Build a Worked-Example Index From YouTube STEM Lectures Before the Final

Published 2026-09-30 · Francesco Masciopinto

A final does not grade how many lecture hours you "almost finished." It grades whether you can recognize a problem type, pick a method, and avoid the trap you keep falling for. That skill lives in worked examples — the board problems, demos, and "let's do one together" stretches buried inside your YouTube STEM playlist. Most students either rewatch whole videos or copy every equation onto a formula sheet. Neither builds a searchable map of examples you can drill.

This article is a one-sitting workflow for a worked-example index: a short table (or note page) that turns a stack of lecture videos into problem-type → method → trap → timestamp rows. Soft on-page study tools can speed the hunt for example segments; the classification and closed-book drills are still yours.

What a worked-example index is (and is not)

An index row answers four questions:

What kind of problem is this? (name it the way the exam will — "related rates with similar triangles," not "the tank problem from Lecture 6")
What method did the instructor use? (one line: setup → key equation → finish)
Where does it usually break? (the trap, the illegal step, the assumption that fails)
Where is the footage? (lecture label + start timestamp)

It is not:

A formula dump (symbols without problem types)
An Anki deck rebuild from scratch (that is a semester habit)
A night-before triage of which videos to abandon (useful, different job)
A closed-book quiz stem list (stems test you; the index feeds stems)

If your sheet only has LaTeX and no "when I see X, I start with Y," you built a collage. The index forces the when.

When to build it

Best window: finals week minus 3–7 days, after you know the exam scope and before panic rewatch. You need:

The syllabus / final topics list
The playlist (or folder) of lecture uploads that cover that scope
60–90 minutes for Pass 1 harvest + 30 minutes for Pass 2 drill

If you have three days and twenty hours of video, you still index — you just harvest fewer examples per topic and skip admin-heavy uploads.

Pass 0 — Charter the index (5 minutes)

Write at the top of a page or doc:

Final covers: [unit range]. Index goal: recognize and start 8–12 problem types without opening the player.

Then list exam topics as the instructor named them (not as YouTube titles). Leave two blank columns for later: example count and confidence 1–5.

Hard rule: if a video is outside the final scope, it does not get a row — even if the thumbnail looks friendly.

Pass 1 — Harvest examples, not transcripts (45–60 minutes)

Work topic by topic, not video by video from Lecture 1. For each exam topic:

Open the 1–2 lectures most likely to contain worked problems.
Jump to example segments. Use chapters, transcript search ("example," "suppose," "let's try"), or an on-page chapter/summary panel to land faster — then watch with a pencil.
For each example you keep, fill one row:

Problem type (exam language)	Method (1–2 lines)	Trap / fail condition	Source (lecture + time)
…	…	…	…

Harvest rules that keep the sheet short

Cap at 2–3 examples per exam topic. A fourth example that is the same method with different numbers is noise. Prefer a second example only when the setup changes (different given, different illegal move).
Name the type before you love the numbers. If you cannot rename "the ladder problem" into a type the exam might use, you do not understand the example yet — rewatch that span once, then name it.
Method is a skeleton, not a proof. "Identify similar triangles → write proportion → solve for dx/dt" beats copying six lines of algebra.
Trap is mandatory. If you leave Trap blank, you will re-fail the same way on the final. Common traps: wrong similar-triangle pair, forgetting a chain-rule factor, applying a theorem outside its hypothesis, mixing units, treating a discrete sum like a continuous integral.
Skip pure theory stretches unless the instructor immediately applies them. Theory without a worked instance goes on a separate claim list — not this index.

Optional speed: generate a first-pass chapter list for a long upload so you can jump to "Example 2" and "Application" segments without scrubbing. Close the panel before you write the row so the wording is yours.

Budget roughly 8–12 minutes per exam topic. If a topic has no worked example in the playlist, mark it needs practice set and move on — do not invent an example from memory in Pass 1.

Pass 2 — Drill from the index with the player closed (25–35 minutes)

Cover the Method and Trap columns (fold the page or hide columns). For each Problem type row:

Say out loud how you would start (first two moves only).
Name the trap you expect.
Uncover and score: green / yellow / red.

Yellow and red rows get a targeted rewatch of that timestamp only — not the whole lecture. After the jump-in, rewrite the Method line in your own words. If you still cannot start closed-book, pull one homework problem of the same type and solve it with the index face-down.

Stop when you have greened the high-weight topics. A half-green index you drilled beats a perfect index you never tested.

How many rows is enough?

Final weight	Target rows	Notes
Core unit (likely multi-question)	2–3	Different setups if possible
Supporting unit	1–2	One solid example + trap
Cameo / low signal	0–1	Skip if time is tight
No worked example in videos	0	Flag for practice set / office hours

A usable finals index for a typical STEM final is often 12–20 rows, not 50. Density without recognition practice is decoration.

