How scores work
The five criteria
Every rating scores the talk from 1 (poor) to 5 (excellent) on five criteria:
- Content value — Was the content useful, new or insightful?
- Delivery — Clear, well-paced, engaging presentation style?
- Expertise — Did they know the topic in depth?
- Audience engagement — Interaction, examples, handling of Q&A?
- Matched the title — Did the talk deliver what the title and abstract promised?
"Matched the title" is there because most attendees choose sessions by title. A talk that promises one thing and delivers another scores low here, even if the content was fine.
Averages shown on pages
A rating's overall score is the mean of its five criteria. Talk and presenter pages show the plain average of all visible ratings, per criterion and overall. "Would see again" is the share of raters who answered yes. A presenter's numbers include every rated talk they gave, at every conference.
Rankings (Top lists and "Audience pick")
Rankings use a weighted score so that a single enthusiastic rating can't beat a long, consistent track record:
- Recent talks count more. Ratings from the last 12 months have full weight, 1–2 years old 60%, 2–3 years old 35%, older ones 10%. (markmyprofessor uses a similar three-year weighting.)
- Few ratings are pulled toward the middle. Every presenter starts with 3 imaginary ratings of 3.5, so the score only moves far from 3.5 once real ratings pile up.
- Presenters need at least 3 ratings to appear in the Top list, conferences at least 5.
- In a conference schedule, the session in each time slot whose presenter has the best weighted score gets the ★ Audience pick badge.
What doesn't count
Ratings waiting for moderation or hidden by a moderator are left out of every average. Each browser can rate a talk once; rating it again updates the earlier rating.
Are the numbers reliable?
They show what the people who chose to rate thought, nothing more. A handful of ratings is a hint, not a verdict. Read the comments, and look at how many ratings sit behind a score.