What happens after you finalize?

When you click Finalize in the contact card, ST300 sends the entire conversation to the AI evaluation engine. The engine reviews the exchange against the persona's defined end conditions and your organisation's configured metrics, then writes corrections for any messages where a different approach would have been more effective.

Evaluation takes 10–20 seconds. Once complete, the scenario appears as finalized in your list and is ready to review.

How to read corrections

Open any finalized scenario to see the full conversation replay. Messages that received a correction are highlighted with a warning border and a “Correction” badge. Below each flagged message, a correction card shows:

• A suggested alternative — an example of what you could have said instead, written as a ready-to-use message with no preamble
• Per-metric explanations — for each metric the correction addresses, a concise note on why the suggested message scores better on that dimension

How to interpret the feedback

Corrections are not about finding the single “correct” answer. Sales conversations do not have one right path. What the correction engine looks for is:

• Specificity over vagueness. Generic claims (“great platform,” “really flexible”) rarely move buyers. Specific, relevant statements do.
• Discovery before pitch. Jumping to a solution before understanding the problem is one of the most common patterns flagged by the engine.
• Objection resolution. Acknowledging an objection and then moving on is not the same as resolving it. The AI tracks whether each defined objection was genuinely addressed.
• Tone calibration. For personas with formal or indirect communication styles, casual language is flagged even when the content is good.

Requesting additional corrections

The automatic evaluation covers the messages the engine flagged on its own pass. But you may have messages you want a second opinion on — moments where you felt uncertain about your phrasing or tried a new approach.

On any of your messages in a finalized scenario, a Request correction chip appears. Click it to submit that specific message for targeted AI evaluation.

How to use results to improve

Reading corrections is step one. Improvement comes from applying them. Here is a pattern that works well:

• Identify one pattern per session. Do not try to fix everything at once. Pick the most common type of correction — “I jumped to features before discovery” or “I was vague under pressure” — and make that the focus of your next scenario.
• Re-run the same persona. Use the same persona profile in a new scenario. You already know the objections — now you can practise handling them with what you learned. Comparing the two conversations is the clearest way to measure improvement.
• Practise the specific exchange. Take the corrected message and the suggested alternative. Write your own version in a new reply and keep going from that point. Muscle memory comes from repetition, not reading.
• Progress to harder personas. When you are consistently finishing scenarios with few corrections on a Medium interest persona, step down to Low. The conversation does not just get harder — it requires a different approach from the opening line.

Manager view — reviewing your team

If you have the manager role, you can access your team's finalized scenarios from the Management → All Results page. Each scenario shows the rep's full conversation and any corrections applied.

Use the Request correction feature on specific messages to add targeted coaching notes — the same mechanism reps use on their own sessions. This lets you run asynchronous coaching without scheduling a separate debrief.

Next steps

Ready to build more personas for your target markets? Read the persona creation guide →