Head of
Contact Centre
I run the service function end to end: the line, quality, training and the economics of a contact. I have formal contact-centre training and experience designing a unit from scratch, not only running one that already exists. A round-the-clock line is under my management right now, and its numbers are measured every week.
The corporate contact centre programme
I prepared a corporate contact centre programme for a holding company: concept, budget and a three-year roadmap. Programme owner since February 2024, with a horizon to mid-2028 and certification against ISO 18295 as the goal. Sponsors from the sales and the customer experience directorates, sign-off at chief executive level.
This is not a concept written for a slide deck. It rests on a gap analysis against the standard across twenty-two clauses, with an honest split: seven requirements closed, eleven partial, four gaps named outright. The three-year plan grew out of that analysis.
The line I run today
- Round the clock with no night-time dip: median first reply from a live agent is 12-13 seconds against a 30-second target, and that is not an auto-greeting
- Time to a substantive answer cut from 100 seconds to 40
- Ninety-one per cent of conversations answered inside a minute, at roughly 2,800 contacts a month
- Customer satisfaction lifted from 42% to 93-94% against a target of 80%
Cost per contact and workforce planning
I calculate cost per contact from a method rather than from industry benchmarks, and it depends heavily on the type of traffic: 213 roubles on a high-volume line against 4,228 roubles in industrial software, where a contact means investigating a technical case and reproducing a fault. Both figures use the same boundaries, including the manager's payroll.
The second unit figure is cost per minute of departmental work: about 10 roubles on the high-volume line against 210 roubles in industrial software. Two independent measures produced the same order of difference, which is how I check that the calculation holds together.
Headcount is planned from contact volume, not from a sense of how busy people feel. Once the cost of a minute is known, "do we need another agent" becomes arithmetic.
AI on the line, under the same quality standard
I put an AI agent into production as the fifth member of the night shift. The share of contacts carried through to a resolution rose from 34% to 82%, availability is 99.95%, with no outages. It handles about a third of the traffic.
- It is scored on the same sixteen parameters as people: 97.7%
- There is deliberately no softer standard for automation
- Otherwise the automated share grows while its quality becomes unmeasurable
Separately I built a real-time AI assistant for agents, covering over ninety per cent of routine contacts.
Major incidents
I own the company's incident plan: twelve failure directions, roles, activation thresholds, a communication order and compensation rules. Its scope reaches beyond support into product, engineering and finance. It also states which quality parameters are suspended during an incident window, so that an agent is not penalised for a fault someone else has to fix.
I took the department over the day after a major outage at an infrastructure supplier: the shift lead had resigned, the queue was flooded and daily satisfaction had fallen to 49%. The line did not stop for a single hour.
Formal background
- Three programmes at Apex Berg Contact Centre Consulting: contact centre operations management, ISO 18295, and finance in service for non-financial managers
- Jury member of the Russian national contact centre awards, CCGuru Awards, since 2025
- ABPMP member, internal auditor for ISO 9001
- Author of two books on technical support