Score — By Project Moneyball

Message Center

AI-generated & advisory. Scores are produced by AI to support human judgment — the scores must not be the sole basis for employment, disciplinary, or compensation decisions. How we use AI →

I have read and understand this.

Identified Lifecycle Stage

Runtime Monitor

Call Details

Rep's Name:

Rep's Role:

Business Name:

Call Type:

Call Duration:

Final Score

Meaning:

Evidence Coverage:

Score Analysis

Recommendations (Next Steps)

MEDDPICC

Cross-Skill Diagnostic

How the connected skills lifted — or dragged — each other on the most recent scored call.

Run a call diagnostic to see the connected-skills story here.

Buyer-Side Diagnostic

The other half of the call — buyer engagement, sentiment shifts, buying vs. risk signals, and decision-readiness.

Planned

The CRO Dash

Team-level rollups — performance trends, coaching priorities, and deal-risk across your reps.

Planned

Integration

Connect your notetaker (Gong, Fathom, Fireflies…) to score calls automatically.

Roadmap

Anonymize Your Transcript

First, the reassurance: you may not need this. Score never stores your transcript. Every call is processed in memory for a single scoring run and then discarded — we keep only the derived results your report is built from (scores, confidence, and short evidence excerpts), keyed to a one-way hash, for a short window. The full details are in our Trust Center.

Anonymizing before you upload is an optional extra layer — the highest-security practice for teams under strict privacy, legal, or compliance rules who prefer to keep confidential details on their own machine entirely. It's playing it safe by choice, not patching a gap.

Here's how it works: scrub the confidential details out of a call transcript on your own machine — before it ever leaves it. Paste the prompt below into any AI assistant, add your raw transcript, and import the synthetic copy it hands back.

Why use it

Your raw call never leaves your control. De-identify it on your own machine — or a private/offline model — before uploading. Score only ever sees a synthetic copy.
The score doesn't change. The prompt preserves the call's structure, timing, and every behavioral gap, so the synthetic transcript grades exactly like the original.
Nothing real slips through. Company, people, and product names, the buyer's quoted pains, and commercial figures are swapped for consistent synthetic equivalents.
Built for compliance-sensitive teams who can't share customer names or PII with third-party tools.

How to use it

Click Copy prompt below.
Open your preferred AI assistant — ChatGPT, Claude, Gemini, or a local/offline model for maximum privacy.
Paste the prompt, then paste your raw transcript beneath it, and run it.
Review the synthetic transcript it returns — skim once for anything real that slipped through.
Save it as a .txt file and import that synthetic version from the Skill-Based Diagnostic page.

The anonymization prompt

You are a transcript anonymization assistant. Take the raw sales or customer-success call transcript I paste below and return a synthetic copy that is safe to share — one that reveals no real company, person, product, or confidential detail, while preserving everything about the conversation that reflects how the rep performed.

CARDINAL RULE: Substitute, don't rewrite. You are find-and-replacing identities and paraphrasing quoted content in place. You are NOT improving, summarizing, cleaning up, or re-authoring the call.

REPLACE the following with consistent, realistic synthetic equivalents:
- Company names — both the seller's company and the buyer's/customer's company, in every spelling, language, abbreviation, or typo. Recast each as a plausible fictional company in the same industry and of the same size.
- People's names — every participant and anyone named during the call. Map each real person to one fixed fake name and reuse it everywhere that person appears.
- Product, feature, and internal project names — replace with neutral, generic equivalents (e.g. "our platform", "the reporting module").
- Other named organizations mentioned in passing — other customers, partners, past employers — replace with unnamed or fictional equivalents.
- The buyer's quoted pains, needs, answers, and objections — paraphrase each into a synthetic equivalent of the SAME length and specificity that keeps its conversational function (a vague answer stays vague; a strong buying signal stays a strong buying signal). Do not make a thin answer richer.
- Small talk and personal details — kids, hometowns, hobbies, military service, health — re-skin into different but similar-length small talk.
- Commercial figures — prices, discounts, quantities, contract terms — replace with different numbers of the SAME magnitude (a six-figure deal stays six-figure; a 12% discount stays low-double-digit).
- Locations and any other detail that could re-identify a real party.

KEEP exactly as-is:
- Every timestamp, the turn order, the speaker labels/roles, the call duration, and the call type.
- Filler, false starts, repetition, and transcription noise ("um", "you know", half-finished sentences) — they carry conversational texture.
- Generic industry vocabulary and ubiquitous third-party tools/brands (e.g. Salesforce, Zoom, AWS, Excel) — industry furniture, not a fingerprint.
- Most importantly, every behavioral gap and weakness. If the rep never asked a discovery question, never recapped, or fumbled the close, the synthetic transcript must fumble it the same way. Do NOT add discovery questions, polish objection handling, or invent next steps — that would change how the call scores.

CONSISTENCY: The same real entity must map to the same synthetic value everywhere in the transcript (and across multiple transcripts, if you process more than one in this session). Keep an internal mapping as you go.

BEFORE YOU FINISH, self-check for leaks: re-scan your output for any real company name, person, product name, number, or identifying detail from the original. If any remain, fix them.

OUTPUT: Return only the anonymized transcript, in the same format and structure as the original. No commentary, no summary, no mapping — just the clean synthetic transcript.

Paste your raw transcript below this line:
------------------------------------------------------------

Account Settings

Manage your account, team members, plan, credits, and usage.

Roadmap

Trust Center

Our privacy, security, and data-handling commitments — how we handle your calls and your data.

Trust Center Overview Privacy Policy Terms of Service Security Data Handling AI Use