Your Contact Centre Is Sitting on a Goldmine. Spotlight Helps You Dig

There’s a number that should bother every customer experience leader: 2-5%. That’s the share of contact centre interactions that typically get reviewed manually. In a team handling thousands of conversations a week, that means the overwhelming majority of what your customers are telling you goes completely unexamined. Not because nobody cares, but genuinely because there aren’t enough hours in the day!

This is the fundamental problem we built Spotlight over a year ago to solve, and after watching how our customers have been using it, we think it’s worth taking a proper look at what it does, how it works, and where it’s gone since we first launched it.

The problem with reading the transcripts

When something goes wrong in a contact centre, the instinct is usually correct: go to the conversations. Find out what customers are actually saying by reading through the chats, the emails, the call notes. Get close to the data, essentially!

The instinct is right, but the method doesn’t scale.

Say your NPS drops after a service change. You filter down to the relevant interactions, and you’re looking at 400 conversations. You could start reading. You could assign someone to read. You could build a spreadsheet and tag themes manually. Or you could still be doing that three weeks later, by which point the moment has passed and the insight is stale.

The contacts keep coming regardless of whether you’ve caught up.

This is where Spotlight fits in. Rather than replacing the conversations, it synthesises them. Run it on any filtered set of interactions in EdgeTier, like contact reason, time period, sentiment, tag, team, or any combination, and within about a minute, you get a structured summary of the themes driving that group of conversations. What customers are frustrated about. What keeps coming up. What the shape of the problem actually looks like.

Critically, Spotlight analyses up to 2,000 interactions at once and samples intelligently from within that set. So you’re not getting one analyst’s read on fifty tickets, but a pattern across your real data, at a scale no human team could match.

This is not a novelty feature

We want to be direct about something: AI in customer service has attracted a LOT of noise. Plenty of tools promise insight and deliver dashboards that tell you what you already knew. Spotlight is EdgeTier’s approach to contact centre analytics, and we’ve been careful to make sure it earns that description rather than just wearing it.

The reason it works is that it builds on something concrete. EdgeTier already tags and categorises your interactions through AI Tags, which means by the time you run Spotlight, you’re working with data that’s already structured and filtered to the conversations that matter. The LLM layer analyses each conversation individually before combining them into a single coherent summary. 

AWS Bedrock handles the infrastructure to make that fast and consistent across large samples.

The result is a summary that tells you what the specific frustrations are, say, incorrect supplier contact details and slow follow-up, or refund delays and confusing policy language. Named, specific, and actionable.

That specificity is what makes it worth acting on. And that’s what separates a useful AI feature from a gimmick: whether the output changes what you do next.

Only 30% of contact centres are currently using AI to generate insights, even as broader AI adoption has accelerated (Verint). That gap exists because a lot of AI tools haven’t earned the trust required to change how teams work. The bar is higher than it looks.

What Spotlight has become

When we launched Spotlight, the core value was clear: stop reading transcripts one by one, and start seeing patterns across hundreds of interactions at once. That still holds, but we’ve kept building, and the capability has grown in ways that have changed how teams actually use it day to day.

Let’s make it concrete: 

Imagine a travel brand on a Thursday afternoon when a wave of flight cancellations hits and the contact centre volume doubles within two hours. By the time a manager has read fifty conversations, five hundred more have arrived. They run Spotlight on the full volume. Within a minute, the picture is clear: the top contact driver is not the cancellation itself. Customers understand disruption happens. What they can not find is clear information about how to rebook. The frustration is concentrated around one specific failure in the communication flow, not the event. That’s a finding that changes what happens next, and it’s a finding that would have taken days to surface manually (if it surfaced at all!) The manager then flips to the Agent Review tab and sees something else worth knowing: agents are handling the emotional weight of the calls well, but a knowledge gap around the rebooking policy is creating inconsistent answers. Two problems, identified in the same workflow, in the time it used to take to read a handful of tickets. By Friday morning they’ve updated the rebooking SMS, briefed agents on a consistent policy answer, and the contact volume has already started to drop.

That kind of depth is what the most recent round of Spotlight updates has been designed to support. Themes are now backed by a browsable view of the actual interactions behind them, so findings can be verified and shared with confidence. Any theme can be expanded into its own Spotlight run, drilling from a summary into the specific detail underneath (what we call Spotlighting the Spotlight!) And Agent Review, available across both Explore and Agent Activity, brings agent performance into the same analytical frame as contact reasons. This is conversation intelligence in the truest sense: understanding not just what customers are saying, but how well your team is meeting them. 

It matters particularly given that 61% of contact centres are now reporting more emotionally charged customer interactions than before (Calabrio) Knowing what customers are contacting about and how agents are responding to it are questions that belong together.

Learn more: Go Deeper with Spotlight

The next evolution for Spotlight

Spotlight has kept getting sharper at surfacing what’s in your data. But one friction point remained: getting a specific answer still required knowing what to filter for and interpreting the output. Brilliant for deep investigation. Less ideal when you just need a quick answer at 9am on a Monday.

So we built Ask Spotlight.

Type a question in plain English — who is struggling with renewal queries this week, where did frustration spike on Tuesday, what are customers saying about the new returns flow — and get an answer in seconds, complete with charts and tables, drawn from your real EdgeTier data. No filters, no wait, no knowing where to look.

If Spotlight gave teams the power to stop reading transcripts one by one and start seeing patterns, Ask Spotlight gives everyone in the business the power to just ask.

Learn more about Ask Spotlight

What this actually unlocks

Contact centres have always been one of the richest sources of voice of the customer data a business has. The conversations happening every day contain real signals about product problems, policy failures, unmet expectations, and where the experience breaks down. Most of that signal has historically been locked away, not because it wasn’t valuable, but because accessing it meant manual effort that most teams couldn’t sustain.

Spotlight changes that equation. Not by replacing human judgment, but by doing the work that makes human judgment possible at scale. You still decide what to filter for. You still decide what to do with the findings. You still read the conversations that matter most. But you get there in minutes rather than weeks, and you get there with confidence that you’re looking at the full picture, not a sample of a sample.

The goldmine was always there. Now you have better tools to dig!