Getting customer support onto one system

They were about to hire

Growth had outrun the support team. Questions were coming in through email, YouTube comments, Instagram, Facebook groups and a private members’ forum, and each one was being handled separately by whoever had it open at the time. The plan was to add headcount. Before they did, the founder asked us to check whether this was really a volume problem or a structure problem.

It was structure. Five channels, monitored independently, meant nobody could see total inbound volume, and things were getting answered twice or not at all. There was no prioritization, so a billing failure and a general question competed purely on who arrived first. There was no escalation path, so anything unusual sat until someone happened to mention it. And there was no response time or resolution data, so nobody could say whether service was getting better or worse.

What we did

The first requirement was one queue. Nothing could be measured or prioritized until all five channels landed in the same place, so tool selection followed from that decision instead of leading it. The second requirement was that prioritization be automatic. Asking a team that’s already underwater to triage by hand adds a decision to every ticket; routing rules and SLAs by category move that decision to the system.

We also kept the tagging deliberately coarse at the start and let the recurring tickets show us where the real splits were. Categories designed in advance have a way of covering cases that never show up.

Audit and tooling. We mapped every channel, its volume, its response expectations and who was covering it. We picked Front as the central platform because it could pull email, social and community channels into one shared inbox, which spared the team five separate tools, and integrated it with Slack and the relevant APIs so a high-priority conversation surfaced when it arrived rather than at the next inbox check.

The system. Routing rules, a tagging taxonomy and SLAs per category. We defined the categories that genuinely needed different handling, like refunds, technical issues and account access, and routed each to the right person on arrival. We wrote the escalation path down, including what qualifies and who receives it.

The team. Training on the new workflows, and dashboards for response time, resolution rate and satisfaction, so the team could see their own numbers instead of hearing about them in a meeting.

Tuning. Weekly reviews with the support leads to flag recurring issues, update macros and adjust SLAs, and a standing route for agents to raise the cases the system didn’t handle.

Rebuilt artifact

Before

After

1. Five channels, five toolsOne shared inbox in Front
2. No shared queueRouting rules on arrival
3. Triage by arrival orderSLAs per category
4. Informal escalationWritten escalation path
5. No response time dataPerformance dashboards

Where it landed

• Response times reduced 35%.
• Customer satisfaction scores increased 25%.
• Agent productivity improved 30%.
• Four months on, the team was still running the routing rules, SLAs, escalation protocol, macro library and dashboards without us.
• Support volume kept growing. Headcount didn’t.