Optimizing Support Ticket Management

Support teams face constant pressure to resolve cases faster without overloading engineering. For one Glean customer, valuable resources were tied up in avoidable tickets, MTTR (mean time to resolution) hovered at nearly two days, and agents spent hours manually triaging cases. Their goal: boost self-solves, improve MTTR, and reduce R&D reliance – without adding more tools. So they embedded Glean in Zendesk, giving agents prompts to quickly gather knowledge across all company data. In triage, agents use Glean to find similar tickets, summarize runbooks and past Jira investigations, and compile clear updates for customers or well-packaged escalations. That streamlined process now drives faster resolutions, smoother knowledge transfer, and consistent workflows—leading to: • 34% increase in self-solves with more future automation planned - this is incredible progress • 24% faster MTTR (1.9 → 1.5 days) • 2–4 hours saved per week for 85% of users (13–26 business days/year) • Reduced R&D involvement in lower-tier tickets By streamlining resolutions, knowledge transfer, and process consistency, the team achieved remarkable results – proof of what’s possible when AI is embedded into everyday workflows. Stories like this are energizing – showing how teams are using Glean to reimagine what they can accomplish.

Early in support, I responded to tickets in the order they arrived. Bad idea. I was constantly stressed, customers with urgent issues waited too long, and I missed patterns that could've prevented repeat tickets. Here's a simple triage system I used and you can start using it today. The 4-Tier Triage Framework Every morning (or start of shift), spend 10 minutes sorting your queue into these four tiers: Tier 1: Blockers (Handle first, within 1 hour) Customer cannot use core product functionality right now. Examples: "I can't log in" "Payment failed but I was charged" "Data is missing from my account" Action: Fix or escalate immediately. Tier 2: Escalation Risk Customer is angry, mentions legal action, or represents significant revenue. For tickets like this responding with speed without clarity will only create problems for you. Pace yourself to go fast. Understand the situation before responding. Watch for phrases like: "This is unacceptable" "I want to speak to your manager" "I'm cancelling my subscription" Action: Personalised response. No templates. Show you're listening. Offer a direct solution or timeline. Tier 3: Repeat Patterns (Batch and document) Multiple customers reporting the same issue. If you see 3+ tickets about the same thing: → Stop responding individually → Alert your team/engineering → Create a saved response for this specific issue and let the team know → Add it to your knowledge base or just update By doing this, you'll prevent 20 more tickets instead of answering them one by one. Tier 4: Everything Else (Handle within 24 hours) Questions, feature requests, general guidance. These matter, but they won't escalate if they wait. Action: Use templates as structure, but customize the opening line based on their tone and the closing with a relevant next step. When I implemented this, I had more time to focus on really complex tickets and work projects. I could actually think instead of just reacting. 2 Mistakes I Made (So You Don't Have To) → Skipping the morning triage: When I tried to triage "as I go," I always ended up in arrival order anyway. The 10-minute investment saves hours. → Not documenting T3 patterns: I'd notice the same issue 10 times but forget to tell anyone. Now I have a Friday ritual: review the week's patterns and flag or document. If you're feeling overwhelmed right now: → Tomorrow morning: Spend 10 minutes sorting your current queue into the 4 tiers → This week: Track one pattern (just one) and document it You're not bad at this. You just need a decision framework that's better than "whatever came in first." This system isn't revolutionary. But it works, and you can implement it in your next shift.

From 87 days to 16 days - an 82% reduction I’m working with a client who, as part of my assignment, put me in charge of a support team for one of its enterprise applications. When I was officially given the reigns of the team, the average open ticket age was 87 days. I had already done some work with the team that reduced that number prior to being put in the leadership role, but 87 is the first measurement I have. (Note to self: get better at collecting “before metrics - always bites me.) As of this morning, we are at a 16 day average. That’s an 82% reduction in the average age of open tickets - achieved in roughly 3 months. We’ve also: *Reduced the quantity of open tickets by 67% *Reduced cycle time by ~70% *Increased customer satisfaction *Increased throughput How did we (I say “we” bc I couldn’t have done it without the fabulous team sticking with me through multiple small changes) achieve this? * Simplify the workflow. When I first saw the workflow the team was using in Jira, it looked like a bunch of scribbles on a page. It was impossible to understand how work flowed through their system. We moved to a very simple, left to right flow. * “Blocked” is a state, not a workflow step. It means we need to resolve the block, not that we put it in the closet and forget about it. * Clarify next action to be taken when a team member frees up: we work items on a Kanban board sorted by priority and date. We work top to bottom, right to left. * Remove sub queues: each team member only “owns” tickets they are actively working. Also, we leave comment trails so any team member can work a ticket at any stage in the workflow. Team members being sick, taking vacation, etc. doesn’t mean tickets stop being processed. * Removed SLAs. The team had SLAs to “respond” to tickets within a small time window. This led to tickets being responded to immediately, causing distraction and leaving tickets owned by whoever responded first. A customer would be told “we’re working on this” but in reality no work was being done- often for weeks or months. These changes aren’t huge or revolutionary, but they allow focus. They ensure we “stop starting and start finishing.” If you’ve got a struggling team or two and could use help to deliver more effectively, let’s chat!

