About the Scores

Each sermon is analyzed by an AI model that evaluates specific dimensions: theological depth, clarity, textual fidelity, and others. These are signals, not endorsements.

A high score means the AI detected strong evidence of that quality in the transcript. It does not mean the sermon is "good" or that you'll agree with its theology. Use the scores to filter and explore, then listen and judge for yourself.

Each metric has a specific rubric that tells the AI exactly what to evaluate. The rubric shows what the AI was looking for, so you can decide if its criteria match what you care about.

Clarity

9.0–10.0 Exceptional: perfect clarity; single idea emphasized throughout; flawless organization; memorable
8.0–8.5 Strong: clear dominant idea stated early; logical progression; accessible language; few tangents
7.0–7.5 Solid: main idea stated; generally logical flow; mostly accessible; some minor issues
5.0–6.0 Main idea present but fuzzy; some scattered sections; occasional jargon or unclear shifts
3.0–4.0 Main idea unclear; many tangents; poor logical connections; confusing language
0.0–2.0 No main idea; chaotic organization; unexplained concepts; listener completely lost

Exposition

10.0 Exceptional: deep attention to context/argument; extremely clear and consistently text-governed
8.0–9.0 Strong: faithful, clear, text drives the sermon throughout
6.0–7.0 Solid: shows contextual understanding and text-driven reasoning
4.0–5.0 Some exposition, but shallow, inconsistent, or frequently untethered from the text
2.0–3.0 Mentions a passage but does not explain it; poor context awareness
0.0–1.0 Scripture mostly absent; no real explanation of any text

• AI can be wrong. Models miss nuance, misread tone, and occasionally hallucinate. Treat scores as suggestions.
• Transcript quality matters. Poor audio or transcription errors affect scores. Some sermons are penalized unfairly.

Prompts and models evolve. When I make significant improvements to a metric's rubric, I re-run analysis on affected sermons. This means scores can change over time.

Each episode shows when it was last analyzed. If you notice a score that seems off, it may be from an older rubric version. Scores are not permanent.