COURSE & E-LEARNING BLOG / 9 MIN READ

How do I improve my course and e-learning videos?

Students sit with your lesson for a long time, so the things that wear them out are sound, readable text and pacing, not camera megapixels. Here is what actually moves the needle, and how to check it before a module goes live.

−14 LUFSvoice loudness target
1hr+typical watch time
~50%watch on mute
0–100craft score

By Thomas, founder of CutScore · Updated June 2026

LESSON CHECK · module_03.mp4
An instructor recording an online lesson at a desk, the kind of course video whose quality is decided by clear audio and readable on-screen text rather than the camera.
CRAFT SCORE
FIXES ADVISED
a 24-minute lesson, checked
Voice loudness on target · −14 LUFS
On-screen code too small · bump font size06:42
Long pause, dead air · trim 4s13:20
The 30-second answer To improve your course and e-learning videos, fix the things a student feels over a long watch, not the camera. Get the voice clean and consistent at around −14 LUFS with peaks under −1 dBTP, so module five is not louder than module one. Make every slide, caption and line of code large and high-contrast inside the safe zone, because around half of viewers watch on mute. Light your face evenly, cut the dead air and slow demos, and export at a high enough bitrate that text stays sharp after upload. Sound and readable text decide whether people finish the lesson. If checking that across a whole course sounds tedious, that is the exact job CutScore does in one pass.
WHY COURSE VIDEO IS DIFFERENT

A course video is not a thirty-second hook. Nobody watches a lesson for the vibes. They are here to learn a thing, and they will sit with you for ten, twenty, sometimes sixty minutes to get it. That changes the rules. Small problems you would never notice in a Short become genuinely exhausting over an hour. A faint hiss, a voice that creeps up and down in volume between clips, code you have to squint at: none of it is fatal once, and all of it is brutal repeated for the length of a module.

I have made e-learning videos that I was proud of and that students quietly bailed on, and it took me far too long to admit why. The teaching was fine. The experience of watching it was tiring. The audio wandered, the screen text was too small on a laptop, and a five-minute setup that bored me to record bored them twice as much to watch. None of that showed up when I previewed thirty seconds on my big monitor at full brightness with good headphones.

So the goal here is not "make it cinematic." The goal is make it easy to sit through and easy to read, module after module, on whatever device the student happens to be holding. That comes down to four things, and they are not the four most beginners worry about.

THE FOUR THAT MATTER

What actually improves a course video.

Ranked by how much your students feel them. Fix from the top down, because the first two carry most of the perceived quality of any lesson.

FixTarget to hitWhat it costs you if you skip it
Voice loudness≈ −14 LUFSQuiet, timid audio over an hour is exhausting and reads as low effort.
Loudness consistencysame across modulesRiding the volume between lessons makes students reach for the dial.
True peak≤ −1 dBTPHot peaks crackle on cheap laptop speakers, which is most of your class.
Background noiselow, steadyHiss and room hum become unbearable over a long, focused watch.
On-screen text + codelarge, high-contrastTiny code is the number one complaint on coding and software courses.
Captions + safe zonesreadable, in-frameAround half watch on mute, and many are non-native speakers.
Exposure + white balanceeven, neutralA dark or colour-cast talking head looks unfinished beside clean slides.
Pacing · dead airtrim the gapsReal-time demos and silent clicking are where students quit a good lesson.
Export bitratehigh enough for textLow bitrate smears slides and code into mush after the platform re-encodes.
The webcam is not on this listNotice what is missing: camera resolution. For most courses the slides, the screen recording and your voice teach the lesson. A clean webcam in good light is plenty. Spend on a microphone and your screen-capture setup long before you spend on a camera.
CHECK EVERY MODULE, NOT JUST ONE

A course is twenty videos, not one. Drift in loudness or text size across a dozen modules is the part that is impossible to catch by ear. CutScore checks each lesson against the same standard and flags the ones that fell off.

Join the waitlist
HOW TO ACTUALLY FIX EACH ONE

Four passes, in order of impact.

1. Sound: the thing that decides whether they finish

Over a long lesson, audio is everything. Get the voice sitting near −14 LUFS so it is comfortable without the student reaching for the volume, and keep true peak at or below −1 dBTP so nothing crackles on laptop speakers. Then chase the two things that wreck a course specifically: consistency and noise. Module one and module twelve should be the same loudness, recorded weeks apart or not. A faint hiss you can ignore for thirty seconds becomes torture across an hour, so clean it up. If your music bed is louder than your voice during a demo, you have already lost half the room.

A hand adjusting faders on an audio console, the part of a course recording that students feel most over a long lesson when the voice drifts in loudness or sits under hiss.
Consistent, clean voice across every module beats any camera upgrade. Photo: Tima Miroshnichenko / Pexels.

2. On-screen text and code: make it readable on a phone

This is where course videos quietly fail, especially software and coding lessons. You record at a comfortable size on a 27-inch monitor, then a student opens it on a phone and the code is a grey smudge. The fix is unglamorous: bump up the font in your editor, your terminal and your slides before you record, not after. Zoom into the part you are talking about. Keep captions and key text inside the platform safe zone so the interface does not eat them. The test is simple, and I do it on every lesson now: play it on your phone at arm's length. If you squint, your students are squinting harder.

