What we measure against
The Bible. We check what a preacher says about a passage against what the passage actually says, using the Berean Standard Bible. The text of every passage the preacher names is set beside the sermon; where a passage is not named aloud, the assessment says it is working from its own knowledge. Creeds and commentaries can help, but the Bible is the measure.
We do not take sides on questions faithful Christians disagree about — for example, baptism or how a church should be governed. Nobody is marked down for their tradition.
A few passages are missing from the oldest copies of the Bible — for example, the longer ending of Mark. Preaching from them is never held against anyone.
The first question
Does the sermon deny something at the heart of the Christian faith — who Jesus is, that he rose from the dead, that we are saved by grace through faith, that the Bible is God’s word, that God is Father, Son and Spirit? If it does, it scores near zero, and the page says why. Almost every sermon passes.
What lowers a score, and what does not
We sort what we notice into four kinds.
- Slips. A word misremembered, a verse quoted from memory. Everyone does it. Counted lightly, if at all.
- Fair readings. An emphasis another faithful reader might not choose, a point pushed a little further than the passage goes, a question Christians genuinely differ on. Noted, never counted against it.
- Making a passage say what it does not. Turning its meaning round, or adding something that is not there, in a way that would lead you to believe something untrue. This is what lowers a score.
- Denying a core belief. The first question again, if it comes up later in the sermon.
A score can also be held back by how the sermon handles its text as a whole — for example, drifting away from what the passage is about, or resting its argument on stories rather than on the text — even when no single point is serious enough to list. When that happens, the summary says so.
When it is a close call, the preacher gets the benefit of the doubt. We would rather miss a small fault than wrongly mark down a faithful sermon.
Quoting false teaching in order to show why it is wrong is never held against a preacher.
We judge only how a sermon handles the Bible. What a preacher says about the world — the news, history, figures, other faiths or politics — is not checked and never affects the score. Saying nothing about it is not agreement with it; those words are the preacher’s own.
What you see on a sermon’s page
The score, a short summary of what drove it, and anything that lowered it — each shown with the preacher’s own words and the Bible passage beside them wherever there is a line to quote, so you can check it for yourself. Slips and fair readings are kept but not shown: they are not faults, and a long list of them would make a good sermon look bad.
How sure we are
Sermons are scored by AI, using this standard. Some are checked more than once. Where the results differ more than usual, the sermon’s page tells you, and the reasons are the better guide than the number. Our standard says a person should look at those before they go up; nobody does, and we would rather tell you than let you assume it.
AI models are replaced over time. When we change, either the new one is tested on the same sermons as the old and agrees closely, or every sermon is scored again with the new one. While that happens, a sermon not yet re-scored keeps its earlier score, and each sermon’s page says which model scored it.
When this standard changes in a way that could change a score, its version number above goes up, and a score made under an older version is never shown as if it were current.