# Should You Show Your Face in Software Tutorials? What Studies Found

> Studies of instructor presence found higher motivation, no reliable learning gain and an attention cost. Get a decision table, layout options and a lighting checklist.

- Canonical page: https://learnhubplay.nl/blog/talking-head-vs-screen-only
- Published: 2026-10-15
- By: LearnHubPlay, an independent software studio in Amsterdam
- Topic: Screen Recording & Tutorials
- Tags: Video production, Screen recording, Research explainers, Checklists

Every screencast maker hits the same fork before recording: a webcam bubble in the corner, a few seconds of face at the start, or just the cursor and your voice. The advice is usually conviction ("people connect with faces", "faces are clutter") rather than evidence.

The evidence exists and is more useful than either slogan. Across lab, field and meta-analytic studies, a visible instructor tends to make viewers like a video more, does not reliably make them remember more, and takes some of their attention. Below are the main studies with sample sizes and limits, what they cannot tell you about software demos, and a decision table, layout options and a camera and lighting checklist.

## What three studies measured and what they found

| Study | Design and sample | Measured | Result for a visible face | Limits |
|---|---|---|---|---|
| Guo, Kim & Rubin, 2014 | Retrospective; 6.9 million sessions, 862 videos, four edX courses | Watch time; next-problem attempts | Higher engagement for videos cutting to the instructor's head, in both courses that had them; in one, 46% attempted the next problem versus 33% | Not an experiment; STEM courses, Fall 2012; the face comparison covers two of the four |
| Kizilcec, Papadopoulos & Sritanyaratana, 2014 | Lab; 22 students; one 14-minute lecture; face on or off by segment; eye tracking | Recall at 3 minutes and 5 days; gaze; ratings | No recall difference; 41% of looking time on the face; 15 of 22 strongly preferred it | 22 people; tested and tracked; no pausing or notes |
| Kizilcec, Bailenson & Gomez, 2015 | Observational (2,951 learners, 8 weeks) and field experiment (12,468, 10 weeks) | Experience, cognitive load, learning, dropout | Face viewers rated it better; 35% chose to watch without it; constant versus occasional face gave the same learning and dropout | Observed groups chose for themselves; lectures, not demos |

### Reading the edX numbers with care

The face was an editing choice, not an assignment: producers spliced it in whenever the timing "felt right", in 60% of one course's videos and 25% of another's, so those videos may differ in other ways. Watch time cannot separate attention from a video left playing, as the authors note. The informal, desk-filmed course also out-engaged a studio-filmed one, but that is one pair of courses, so instructor skill could explain it.

### What the lab and field studies add

The lab study controlled what logs cannot: the same lecture, face present for some segments and absent for others, in randomly assigned order. Its power analysis says only medium-to-large effects were detectable with 22 people. Participants knew they would be quizzed and eye-tracked, and could not pause or take notes, which the authors say may have raised cognitive load.

The 2015 field experiment found that showing the face constantly or only strategically gave the same learning and dropout, while the strategic version raised cognitive load and perceived social presence. The authors read this as consistent with Mayer's image principle, which they summarize as: showing the face does not promote learning, because social benefits are outweighed by extra processing.

## What the reviews and meta-analyses conclude, and where they disagree

| Review | Studies | Learning | Feelings and motivation | Attention and load |
|---|---|---|---|---|
| Henderson & Schroeder, 2021 (systematic review) | 12 | "No consistent, compelling evidence" for including the instructor, and none for leaving them out | Some studies: learners more satisfied | Not in the abstract |
| Alemdag, 2022 (meta-analysis) | 20 experiments | No significant effect on learning or social presence | Motivation significantly up | Cognitive load significantly up |
| Beege and colleagues, 2023 (meta-analyses) | 35 studies, 6,339 participants | Small retention gain; no significant transfer effect | Social presence, affect, motivation up | Less dwell time on relevant visuals; lower rated extraneous load |

Two things hold across the rows: motivation and affect rise, and no review finds a learning gain large or consistent enough to bank on. The meta-analyses split on learning, social presence and cognitive load; the abstracts do not say why, though they pooled different studies. Polat's 2022 review of 41 studies landed similarly: positive feelings, no reliable learning gain from presence alone, inconclusive social effects. And "no learning effect" is not "harm": Henderson and Schroeder found no convincing reason to leave the instructor out.

