Contact-Center Domain Primer

For the NiCE technical round (stage 2) · the layer cake, the vocabulary, the metrics, your bridges

The technical round is your exposure point — but the bar isn't "be a 10-year contact-center veteran." It's fluency: sound like you understand how the machine works and what good looks like. This sheet gets you there. Read the layer cake and the metrics until they're automatic; those two carry most conversations. Your honest position: "new to the contact-center stack specifically, fluent in the job it does — deflection, adoption, integration, ROI."

Mental model

The layer cake

A contact center is a stack of layers. If you hold this model, you can place any product or buzzword someone throws at you.

Layer	What it does	NiCE	Google
Platform (CCaaS)	Runs the whole center: phone lines, routing, agent desktop, omnichannel, reporting	CXone Mpower	CCAI Platform
Self-service / bot	The "brain" that resolves a request with no human (voice + chat)	Enlighten Autopilot, Cognigy	Dialogflow CX
Agent assist	Real-time help to the human agent (suggestions, answers, summaries)	Enlighten Copilot	Agent Assist
Analytics	Insight on every interaction (quality, sentiment, trends)	Enlighten / Interaction Analytics	CCAI Insights
Workforce (WEM)	Forecasting, scheduling, quality + coaching of agents	CXone WEM	—

Use it CCaaS = Contact Center as a Service (the cloud platform). WEM = Workforce Engagement Management. The platform layer is where NiCE competes with Genesys, Five9, Amazon Connect, Talkdesk — good names to know.

How it works

How a contact moves

Be able to narrate this. It proves you understand the plumbing the AI sits on.

1 · Arrive

Channel

Customer calls, chats, emails, or messages.

→

2 · Greet

IVR / bot

Voice menu or virtual agent triages and tries to self-serve.

→

3 · Route

ACD / routing

If a human is needed, route + queue to the best agent.

→

4 · Resolve

Agent (+ assist)

Human handles it, helped by AI suggestions + knowledge.

→

5 · Learn

Analytics

Wrap-up, sentiment, QA, feed insights back.

The whole game Everything NiCE's AI does is to push resolution left — resolve at step 2 (self-service) instead of step 4 (a person). That's deflection / containment. It's literally what your Ting self-scheduling and Autzu redesign did.

Vocabulary

The alphabet soup

Term	Plain meaning
IVR	Interactive Voice Response — the automated phone menu. Voice only.
ACD	Automatic Call Distributor — routes + queues a voice call to the right agent.
Routing	The rules deciding where a contact goes; "omnichannel routing" extends it to chat/email/social.
IVA / VA / bot	Intelligent Virtual Agent — the self-service bot (voice or chat).
NLU	Natural Language Understanding — turning "I want to cancel" into a known intent.
Intent / entity	Intent = what the user wants; entity = a detail inside it (date, account #).
Containment / deflection	Resolving a contact in self-service, with no human. The core ROI lever.
Self-service	"Unassisted" = bot/IVR only; "assisted" = a human in the loop.
Knowledge mgmt (KM)	The knowledge base that feeds both self-service answers and agents. (NiCE: Autopilot Knowledge.)
Dialogflow ES vs CX	Google's bot builder. ES = simple; CX = enterprise, complex multi-turn flows.
WFM / WEM	Workforce Mgmt (forecast + schedule agents) / Workforce Engagement Mgmt (adds QA + coaching).
CCaaS	Contact Center as a Service — the cloud platform model (CXone is one).

The currency

Metrics that matter (this is the job)

The JD says "establish and monitor business success criteria." These are those criteria. A consultant's whole value is moving these numbers — speak them fluently and you sound like an insider.

Metric	What it is	Good =
Containment / deflection rate	% of contacts resolved by self-service	↑ up
AHT	Average Handle Time — time an agent spends per contact	↓ down
FCR	First Contact Resolution — solved on the first try	↑ up
CSAT / NPS	Customer satisfaction / loyalty score	↑ up
Cost per contact	What each interaction costs to handle	↓ down
Adoption / utilization	% of users/agents actually using the deployed AI	↑ up — your deliverable

Tie-in Your wins already speak this: Ting self-scheduling = deflection + onboarding time −25%; Autzu = call volume −15%; the platform GTM = adoption 90%. Relabel them in these terms and they land as contact-center results.