3. Picture: light the face, neutralise the colour

Your talking-head insert does not need to be cinematic, but it should not look like a hostage video next to a crisp slide deck. Get even light on your face, no harsh shadow on one side, and make sure your white balance is neutral so you are not faintly orange or blue. A clean, well-lit webcam reads as professional. A dark, colour-cast one reads as "recorded in a hurry," and it drags down the slides sitting right next to it. This whole family of image checks is part of what we analyze, because the picture is the first thing a student reads before you have said a word.

4. Pacing and export: respect the student's time, then test it

A good lesson delivered slowly still loses people. Watch your own video and be honest about the dead air: the silent clicking, the "let me just find that," the demo running in real time when a cut would do. Trim it. Speed up routine setup. A few well-placed jump cuts remove the slog without reshooting anything, and counting your filler words will tell you whether the delivery is tightening or wandering. Then the boring last step: export at a bitrate high enough that small text survives compression, upload, and watch the published version on a phone. If the code went soft, your export settings are the suspect, not your screen recorder.

RATHER SEE IT THAN READ IT?

Here is a real CutScore coaching report: loudness, on-screen text, pacing and export, all scored, with timestamps and the exact fixes for the lesson.

See a sample report
SHORT ON TIME

If you only fix three things.

Most of the jump from "homemade lesson" to "this instructor knows what they are doing" comes from these three. Fix them across the whole course, not just the intro.

1
2-MIN FIXAUDIO
Normalise every module to about −14 LUFS
Quiet, uneven audio is the fastest way to tire a student out, and it has nothing to do with your microphone. Normalise each lesson toward −14 LUFS with a true peak under −1 dBTP, so module one and module twelve feel identical and nobody touches the volume.
How Run a loudness meter over each export, or let CutScore measure every module and flag the ones that drifted.
2
SETUPTEXT
Make every line of text and code phone-readable
Increase your editor, terminal and slide font sizes before you record, and zoom into whatever you are explaining. Tiny code is the single most common complaint on technical courses, and it is the easiest one to prevent entirely.
How Play the lesson on your phone at arm's length. If you squint, the font is too small.
3
EDITPACING
Cut the dead air and the real-time waiting
Trim the silent clicking, the "let me find that," and the demos that run in real time. Students forgive a webcam, but they will not sit through five minutes of you waiting for something to install. A cut respects their time and keeps the substance moving.
How Watch it back and remove anything you would skip if it were not yours. See pacing and shot length.
THREE WAYS TO CHECK A LESSON

By eye, by meter, or in one pass.

OPTION 01

By eye and ear

Free, and a real start. Watch the whole lesson on a phone with cheap earbuds, not your studio setup. The catch over a long course: your ear adapts to your own voice, and you cannot reliably feel a two-decibel drift between module three and module nine. Best used on yesterday's recording, not today's.

OPTION 02

With scopes and meters

Accurate and honest. A loudness meter on every export, a scope for exposure, a hard look at text size. The cost is time: you have to run them on each module of a long course and read them correctly every single time. Great if you like the process. Across twenty lessons, most instructors quietly stop.

OPTION 03

With a coach in one pass

Hand each lesson (or a link) to CutScore. It measures loudness, consistency, on-screen text, pacing and export against the same standard for every module, then gives you a 0 to 100 score with timestamped evidence and the fixes. See a sample report.

How CutScore checks a whole course CutScore is an AI video quality coach for the craft of a video. For a course, that means it measures the things that wear a student down: loudness near −14 LUFS, peaks, noise, on-screen text legibility, exposure, pacing and the export, computed deterministically where it can be (loudness with an EBU R128 meter, for instance) and reserving AI for the genuinely subjective parts. Run every module through it and you get one score per lesson, the evidence behind it, and a prioritised list of fixes, so module twelve is as clean as module one. It judges the craft of the lesson itself, not your course's marketing. More on the method and the standards.
QUESTIONS

Frequently asked.

Fix the things students actually feel: clean, consistent audio near −14 LUFS with peaks under −1 dBTP, even and neutral exposure so the slides and your face are easy to read, on-screen text and code that is large and high-contrast inside the safe zone, pacing that does not crawl, and an export that survives upload. Sound and readable text matter more than camera resolution in a lesson someone watches for an hour.
Audio, followed closely by readable on-screen text. A student will quit a great lesson with tiring, uneven sound long before they quit a clear lesson shot on a webcam. Get the voice clean and consistent across every module, then make sure every word and line of code on screen can be read on a phone.
Usually pacing and dead air, not the material. Long unbroken takes, slow demos in real time, and silence while you click around drain attention even when the teaching is solid. Trim the gaps, cut the throat-clears, and speed up routine setup so the substance arrives faster. Good information delivered slowly still loses people.
No. For most courses the screen recording, the slides and your voice carry the lesson, not the camera. A clean microphone, even lighting on your face, and crisp, readable text beat an expensive camera every time. Spend on sound and a good screen-capture setup before you spend on a lens.
EARLY ACCESS

Make every module as clean as the first.

CutScore checks each lesson against the same standard and tells you exactly what to fix, with the evidence to back it up. Join the waitlist for early access.

Join the waitlist