We recently analyzed our support tickets at Yoodli AI Roleplays, and the results challenged a lot of our assumptions. Our team felt stretched and the instinct was to hire. But first, we asked a very important question: What is actually happening in our support data? At Yoodli: Great support isn’t about zero wait times, it’s about solving customer problems quickly and painlessly. Here’s what we found: 1. Tuesday, not Monday, is the real spike. Ticket volume jumps ~19% on Tuesdays, making it the busiest day of the week. Monday feels worse because of backlog, but the real surge comes after. 2. Support demand follows a global rhythm . Peak volume hits at 10am ET, but there’s a second spike around 7pm ET, driven by international users starting their day. Support isn’t just a 9 - 5 problem. 3. Early-morning demand is real, and growing. ~14% of tickets arrive before the US workday even begins. If you’re only staffed locally, those tickets are aging before anyone sees them. 4. Weekends aren’t quiet. ~13% of tickets come in on weekends. Ignore them, and you’re starting every Monday behind. 5. Growth shows up in support first. We saw a ~38% week-over-week increase in tickets. If you don’t plan for that, your team feels it before your dashboards do. The takeaway: We don’t need more people everywhere. We need the right coverage at the right times. That means: → Staffing for peak hours and global surges → Planning for weekends so Monday isn’t chaos → Treating midweek spikes as a distinct problem If your support team feels underwater, don’t just hire without studying the shape of your demand first. Shoutout to our support team for the incredible work their doing to give our customers a great experience. Alan Camperson, Christian, Dewey.

Everyone in support obsesses over 𝘵𝘪𝘮𝘦 𝘵𝘰 𝘳𝘦𝘴𝘰𝘭𝘶𝘵𝘪𝘰𝘯. But the real bottleneck isn’t resolution. It’s 𝘵𝘪𝘮𝘦 𝘵𝘰 𝘤𝘰𝘯𝘵𝘦𝘹𝘵 . Before you can fix anything, the customer spends most of their time just waiting for us to understand the problem. That’s why I started tracking something different: 𝐀𝐯𝐞𝐫𝐚𝐠𝐞 𝐃𝐞𝐥𝐚𝐲 𝐩𝐞𝐫 𝐈𝐧𝐭𝐞𝐫𝐚𝐜𝐭𝐢𝐨𝐧. Not how fast the ticket was closed. Not how fast we eventually replied. But how long the customer was left hanging between each step in the conversation. The tricky part: it’s surprisingly hard to calculate. ▪️ Count every gap between messages → you’re mostly measuring customer delays, not ours. ▪️Count only customer → agent gaps → closer. Now you’re isolating the wait they actually feel. ▪️Make it SLA-aware → best signal. Under SLA = 0. Over SLA = only the extra hours count. Here’s how we calculate it: 1️⃣ Look at every gap between a customer message and the next agent reply. 2️⃣ If the gap is under SLA → count it as 0. 3️⃣ If it’s over SLA → count only the extra hours (in business hours). 4️⃣ Average those numbers across the ticket. This metric doesn’t tell you how good your agents are at solving problems. It tells you how good your system is at not leaving people hanging. And that’s what customers remember. Because solving the problem is expected. But the waiting? That’s the part that feels like friction. Takeaway: Stop asking “How long until the ticket was closed?” Start asking “How long did the customer wait at every step along the way?” That’s where trust is either built — or lost.

