Manual QA saw 5% of calls
Supervisors listened to a random 5%, so the bank had a fractional picture of service quality — and the supervisors were buried in scoring instead of analytics and coaching.
Customer story · Banking
From scoring 5% of calls by hand to every conversation analyzed — in a retail bank's contact center
OTP Bank's retail contact center serves help desk, retail and premium banking, debt collection, identity verification and customer-experience surveys. Supervisors could review about 5% of calls by hand; call categories were guessed by busy agents; coaching was a distant priority. Here is what changed when every call and chat went through Ender Turing.
- Industry
- Retail banking
- Channels
- Calls, chats, emails and web forms
- Functions
- Automated categorization, call-avoidance detection, AutoQA on 100% of conversations, self-training for agents, per-customer CSAT, sales funnels
- Source
- Two case-study posts published by the Ender Turing team in 2024, with product screenshots from the deployment
The results
What the contact center measured after the rollout.
Customer-experience results from the first case study, sales results from the second — each as the team published it.
Customer experience
- 81% → 86%First call resolution
- 96%Service level reached
- −25%Average handling time
- 47% → 62%NPS and CSAT scores
- −2%Abandoned calls
- −34%FTE: staffing needs down by 6 FTEs
Sales call center
- 7% → 15%Deposit penetration rate
- 15% → 45%Card reactivation
- +5%Upsell conversion in the call center
Customer satisfaction improved by 18 points, from 6.8 to 8.0 on a 10-point scale. Full write-ups: customer experience · sales growth.
The challenge
Quality was run on a 5% sample and guessed categories.
A retail bank's contact center is where most customer contact happens — about 30% of retail customers reach their bank by phone. The quality of those conversations decides NPS, and NPS moves the number of financial products per customer. OTP Bank could see almost none of it.
Call categories were a guess
About 2 of every 100 FTEs in banking contact centers go to sorting calls, yet agents pick the first category they see; reports came back with half the calls marked "Other".
Call avoidance went unnoticed
Fatigued agents stretched handle time, stayed on the line after the customer hung up, or left customers in silence — behavior nobody could see across thousands of calls.
Training was a distant priority
One supervisor for 10–15 agents, all of it manual: listen, score, give feedback, handle escalations. A one-to-one once a week was the best case.
The solution
Every call and chat through Ender Turing — six functions the team switched on.
Automated call categorization
Categories as granular as the operation needs, assigned to every call automatically by the AutoQA module — the head of the contact center sees exactly why customers call, without wrap-up codes.
How it works today →Call-avoidance detection
Handle-time tricks, dead line after the customer hangs up, long silences: tracked close to real time per agent, with the evidence one click away — separating burnout from a lack of knowledge from gaming the system.
How it works today →100% of conversations evaluated
Every call, chat, email and web form scored on unlimited scorecards seconds after it ends; managers get the reasons behind low CSAT in real time and supervisor time moves to focused evaluations and feedback.
How it works today →Self-training for agents
Individual training plans, playlists of the best and worst calls, quizzes and uploaded materials — so newcomers learn from real conversations without waiting for a supervisor.
How it works today →A CSAT per customer, not per survey
Surveys reach a selective sample; the sales team needed the satisfaction of every customer before calling. Ender Turing scores each conversation, so outbound lists start with customers who are happy with the bank.
How it works today →Sales funnels from the conversations
Why offers fail and where — automatically, per agent and per script — so the team builds funnels, gathers insights, tweaks the process and sells better again.
How it works today →
“Give your staff the opportunity to work with people and develop their communication skills, instead of wasting time on the unproductive task of manually selecting wrap-up codes.”
“My advice to everyone about to start using Ender Turing for outbound sales calls: make your first sales, build funnels, gather insights from the funnels, tweak the process, and sell better again!”
OTP Bank story FAQ
What readers ask about this deployment.
Which OTP Bank operation does this story cover?
The retail contact center: help desk, retail and premium banking, debt collection, identity verification and customer-experience surveys — calls, chats, emails and web forms. The numbers are that operation's own, measured against its baseline before the rollout.
What did OTP Bank switch on?
Automated call categorization, call-avoidance detection, evaluation of 100% of conversations on its own scorecards, self-training for agents, a per-customer CSAT for the sales team and sales funnels built from the conversations — each is a documented product function today.
Why do two customer-satisfaction figures appear?
The 2024 case study reports NPS and CSAT scores rising from 47% to 62%; the separately approved statement is that CSAT improved by 18 points, from 6.8 to 8.0 on a 10-point scale. They are kept in their own units and not recalculated into each other.
Can my contact center start the same way?
Yes: the free plan has no expiry — up to 5 agents and 100 calls per agent a month — and auto-scoring starts from the first call. Connect the platform you run or upload a week of recordings.
Security, hosting regions, sub-processors and response times are documented on the Data Security page; SOC 2 Type II examination completed.
Across our deployments
What changes when every conversation is analyzed.
- 100%Calls automatically scored (vs. 3–5% manual baseline)Quality coverage
- +19%Average lift in customer satisfactionCSAT
- −15%Reduction in average handle timeAHT
- 68hQA hours saved per specialist per monthProductivity
Averages across our contact-center deployments — each customer's own numbers are in the case studies. Case studies →
Free for up to 5 agents
Your contact center's version of this story starts with a week of calls.
Connect the platform you run or upload recordings: every conversation categorized and scored in seconds, call avoidance visible, coaching from real calls. Free plan with no expiry.