Using on-page AI without outsourcing the index

Soft tools on the YouTube watch page can draft chapters, summaries, or study prompts so Pass 1 finds example segments faster. Use them as a fog map: candidate timestamps, rough section titles, maybe a sketch of "what was covered." Then close the panel and fill the four columns yourself. The final grades your ability to start a problem type, not a model's paraphrase of the lecture.

For a free on-page study panel on YouTube (summary, chapters, and study prompts), see https://summarizai.ink. Start there if a two-hour upload has no human chapters; the index table above still owns naming, traps, and closed-book drills.

Common failure modes

Indexing every board scribble. If it is not a problem type the exam could ask, it is not a row.
Copying the entire solution into Method. You will reread instead of retrieve. Skeleton only.
Skipping Trap because "I'll remember." You will not. Write the fail condition while the mistake is still embarrassing.
Building the index in playlist order. Exam topics first; videos are sources, not the outline.
Never doing Pass 2. An unread index is a second transcript. Cover Method/Trap and speak the start.
Turning Pass 2 into full rewatch. Timestamp only. If the stamp is wrong, fix the stamp — do not restart at the intro.
Mixing formula-sheet duty into the same page without labels. Formulas can live in a sidebar; the index's job is problem-type recognition. Keep the jobs visually separate.

One-page checklist (copy this)

Charter: final scope + goal line at the top
List exam topics in instructor language
Per topic: jump to 1–3 worked examples only
Fill: problem type · method skeleton · trap · lecture+time
Cap rows; merge duplicates that share the same method
Cover Method/Trap; drill starts out loud
Yellow/red → timestamp rewatch → rewrite Method
Stop when high-weight topics are green — sleep or practice set next

Do this once before a STEM final and the playlist stops being a guilt stack of hours. It becomes a short, searchable map: which problem types you can start cold, which traps still own you, and exactly which minute of which lecture to reopen if a row goes red — without pretending that rewatching everything was ever the plan.

Frequently asked questions

What is a worked-example index?

An index row answers four questions: What kind of problem is this? What method did the instructor use? Where does it usually break? Where is the footage? It is not a formula dump, an Anki deck rebuild from scratch, a night-before triage of which videos to abandon, or a closed-book quiz stem list. If your sheet only has LaTeX and no "when I see X, I start with Y," you built a collage. The index forces the when.

When should I build a worked example index from YouTube lectures?

Best window: finals week minus 3–7 days, after you know the exam scope and before panic rewatch. You need the syllabus / final topics list, the playlist (or folder) of lecture uploads that cover that scope, and 60–90 minutes for Pass 1 harvest plus 30 minutes for Pass 2 drill. If you have three days and twenty hours of video, you still index — you just harvest fewer examples per topic and skip admin-heavy uploads.

How many examples should each exam topic get?

Cap at 2–3 examples per exam topic. A fourth example that is the same method with different numbers is noise. Prefer a second example only when the setup changes (different given, different illegal move). If a topic has no worked example in the playlist, mark it needs practice set and move on — do not invent an example from memory in Pass 1.

How many rows is enough for a STEM final?

A usable finals index for a typical STEM final is often 12–20 rows, not 50. Density without recognition practice is decoration. Core units (likely multi-question) get 2–3 rows with different setups if possible. Supporting units get 1–2. Cameo or low-signal topics get 0–1 and can be skipped if time is tight. Topics with no worked example in the videos get 0 rows and a flag for a practice set or office hours.

How do I drill from the index with the player closed?

Cover the Method and Trap columns (fold the page or hide columns). For each Problem type row, say out loud how you would start (first two moves only), name the trap you expect, then uncover and score green / yellow / red. Yellow and red rows get a targeted rewatch of that timestamp only — not the whole lecture. After the jump-in, rewrite the Method line in your own words. Stop when you have greened the high-weight topics. A half-green index you drilled beats a perfect index you never tested.

Can on-page AI write the worked-example index for me?

Soft tools on the YouTube watch page can draft chapters, summaries, or study prompts so Pass 1 finds example segments faster. Use them as a fog map: candidate timestamps, rough section titles, maybe a sketch of what was covered. Then close the panel and fill the four columns yourself. The final grades your ability to start a problem type, not a model's paraphrase of the lecture.

Related guides

Try SummarizAI on your next lecture

SummarizAI is a Chrome extension that adds a summary, chapters, and Study flashcards on the YouTube watch page. The free plan is a student trial—no need to leave the lecture tab.

Start the free student trial

Students · Responsible summaries · Install