@BowTiedDingo Classic script-as-checklist problem. QA agents on 'did you capture the answer', not 'did you ask the question'. The moment you score asking, reps read Q4 aloud to a prospect who already answered it.
@Hyken@BrittanyHodak Chase handle time alone and repeat calls quietly climb. A fast answer that brings the customer back twice isn't fast. First-contact resolution is the real speed metric.
@saguppa We learnt this with outbound call scripts: fixed script, one live line about the customer's last interaction. A wrong guess in that slot did more damage than no guess. How often was the AI's [X] off?
@annettefranz@castdotapp The 'I have a CSM' feeling was never about the human. It was someone remembering what you were promised last time. At 1,000 seats with rotating agents we lost that. AI with real memory can win it back.
@mardehaym The 'agent prepares, a person posts' line is the one boards skip. In BPO QA we ran maker-checker for years, and it only held when the checker's name sat on the entry, not the system's.
Thryv's net revenue retention fell from 94% to 89%.
New AI products grew. The old base leaked faster.
3 churn rules from my Airtel floors:
1. Save them before they cancel
2. Track repeat complaints, not CSAT
3. Pay retention like sales
Who owns churn on your team?
#SaaS#GTM
Hot take: 95% occupancy isn't efficiency. It's attrition on a delay.
No breath between calls.
No time for notes.
No time to learn.
Dashboard looks great. Then your best agents quit.
Plan the slack in. Dhyan rakho.
What occupancy do you plan for?
#ContactCenter#FutureOfWork
@HarriKonola Cold Calling 101 next. Most teams build RevOps tracking for meetings that never show. From 20 years on call floors: the daily calling rhythm and callback discipline decide show rates more than any dashboard.
@fivosaresti The step most orgs skip between stages: someone who owns speed-to-lead on inbound. On contact-centre floors, an inbound call answered late was worth a fraction of one answered fast. Tools don't fix that, ownership does.
@HamelHusain Same trap as contact-centre QA scorecards: shift the weights quietly each quarter and the 'quality' trend becomes a weighting trend. Did any model's rank flip purely from reweighting?
@aakashgupta Best kind of product decision: users kept saying GIF, the plumbing quietly became video, and nobody had to learn a new word. In support ops it's the same rule: fix the backend, keep the customer's vocabulary.
@jasonlk Ran BPOs for 20 years: the clients who demanded calls, SLAs and hard negotiations were the ones that stayed for years. Low-touch revenue feels easy until you realise you never learned what the customer needed.
@levelsio Works when both sides already trust each other. On contact-centre floors, the angriest customers were the ones stuck in async loops; one short call fixed what a week of messages couldn't. Calls aren't the problem, pointless ones are.
@pineaultamy Human agents learn this in week one: 'okay...' means 'not convinced yet'. We coached reps to pause and ask one more question. A voice agent that can't hear the dots will close calls that should stay open.
@kunal_twts Taking the call is the easy part. The money is in what happens after: the lead routed right, the appointment that actually shows, the angry caller handed to a human in time. That's where voice bots break.
@bradmenezes Embedded FDEs is the right call. On ops floors the best automation ideas never came from IT, they came from team leads who knew which step broke at 10 PM on a Saturday. Put the FDE next to them.
@darynakulya@bot Win-back is where bots earn their keep. In contact centres the retention desk saved more revenue than sales booked, mostly by calling within a day of the cancel. Does your bot trigger on intent or on churn?
@aaditsh Same shift in ops hiring. A resume says you ran a 500-seat floor. A post on how you'd fix a broken SLA shows how you think under pressure. The second one gets the call.
@hasantoxr We recorded every customer call for 20 years. Capture was never the problem. Finding the one moment that mattered was. Memory is a retrieval problem, not a recording problem.
@aakashgupta Contact centres run on this exact math. One agent late for shift doesn't cost one salary, it costs the queue: abandoned calls and missed SLA for the next hour. Lateness is never a personal cost.