One agent suddenly had 500 leads. Conversions dropped silently. 💀
Our lead assignment flow was simple:
Agent 1 → Agent 2 → Agent 3 → repeat
Worked perfectly for months.
Then production reality kicked in.
Agents go on break. Agents change shifts. Agents leave the team. Agents go offline mid-day.
Round robin happily kept skipping them. Remaining agents kept absorbing leads. Some started drowning. 🌊
Round robin assumes every agent is available all the time. Production never works that way.
How we fixed it
1. Hard cap per agent
Every agent gets a max active lead count — say, 50. Before assigning, check the cap. At cap? Skip or queue. Simple, but it stops one agent from absorbing everything when others go offline.
2. Smart scoring
Instead of position in a rotation, each agent gets a score:
score = current_load + activity_weight + availability_flagLowest score wins the lead. An agent on break scores high (unavailable), an agent with 20/50 leads scores lower than one at 45/50. The system routes to whoever can actually handle it.
If everyone is at cap, the lead goes to a queue — not to the most overwhelmed person.
Round robin is a great algorithm for stateless, always-available resources. Agents are neither.
Perfect algorithms fail when reality doesn’t match the assumption. Production humbled us. 😅