Their products

NiCE's stack in this language

Product	Layer	What it does
CXone Mpower	Platform	The flagship CCaaS — IVR, ACD/routing, agent desktop, omnichannel, WEM.
Enlighten	AI models	NiCE's CX-specific AI, trained on a huge labeled customer-conversation dataset.
Enlighten Autopilot	Self-service	Customer-facing bot for deflection/containment. Autopilot Knowledge = the KM behind it.
Mpower Agents	Agentic AI	No-code AI agents that take action end-to-end (self-service → mid-office → fulfillment). Built in Mpower AI Studio.
Cognigy	Conversational AI	Voice + chat bots; acquired 2025 to own the bot layer.
Enlighten Copilot	Agent assist	Real-time help to human agents.
Virtual Agent Hub	Integration	Plugs 3rd-party bots (Google Dialogflow ES/CX, Microsoft, Amazon) into CXone.

Google angle The JD names "Google Dialog Flow" because it's the canonical self-service engine. NiCE both integrates it (Virtual Agent Hub) and competes with it (Autopilot/Cognigy). The Cognigy buy = NiCE wanting customers on its bot, not Google's. A consultant often drives that migration.

Your bridges

Real experience → domain language

Don't claim contact-center tenure you don't have. Do connect what you've done to what they do — in their words.

You've done	Say it as
Autzu: integrated an offshore contact center into your app; business–tech liaison	"Hands-on with contact-center operations + the integration layer — I sat between the business and the platform."
Ting self-scheduling (SMS, −25% onboarding time)	"Built self-service that deflected contacts and cut cycle time."
Autzu UI redesign (−15% call volume)	"Drove deflection by removing the reasons people contacted support."
50K-user migration, 10+ teams, 99% uptime	"Delivery + program leadership on a complex enterprise cutover."
Building agentic AI solo (Claude Code, tool-calling, flows)	"I design conversational/agentic flows hands-on — same muscle as Dialogflow CX or Mpower Agents, different tool."

Honest line If pressed on depth: "I haven't run a CXone deployment — that's the part I'd ramp on. What I bring is the adoption-and-ROI job around it, plus hands-on AI fluency, and I learn the stack fast." Confident on the job, curious on the tool. No hedging on your wins.

Do this

Develop the Autzu story

This is your single best contact-center asset and the technical round will dig into it. Fill these in from memory so you have a crisp 60–90 sec version with real specifics.

The platform: what software did the offshore contact center run on? (Zendesk? Freshdesk? a dialer/telephony tool? — name it.)
The integration: what did "integrated into our app" actually mean? (API calls, ticket sync, embedded support widget, telephony hookup, data passing both ways?)
Your role: as business–tech liaison, what did you decide / unblock / translate? (requirements, edge cases, prioritization between offshore needs + dev capacity?)
Managing offshore + local dev: how did you keep two teams aligned across a handoff? (your matrixed-influence muscle — name the mechanism.)
The outcome: the call-volume −15% — from what, over what period, why did it drop? (self-service? better routing? fewer defects?)
The NiCE translation: "That's the same shape as what a consultant does here — make the support tech actually work for the operation and drive the metric."

Rehearse

Likely technical-round probes & how to approach

"Tell me about your experience with contact-center software."

Lead with Autzu (the integration + liaison), then the deflection wins, then honest ramp on CXone. Don't open with the gap.

"How would you improve a client's containment rate?"

Structure it: look at why contacts come in (top intents) → which are automatable → design/improve self-service for those → measure deflection + CSAT, watch you're not just deflecting into frustration. Show the reasoning, not a product pitch.

"Walk me through designing a self-service flow."

Intent → what info you need (entities) → happy path → fallbacks/escalation to a human → knowledge source → measure containment + where it breaks. This maps to the agentic flows you build solo.

"How do you think about an integration?"

Use a real one (Autzu, or your Stripe/webhook work): the contract (API/data), the failure modes, who owns what, how you test. Demonstrate you reason about systems, not just features.

"What's the hardest part of driving adoption?"

People, not tech: change resistance, trust in the AI, unclear ownership. Your address-serviceability + migration stories show you push change through a matrix.