## Why software demos are a harder case for a face

Treat the above as evidence about lectures and worked-problem videos, applied to software by analogy. The edX courses were computer science, statistics, AI and chemistry, the lab study used a sociology lecture, and Polat names picture-in-picture over lecture slides as the most common format in the research it reviewed. Three differences matter for screencasts.

- **The screen is the lesson.** Viewers read labels and track a cursor. In the lab study a small corner video took 41% of looking time, with a glance between face and slide about every 3.7 seconds. Beege's pooled results include less time on the relevant visuals.
- **Tutorial viewers watch briefly and return.** In Guo's data, which labelled problem-solving walkthroughs as tutorials, they watched 2 to 3 minutes whatever the video's length and re-watched more than lecture viewers. Someone scrubbing to step 4 gets less from a face than a lecture viewer sitting through 10 minutes (our reasoning, not a measured result).
- **You are looking at the screen.** While you demonstrate, an overlay shows you glancing away from the viewer, whereas many instructor-present videos Polat reviewed used direct gaze as a social cue.

## A decision table by video type

This is our synthesis of the evidence above, not a result any study tested. The logic: a face buys motivation and a sense of a person, costs attention and screen area, and has not reliably bought learning.

| Video type | Default | Why | Change it when |
|---|---|---|---|
| One task, 1 to 4 minutes | No face, or a few seconds at the start | The viewer wants the result and scans for a step | You publish a series and viewers return for you |
| Reference video people scrub (flags, settings) | No face | Viewers jump straight to a step; chapters do the navigating | Rarely |
| Dense interface: small text, terminal plus editor | No face, or intro and outro only | Screen area and attention are scarce | Some screen outside the interface is unused |
| Concept before the demo | Face full-frame or corner overlay | Nothing small to read; a person explaining carries well | The concept needs a diagram: show the diagram |
| Lessons in a course or series | Face at intro, outro and topic changes | Motivation is the benefit both meta-analyses found | Learners use it as reference |
| Opinion, review or comparison | Corner overlay, or alternate face and screen | Your judgment is the product | Click-by-click sections: screen only |

Three questions break ties. Must the viewer read small things on screen? Do viewers come for you or for the task? Can you do a face segment well (lit, looking at the lens, relaxed)? A poor face segment costs attention and buys nothing.

### A worked example

Say you are recording a 3-minute "Set up a pre-commit hook" video, with a terminal and editor in small text. The dense-interface row applies: no face during the demo. You publish weekly, so you add 6 seconds of face to open and 5 to point at the next task, about 11 of 180 seconds on camera. The [beat-sheet approach in the software tutorial guide](https://learnhubplay.nl/blog/record-software-tutorials) works with either choice.

## Layouts and what each one costs

| Layout | Cost to attention | Effort | Fits |
|---|---|---|---|
| Voice-over only | None from a face; no social cue | Lowest | One-task and reference videos |
| Face at intro and outro | None during the demo | Low: two short clips | Series, personal channels |
| Corner overlay throughout | Constant motion beside the content | Low if captured with the screen; hard to change later | Concept talks, opinion |
| Corner overlay at transitions | Motion at key moments only | Medium: needs an edit | The "strategic" pattern in the 2015 experiment |
| Side band for the face | Nothing covered; interface smaller | Medium | Wide displays where text still reads |
| Full-frame cut-ins between screen segments | One thing at a time | Highest: a cut per segment | The cut-between pattern in the edX data |

For any overlay: put it over the least important part of the screen, usually a margin or title bar, in the same corner all video, and start small; never cover a control or text the viewer must read; and check it at the size most people watch, such as a phone. Captions normally sit along the bottom edge, so keep the overlay clear of them, as in the [captions and chapters guide](https://learnhubplay.nl/blog/youtube-chapters-captions-descriptions).

If your recorder can keep the camera as its own track or file, you can drop, shrink or move it in the edit. TechCast Studio, our recorder, has optional camera capture alongside screen and microphone; it is in Founding Beta on Apple Silicon and its public desktop download is not open yet. The method works with any recorder.

## Camera, lighting and a pre-record checklist

Camera height, framing, front lighting and the test clip follow the University of Illinois guide cited below; the rest is ordinary reasoning.

