AI Analysis Improvements: Larger Spotlight sample size for broader insights & Experience Score change

Spotlight

Spotlight is a core AI feature on the EdgeTier platform. Instead of reviewing interactions one by one, Spotlight Summaries let you analyse multiple conversations at once, highlighting recurring issues, agent performance trends, and customer frustrations, without the manual effort.

The AI processes message content in two steps:

• First, it summarises each individual conversation.
• Then, it combines those summaries into a single, high-level overview of common themes and trends, allowing you to action these.

If you’re interested in more information in how Spotlight works, please refer to our existing documentation page.

How does this look for you?

Bigger samples, broader coverage

Previously, Spotlight typically analysed a sample of 100 interactions. Spotlight now analyses ~1,500–2,000 interactions per Spotlight, so you can capture insight across far more conversations without needing multiple runs.

Built to scale (with a small time trade-off)

To support the larger sample size, we rebuilt parts of Spotlight’s processing pipeline to handle more data efficiently and reliably. As a result, you may see a small increase in processing time in some cases — around 4–6 seconds — depending on the size and shape of the dataset.

More nuanced root-cause insights

We’ve also improved the prompts behind Spotlight to surface more detailed and nuanced root-cause information — helping Spotlight better explain the “why” behind trends.This works best when Spotlight is applied to a focused subset of interactions, such as when you’ve already narrowed down by a topic, spike, or trend.

How to get the best results

Spotlight still performs best on smaller, targeted sets of conversations — and now it can cover many more interactions within those focused views. If you have any queries on this update, or optimising Spotlight, please reach out to Customer Success.

Experience Score

You can find more about our experience score calculations here.

We’re making a small update to how Experience Score is shown at the individual interaction level:

• Before: single interactions displayed with decimals (e.g. 7.2, 5.4, 9.2)
• Now: single interactions will be shown as a whole number (0–10)

Important: Averages stay the same across the product — we’ll still calculate the same way and display to 1 decimal place (e.g. 7.2), so team/agent aggregate reporting retains granularity and is easier to compare.Why: we’ve had customer questions about meaningfully interpreting small decimal differences on single interactions (e.g. “what’s the difference between 7.23 and 7.66?”), and in practice it’s not significant. Whole numbers align better with how NPS/CSAT-style scores are typically understood.