What’s driving CX: Agent, Bot, or Product/Process? Find out with new tools in Ask Spotlight .

What’s an Experience Driver?

Experience drivers tell you which part of the customer journey caused a good or bad experience and why, helping you find and measure root cause. AI labels are applied for different aspects of the customer experience, rating them as positive, negative, or neutral.

Each of your interactions is now automatically rated across three Experience Driver categories:

• Agent Handling — how your human agent(s) handled the interaction
• Bot Handling — how your bot/IVR handled it
• Product/Process — whether there was a problem with your product or process

Note: these drivers are only applied where relevant, so a conversation with no bot involvement will not receive a ‘Bot handling’ rating.

Each positive, negative, or neutral rating is paired with a one-sentence explanation of why it was applied. For example, if Bot Handling comes back negative, you might see something like: “The bot misunderstood the customer’s question, looped them through the menu, and they left without resolution.”

Now that these ratings and explanations are explorable in Ask Spotlight, they will be invoked when relevant, and it will be easier to quantify the root cause behind your CX questions.

What’s new

Interaction counts, aggregates and search can now be filtered by driver and rating through Ask Spotlight, or any MCP-connected assistant. You can scope a question to one driver and one rating, then break the result down by contact reason, team, market or any other dimension you already use.

That lets you ask which part of the experience is going wrong, rather than just how badly an interaction went overall. You might find Agent Handling is consistently positive across a market where Product/Process is consistently negative, which tells you your team is holding the line on something the product is causing.

Example prompts

• “How many interactions in the last 5 days had a negative Product/Process rating?” – A simple count that tells you how often the product itself is the problem.
• “Break down experience drivers by team, so I can see if negative experience is coming from agents or from the bot.” – Useful when you need to know whether a problem is a people issue or a tech issue before you decide what to fix.
• “CSAT dropped 8 points last week for Billing, what’s driving it?” – This is the kind of question Ask Spotlight can now answer more precisely, filtering down to that contact reason and surfacing the dominant driver instead of someone reading transcripts by hand.
• “Show me every interaction where Agent Handling was negative but Product/Process was positive.” – This is a clean coaching list you can build. The process worked, so nothing outside the agent’s control can explain the outcome. Whatever went wrong, went wrong in the conversation itself. Reviewing these interactions is a better use of a team leader’s time than working down a list of low scores (where the frustration could’ve been the product/process).

Experience Score v. Experience Drivers

Though both are signals helping you understand the overall customer experience for each interaction, these are currently powered by two separate systems:

• Experience Score looks at signals inside the interaction — sentiment, frustration, gratitude, whether the issue was resolved, how the conversation ended — and predicts how satisfied the customer was overall. It’s one number for the whole interaction, and it answers: how did this go for the customer?
• Drivers are powered by an LLM rating each leg of the journey. Product/Process, Bot Handling and Agent Handling are each rated positive, negative or neutral with a one-sentence explanation. They answer a different question: which part of the journey caused the outcome?

Note: Drivers aren’t an input to the experience score (yet), so the two can seemingly ‘disagree’.

Example. A customer contacts you about a cancelled flight, angry from the first message and still angry at the end. The agent answers fast, confirms their entitlement and rebooks them.

The Experience Score is low: frustration=high, final sentiment=negative. But Agent Handling is positive. Read together: the interaction went badly for the customer, and the reason was the cancellation, not the agent.

---

A few things worth knowing

⭐ Experience drivers are stored from the 1st of September for interactions. Interactions before this date will not have the drivers present.

⭐ UI support for filtering by Experience Driver is coming soon. A new filter category will let you do the same filtering directly on Explore, Agent Activity, and other interaction tables.

If you have any queries about using this within Ask Spotlight or your own AI agent, reach out to Customer Success.