• Configuration and break/fix: a broken integration, a failing payroll feed, an incorrect business-process step.
• Judgment calls and exceptions: anything requiring approval, interpretation, or a decision the policy doesn't cover.
• Net-new build and enhancements: new reports, calculated fields, security redesigns, release-driven changes.

The point of deflection is not to remove people. It is to stop spending senior Workday time on password resets so that time goes to the work that actually needs it.

The modeled example (illustrative — not a real client)

Picture a hypothetical mid-market organization running Workday HCM and Financials. Make up a starting point of 500 Workday tickets per month. None of these figures describe a real engagement; they're round numbers chosen to make the arithmetic easy to follow.

We'll assume — again, illustratively — an average of 30 minutes of human effort per ticket and a fully-loaded blended rate of $80 per hour, so each ticket "costs" about $40 of effort in the model.

Step 1: the baseline (all human)

• 500 tickets per month
• 500 tickets × 0.5 hours = 250 hours of effort
• 250 hours × $80 = $20,000 of effort per month

Step 2: apply deflection and triage

Now layer in the AI-native pieces:

• Assistly® (deflection) resolves routine tickets directly. Applying the 68% production figure: 500 × 0.68 = 340 tickets deflected, leaving 160 for humans.
• AI-powered proactive monitoring catches some issues before they become tickets at all. For the model, say it quietly removes another 40 would-be tickets a month (illustrative).
• Resolve (AI ITSM) triages and routes what's left, trimming average handle time on the remaining tickets — say from 30 to 24 minutes in the model (illustrative).

Step 3: the modeled result

Metric (illustrative)	Baseline (all human)	With AI deflection + monitoring + triage
Tickets created	500	460 (40 prevented by monitoring)
Deflected by Assistly (68%)	0	313
Human-handled tickets	500	147
Minutes per ticket	30	24
Human effort (hours/mo)	250	about 59
Modeled effort cost/mo	$20,000	about $4,700

(All figures illustrative. 460 created × 68% deflection = 313 deflected, leaving 147 human-handled; 147 × 24 min = about 59 hours; 59 × $80 = about $4,700. Real engagements vary.)

In this made-up scenario, human-handled volume falls from 500 to about 147, and modeled monthly effort drops from roughly $20,000 to about $4,700. The shape of the change — not the exact dollars — is the point.

What this does to backlog over a few months

Backlog is where the difference compounds, so here's a second illustrative table. Assume the team can clear about 180 tickets per month with its existing capacity, and that 500 new tickets arrive each month.

Month (illustrative)	Without deflection	With deflection (147 human/mo)
Start	200 backlog	200 backlog
Month 1	520 (200 + 500 − 180)	167 (200 + 147 − 180)
Month 2	840	134
Month 3	1,160	101

Without deflection, 500 arriving against 180 cleared means the queue grows by 320 a month, the familiar AMS death spiral. With deflection, only about 147 reach humans, so the same capacity burns the backlog down instead. This is illustrative arithmetic, but it captures the real dynamic: deflection changes the sign of the trend, not just the size of the queue.

Why the incentive matters here

This is also where AMS economics quietly diverge. In a staffing-led, hours-billed model, the baseline column is the good column. More tickets and more hours mean more revenue, so there's little structural reason to drive volume down. In an AI-first / outcome-based model, the right-hand column is the goal: the provider is paid for the result, so deflection is in everyone's interest. Same service category, opposite incentives. We unpack that structural difference in our guide to Workday AMS billing models and incentives, and it's a recurring theme across how to think about staffing vs. AI-first AMS.

Frequently asked questions

Is 68% deflection guaranteed? No. 68% is a production figure for Assistly®, not a promise for every tenant. Actual deflection depends on your ticket mix, how much is genuinely routine, the quality of your policy content, and the scope of what the AI is allowed to handle. A tenant heavy on complex configuration work will see a lower routine share, and therefore lower deflection, than one dominated by self-service questions. Treat 68% as a real-world data point, not a contractual number.

Are the numbers in this post real? Only the 68% deflection rate. The 500-ticket volume, the 30-minute handle time, the $80 rate, the backlog figures — all of it is illustrative, chosen to make the model easy to follow. This is not a client case study. (A true, anonymized client case study is in the works for when real numbers are available to share.)

What kinds of tickets actually get deflected? Routine, well-defined ones: password and access requests, "how do I do X in Workday" questions, and policy or lookup questions. Configuration changes, break/fix on integrations or payroll, exceptions, and net-new build still go to a human. Deflection moves the easy volume off the queue so people can focus on the hard work.

Does deflection mean fewer staff? Not necessarily. It means the same staff spend their time differently. Instead of clearing password resets, senior Workday talent works enhancements, release readiness, and the genuinely tricky tickets. The aim is lower effort and a shrinking backlog at a steady or better service level, not a headcount cut.

Where to go next

• See how the underlying economics line up in Workday AMS billing models and incentives.
• Understand the staffing-vs-deflection question in Staffing Firms Are Entering Workday AMS — What Buyers Should Know.
• Compare the field in our best Workday AMS providers for 2026 guide.
• Or see how our AI-native AMS works on the Workday AMS page. (A real, anonymized client case study will replace this modeled example once verified data is available to publish.)

References

1. Assistly® production deflection rate (68%) — AssistNow internal, assistnow.com.
2. AMS staffing vs. managed-services model framing, theplanetgroup.com.
3. Workday — "Find a Partner," partners.workday.com (partner directory and designations).