I recently sat down with the former VP of Global Customer Support at a $5B org. While there, she cut ticket volume by 25%. I asked her how she did it - and I'm shocked more CX leaders aren't doing the same. BACKGROUND Emily Ebersole, Customer Success Exec was the VP of Global Customer Support at Zapier. She helped reimagine customer support at the org, introducing live chat, automation, and AI - en route to slashing ticket volume by 25% while maintaining a completely flat headcount. But none of this happened with the flip of a switch. Most CX leaders either try to reinvent the wheel all at once or are paralyzed by the thought of innovating their processes. Both lead you to the same place: a dead-end. Instead, Emily took a fundamentally simple yet completely overlooked approach: She focused on one thing first, mastered it, then moved onto the next. First, she optimized for self-service by linking help articles to a contact form that led to a 15% decrease in ticket volumes. Then she made changes to their free user support, which cut another 10% of tickets. With more time and space available, she and her team then, and only then, tackled live chat by reworking the team structure and allocating more people toward high-value service offerings. Only their highest-paying customers had access to live chat to start - ensuring her team could handle the workload while making sure customers would get real value from it. Once proven out, they expanded live chat to mid-tier customers. And then, boom - generative AI hit the masses. Instead of waiting around to see what may or may not be possible with it, Emily and her team dove in head-first to experiment and see how it could help with support. In short order, a team member created a self-serve tool with ChatGPT and embedded it inside of Zendesk. Soon after, they scaled this tool throughout the support organization to drive further efficiency within their operations. TAKEAWAY If you want to improve customer support, you can't boil the ocean. You'll get nowhere fast. But if you isolate one area and optimize it, you gain the momentum, time, and space necessary to move onto another area and do the same. And when you get into the habit of doing this, you not only drive improvement in efficiency and effectiveness - you set yourself up to be ready to act quickly when a massive opportunity (like generative AI) comes along. I recently spoke about this with Emily Ebersole, Customer Success Exec on the CX Innovation Playbook Podcast. Listen on Spotify or Apple Podcasts.

I automated 95% of our customer support flow over the weekend (for real) thanks to AI tools that turn anyone into a technical product builder. There are 3 steps to the workflow: First is a RAG AI agent that's trained on our user guide and a corpus of customer support email threads. When a support ticket comes in, I just copy it in here and get an email response that I can send back to the user This agent works because it's built on a continuously-improving knowledge base, curated by a second AI system. This system periodically reads through a database of support interactions to find new learnings to add, which in turn makes the RAG agent more capable over time At the bottom is a system that collects new support email threads that I resolve and adds them to a database. I have third AI system running in the background that automatically classifies and tags threads in my inbox for this purpose These 3 AI systems working together take all of the CS heavy lifting off my shoulders, allowing me to spend valuable time elsewhere We don't automate to remove people from the equation - we do it to leverage ourselves better where we're needed It's how we manage to scale so effectively at Aomni as a 5-person team

Last week, I shared this idea: “Support will stop being measured by how much work it absorbs and start being valued by how much friction it removes from the business.” This week, I want to go one level deeper because removing friction isn’t accidental. It’s designed. Most friction in support doesn’t come from bad intent or lack of effort.  It comes from fragmentation: - Fragmented systems - Fragmented ownership - Fragmented context - Fragmented decisions Support teams inherit that fragmentation every day. They’re asked to resolve issues that cut across products, engineering teams, configurations, versions, regions, and time zones.  They are often doing this without a unified view of what’s already known or what’s already been tried. So the question isn’t: “How do we get faster?” The real question is: “How do we make better decisions earlier?” That’s where the next shift happens. In the next era of support, productivity won’t come from closing tickets faster, it will come from orchestrating decisions across complexity: - Knowing which paths are proven vs. experimental - Knowing when a problem is self-serviceable vs. engineer-led - Knowing when escalation adds value and when it just adds noise - Knowing how today’s difficult case informs tomorrow’s prevented case This is why the role of support is changing so fundamentally. Support isn’t just reacting to issues anymore. It’s becoming the system that teaches the business where friction actually lives and how to remove it upstream. When support operates this way: - Customers feel less effort, not just faster responses - Engineers spend more time on judgment, not archaeology - Leaders get signal instead of volume - The business learns, not just survives incidents This is what frontline productivity really means in a complex world. - Not doing more work. - Not absorbing more pain. But designing systems that learn, adapt, and reduce friction before it compounds. More next week....this is where the conversation gets very real. #FrontlineProductivity #CX #TechnicalSupport #SupportLeadership #KahunaLabs John Ragsdale Sanjeev Gupta