Summary

Message Scoring is Alta's built-in critic for the AI-written messages in your campaigns. Luna grades each message on multiple criteria, scores them out of 10, and (when something falls short) suggests an exact prompt tweak you can apply in one click.

Who this is for

Anyone tuning a campaign before launch — SDRs, growth managers, RevOps — who wants signal on whether a draft is strong without waiting for replies to find out.

Before you start

The message must already be generated by Alta inside a campaign touchpoint (email or LinkedIn). Scoring evaluates the rendered output, not raw prompts.

Step 1: Open the message in your campaign

Go to the Campaigns tab, open the campaign, navigate to Touchpoints, and click the email or LinkedIn step you want to evaluate.

Step 2: Open the Message Scoring dialog

Trigger Message Scoring from the message preview. The dialog opens with the title Message Scoring, Luna's avatar, and an intro that adapts to your score:

If your average score is high: "This is already a good message, but I think we can make it even stronger with just a few small tweaks, here they are:"
Otherwise: "This message could be better, I think we can make it even stronger with just a few small tweaks, here they are:"

Step 3: Read the average score

On the right side of the header you'll see a circular progress meter showing the average score out of 10. Below it, a blue-bordered pill shows the message reading time in MM:SS.

Step 4: Review each criterion

The body groups individual scores by category. Each row shows:

The criterion name (e.g. clarity, personalization, length, CTA — varies by Email vs LinkedIn).
The score from 0 to 10, color-coded:
- Green — High (strong on this criterion)
- Orange — Medium (room to improve)
- Red — Low (clear weakness)

Click a row to expand a justification — Luna's plain-English explanation of why that score was given. Rows with a suggested improvement are open by default.

Step 5: Apply suggested prompts

If a criterion includes a suggested prompt, you'll see it inside the expanded row with an Add to logic button. Click it to append that instruction to your message prompt. After you click it the button becomes Added and disables, so you can see at a glance what's already in.

Step 6: Apply everything at once

In the footer there's an Apply all prompts button. Clicking it appends every available suggested prompt to your message prompt and closes the dialog. If there are no suggestions available, the button is disabled and the tooltip reads "No suggested improvements available".

Step 7: Improve and re-score

The footer also shows an Improve Message button. Clicking it closes the dialog and routes you to refine the message — once the new version is generated you can re-open Message Scoring and see the score change.

Tips and common pitfalls

Color tells the story fast. Scan for red rows first — those are where you'll get the biggest lift.
Reading time is part of the score story. A 02:00+ reading time on a cold email is usually a flag — trim ruthlessly.
"Apply all prompts" stacks instructions. If you apply individual prompts and then click Apply all, you may double up. Pick one approach per pass.
Email vs LinkedIn criteria differ. The categories you see are channel-specific, so don't be surprised if a LinkedIn message scores against different rubrics than an email.
Re-score after edits. Scores are computed on the rendered output — change the prompt, regenerate, and re-open to see if you moved the needle.
Use it as a teaching tool. Reading the justifications is a fast way for newer reps to learn what "good" looks like.