You just hired a support agent. Your last one took 8 weeks to get fully productive. During those 8 weeks, the rest of the team carried extra load, quality dipped on the new hire's tickets, and by week 6, everyone was frustrated.

Eight weeks is what happens when onboarding is unstructured. "Shadow Sarah for a while and then start picking up tickets." That's not a plan. That's a hope.

Two weeks is aggressive but achievable if you've done two things: documented your most common scenarios and built a clear progression from observation to independence.

Week One: Observe and Practice

Days 1-2: product and tools.

The new agent needs to know what your product does and how your tools work. Not every feature. Not every edge case. Just the core: what does the customer experience look like? What does the support workflow look like? How do you find customer information? How do you respond?

Give them a customer account to play with. Let them submit fake support tickets to themselves. Let them break things. Two days of hands-on exploration beats two weeks of reading documentation.

Set up their tool access on day 1, not day 3. Nothing kills momentum like sitting around waiting for login credentials.

Days 3-4: shadow sessions.

The new agent shadows an experienced agent for two full days. They sit next to them (or screen-share remotely) and watch them handle real tickets in real time. The experienced agent talks through their decision-making: "This customer is asking about X, so I'm going to check Y first. I can tell they're frustrated because of Z, so I'll start with an acknowledgment."

After each ticket, the new agent asks questions. Why did you phrase it that way? Why did you escalate that one? How did you know to check that system?

This is where the real learning happens. Documentation tells you what to do. Shadowing shows you how to think.

Day 5: guided practice.

The new agent handles real tickets, but every response gets reviewed by a senior agent before it goes out. Start with the simplest ticket categories (password resets, basic how-to questions, status inquiries). The senior agent reviews each draft, gives feedback, and approves or suggests changes.

Most agents handle 10 to 15 tickets on their first day of practice. Some responses take 20 minutes to compose and review. That's fine. Speed comes with repetition.

Week Two: Increasing Independence

Days 6-7: expanded categories with review.

The new agent handles a wider range of ticket types, still with review. Add billing questions, feature requests, simple complaints. The reviewer checks every response but spends less time on the simple ones (which the agent has already mastered) and more on the new categories.

By day 7, the agent should be handling simple tickets independently (no review needed) and medium-complexity tickets with review.

Days 8-9: independent with spot-checks.

The agent handles all ticket types independently. A senior agent reviews a random 30% of their tickets (the agent doesn't know which ones). Feedback is given at the end of each day, not after each ticket.

This is the trust-building phase. The agent develops confidence. The team develops trust in the new hire. If quality is consistently good on the spot-checked tickets, you're almost there.

Day 10: full independence with safety net.

The agent handles their full queue. They know they can escalate anything they're unsure about (and they should, liberally, for the first few weeks). A weekly quality review replaces daily spot-checks.

You now have a productive agent. Not a perfect one. They'll still encounter scenarios they haven't seen. They'll still need to ask questions. But they're handling 80% of tickets independently and correctly.

What to Document Before They Start

The onboarding only works this fast if you've prepared. You need:

A top-20 scenarios guide. Your 20 most common ticket types with the correct response for each. Not a 50-page manual. A quick-reference document that covers 70% of what a new agent will encounter in their first month.

A decision tree for escalation. When should they escalate? To whom? What information do they need to include? If this isn't documented, they'll either escalate everything (wasting senior agents' time) or escalate nothing (sending bad answers to complex questions).

Tool guides. Screenshots of your help desk, billing system, and any other tools they'll use daily. Keep these current. If you reorganized your Zendesk views last month, update the screenshots.

A "who knows what" list. When the agent has a question about billing, who do they ask? Product questions? Technical issues? Legal/compliance? This is especially useful in small teams where roles overlap.

Common Onboarding Mistakes

Throwing them in on day one. "Just start answering tickets, you'll figure it out." They'll figure it out eventually, after sending 50 wrong answers and frustrating 50 customers. The cost of bad early responses (customer churn, reputation damage) far exceeds the cost of a structured first week.

Over-documenting. A 100-page knowledge base is not onboarding. Nobody reads 100 pages. Give them the essentials on day one and let them discover the rest as needed. The top-20 scenarios guide should fit on 5 pages.

No feedback loop. If the new agent sends a bad response and nobody tells them, they'll keep sending bad responses. Daily feedback in week one, regular feedback in week two, weekly feedback after that. Frontload the coaching.

Comparing to experienced agents too early. Your senior agent handles 40 tickets per day. Your new hire handles 12. That's normal. Speed comes with practice. Quality comes first.

How AI Changes Onboarding

If you're using AI classification and auto-responses for simple tickets, your new agent doesn't need to learn those categories at all. AI handles password resets, basic how-tos, and status checks. The new agent starts with medium-complexity tickets and works up.

This means the two-week onboarding is even faster for the tickets that matter. The agent spends less time on the easy stuff (which AI handles) and more time learning the complex scenarios that actually need human judgment.

AI also provides a built-in quality check during onboarding. If the AI classifies a ticket and the new agent's response contradicts the classification, that's a flag. "The AI identified this as a billing issue, but you responded about a feature question. Let's look at this together."