The Problem
Every contact centre has a category of call that is, in truth, completely unnecessary — not because the customer is unreasonable, but because they're missing one small fact and the only way to get it is to phone up. The classic signature is the "status check": Where am I in the queue? Has my payment gone through? When is my engineer coming? Where am I on the upgrade list?
These calls are pure cost. They don't need judgement, empathy, or problem-solving — they need a single data point the organisation already holds but hasn't shown the customer. Yet they get optimised like any other call: better routing, a slicker bot, a shorter prompt. All of that misses the real opportunity, which is to make the call never happen.
The AI Use Case
Point AI at your conversational data and ask it a different, out-of-the-box question: for each intent, "is there a single data item that, if the customer could simply see it, would remove the need for this contact entirely?"
This flips the usual analysis. Instead of "how do we handle this intent better," the AI looks for intents whose entire reason for existing is an information gap — and names the specific data item that would close them.
- Cluster the intents — group conversations by what the customer actually wanted (building on the kind of categorisation in emerging-topic detection).
- Spot the "status check" pattern — intents where the customer is really just asking to see a value, not to change anything.
- Name the missing data item — the one field (queue position, upgrade rank, delivery ETA, application stage) that would answer them.
- Estimate the prize — how much volume that one intent represents, so you can prioritise.
Worked Example: The Upgrade Priority List
Imagine a provider where thousands of customers are on a waiting list for an upgrade. A steady stream of them call in asking the same thing: "Where am I on the list? Am I getting closer?"
- AI mines the conversations and surfaces a high-volume intent: "check upgrade priority position."
- It identifies that every one of these calls resolves to showing the customer a single value — their rank on the list — which the business already knows.
- The recommendation isn't "build a better bot for this intent." It's "expose the upgrade position in the app and send a notification when it changes."
- Once that data is surfaced proactively, the customer can see their position any time — and the reason to call evaporates. The intent doesn't get contained; it gets eliminated.
Before: customer anxious → phones up → agent looks up rank → reads it out After: position shown in app + "you've moved up" push → no call needed
One surfaced data item removes an entire call driver. Do that across the top handful of status-check intents and the effect on volume compounds.
The Value
- Demand removed, not shifted — the lowest-cost call is the one that never happens. This is deflection by design, upstream of any bot.
- Better experience — customers get instant, anytime answers instead of queuing to hear one number.
- Agents freed for real work — status-check volume is dull, repetitive, and squeezes out the complex calls that need a human.
- Proactive, not reactive — a timely "you've moved up the list" notification pre-empts the question entirely.
- Compounding returns — each eliminated intent is a permanent volume reduction, not a one-off saving.
The mindset shift: the goal isn't always to answer the question faster — sometimes it's to make the question unnecessary. AI is uniquely good at spotting these because it can read why people really call across thousands of conversations and reason about what single fact would have satisfied them.
What You Need to Build It
- A corpus of conversations — transcripts or intents to mine, ideally with volume data per intent.
- An LLM analysis step prompted to find information-gap intents and name the data item that would close each one.
- A channel to surface the data — app, portal, SMS, or proactive notification — plus the integration to pull the value from the system that already holds it.
- Prioritisation by volume & feasibility — tackle the biggest, easiest wins first.
- Measurement — track the call volume for each targeted intent before and after, to prove the reason really did disappear.
Explore More AI Use Cases
Part of a growing series of practical AI use cases for contact centres. Browse the full set on The AI Use Cases hub, and see the wider picture in 15 ways AI can transform your contact centre. Got one you'd like covered? Get in touch.