A customer calls a broadband provider because the internet is down. The first agent walks them through restarting the router, checks whether there is a known outage in the area, and confirms the account is active. Still nothing. The ticket goes to a network support engineer who runs line diagnostics and finds a configuration fault on the customer’s port. If that engineer cannot fix it, the issue might go further, to the team that manages the equipment or even the vendor that made it.
That ladder is tiered support, usually described as L1, L2 and L3. The labels are universal; what sits in each tier varies a great deal between companies.
What each level typically owns
L1 (first line) is the front door: phone, chat, email and self-service follow-ups. L1 agents log the issue, identify and verify the customer, gather the facts, and resolve anything that has a known answer: password resets, billing explanations, order status, standard troubleshooting from a script or knowledge base. Their key skills are diagnosis through questioning, clear communication, and knowing when to stop and escalate. In most BPOs, L1 is the bulk of the headcount.
L2 (second line) handles issues that need deeper technical or process knowledge, more system access, or authority that L1 does not have. In technical support this might mean log analysis, account configuration, or remote access to a device. In a bank, L2 might be a specialist team that can approve fee reversals above a threshold or investigate a disputed transaction. L2 usually works from tickets rather than live calls, though some models have L2 take warm transfers.
L3 (third line) is where problems go when the answer is not known. These are engineers or product specialists who can change code or configuration, find root causes, and fix defects. In many outsourcing models L3 stays with the client or the software vendor, because it requires access to the product itself.
Some organisations also speak of L0, meaning self-service such as FAQs, chatbots and IVR, and L4, meaning external vendors or manufacturers. The exact split matters less than everyone in the chain agreeing on it.
Why tiers exist at all
Tiers are an economic design. Specialists are scarce and expensive; most customer issues are routine. Putting everyone in front of a specialist would waste their time on password resets. Putting nobody in front would mean specialists answering the phone all day.
A simple illustration shows how the numbers work. Suppose 10,000 tickets a month, a cost per ticket of ₹150 at L1, ₹600 at L2 and ₹2,000 at L3, and that every escalated ticket has already incurred the cost of the levels below it. If L1 resolves 70%, L2 resolves 25% and 5% reach L3, the total is ₹43 lakh, or ₹430 per ticket on average.
Now suppose better troubleshooting guides let L1 resolve 80%, with L2 taking 15% and L3 still 5%. The total falls to ₹37 lakh, ₹370 per ticket, a saving of ₹6 lakh a month without touching L3. This is the logic behind shift-left: moving resolution to the lowest level that can handle it safely, and ideally into self-service.
The hidden cost is the hand-off, not the tier
The cost figures above leave out what matters most to the customer: every escalation adds waiting time and usually a repeat explanation. A customer escalated to L2 might wait a day for a callback; one escalated to L3 might wait a week. If the L1 ticket notes were thin, the L2 engineer calls the customer to ask the same questions again.
Good escalation practice therefore focuses on the quality of the hand-off:
- Clear escalation criteria. L1 should know exactly when to escalate: which symptoms, which request types, which value thresholds. Vague criteria cause both premature escalation (L2 flooded with things L1 could have solved) and late escalation (customers stuck at L1 while agents try things they should not).
- Minimum ticket content. Customer details verified, symptom described, steps already tried and their results, error messages, and the customer’s preferred contact time. L2 teams often maintain a checklist and reject tickets that lack it, which is unpopular but effective.
- Ownership. Some organisations keep the ticket owned by L1 even when L2 works on it, so the customer has one point of contact. Others transfer ownership. Either works if it is explicit; confusion over who updates the customer is a classic source of complaints.
- Feedback down the ladder. When L2 solves something L1 could have solved with the right article or access, that should become a knowledge base update or training point. Without this loop, the same tickets climb the ladder forever.
Measuring a tiered model
Useful measures include the L1 resolution rate (sometimes called first-level resolution), the escalation rate, the percentage of escalations returned as “could have been solved at L1”, time spent at each tier, and reopen rates after closure. Watch them together. Pushing L1 resolution too hard can make agents hold on to tickets they should escalate, and the reopen rate will rise. Pushing escalation rates down with targets alone, rather than with knowledge and access, simply moves frustration around.
When tiers start to blur
Several trends are softening the strict ladder. Self-service and chatbots absorb many simple L1 contacts, so the L1 work that remains is harder than it used to be. AI-assisted search and suggested answers let L1 handle problems that once needed L2. Some organisations use swarming instead of escalation: rather than passing a ticket upward, they pull the right specialist into the case with the front-line agent, so the customer is not handed off at all. Others create “L1.5” roles of experienced agents with extra access.
Whatever the structure, the questions stay the same: how quickly does each problem reach someone able to solve it, and how much does the customer have to repeat along the way?
Written by the RP Strativa Team. Ticket volumes, costs and resolution rates in the examples are illustrative.
Leave a Reply