- [ ] **Layout chosen.** You picked a row from the tables and wrote it at the top of your beat sheet.
- [ ] **Camera at eye level**, not too low or too high; a laptop on a stack of books works.
- [ ] **Head and shoulders framed**, centered, with a little space above your head.
- [ ] **Light in front.** A soft light slightly above eye level and aimed at your face, or a window in front or to the side. No bright light or window behind you, no harsh shadows, no overhead-only light. Screen glow alone lights you from below, so do not rely on it.
- [ ] **Lens versus screen decided.** Look at the lens when addressing the viewer and at the screen when demonstrating; keep the camera near where your eyes go most.
- [ ] **Overlay clear.** It covers no control and no readable text.
- [ ] **Background checked.** Tidy, with no names, numbers or private items readable. The [privacy checklist](https://learnhubplay.nl/blog/screen-recording-privacy-checklist) covers what a camera adds.
- [ ] **Voice first.** Every layout includes your voice, so the [microphone setup guide](https://learnhubplay.nl/blog/budget-microphone-audio-setup) matters whichever you pick.
- [ ] **Test clip.** Record 10 to 15 seconds and check framing, light and levels at the size viewers will watch.

## Where to start this week

Pick your next tutorial, find its row in the decision table, and record the camera alongside the screen even if you will show it for seconds. Cut it the way the table says and write one line about why.

If you publish a run of similar videos, try different layouts on two and compare retention. Small channels give noisy numbers, so treat only a pattern that repeats as a hint. More on planning, recording and publishing is in the [tutorials topic hub](https://learnhubplay.nl/topics/tutorials).

## Frequently asked questions

### Do viewers learn more if they can see the instructor's face?

Not reliably. A lab experiment with 22 students found no recall difference, and a 2022 meta-analysis of 20 experiments found no significant learning effect. A 2023 meta-analysis of 35 studies found a small retention gain and no significant transfer effect. The consistent benefit is motivation and how viewers feel about the video.

### Is a webcam overlay distracting in a screen recording?

For some viewers. In the lab study people spent 41% of their looking time on a small face while it was shown, and in the field study 35% of learners who could choose watched without it, some citing distraction. Neither tested software demos, so keep it small, off the interface and brief.

### Where should I put the webcam overlay on a screen recording?

Use the corner or margin the application uses least, and keep it there for the whole video. It must not cover a control, menu or text the viewer needs, and with captions on it should stay off the bottom edge. Check it at phone size before publishing.

### What if I am uncomfortable on camera?

The research gives no reason to force it. A review of 12 studies found no consistent evidence that an on-screen instructor is needed, and the benefits found are mostly motivational. Spend the effort on clear audio, one task per video and a readable screen.

## Sources and further reading

- [How Video Production Affects Student Engagement: An Empirical Study of MOOC Videos](https://www.cs.rochester.edu/hci/pubs/pdfs/edX-MOOC-video-production-and-engagement_LAS-2014.pdf) - Guo, Kim & Rubin, Learning at Scale, 2014. The 6.9 million sessions, face and problem-attempt figures, limits.
- [Showing Face in Video Instruction: Effects on Information Retention, Visual Attention, and Affect](https://rene.kizilcec.com/wp-content/uploads/2014/01/final_version2.pdf) - Kizilcec, Papadopoulos & Sritanyaratana, CHI, 2014. The 22-person experiment, 41% looking time, recall result.
- [The Instructor's Face in Video Instruction: Evidence From Two Large-Scale Field Studies](https://vhil.stanford.edu/publications/learning/instructors-face-video-instruction-evidence-two-large-scale-field-studies) - Kizilcec, Bailenson & Gomez, Journal of Educational Psychology, 2015. The two field studies, the 35% figure.
- [Effects of instructor-present videos on learning, cognitive load, motivation, and social presence: A meta-analysis](https://doi.org/10.1007/s10639-022-11154-w) - Alemdag, Education and Information Technologies, 2022. Pooled results on learning, load, motivation.
- [The instructor presence effect and its moderators in instructional video: A series of meta-analyses](https://doi.org/10.1016/j.edurev.2023.100564) - Beege, Schroeder, Heidig, Rey & Schneider, Educational Research Review, 2023. The 35-study results on retention, transfer, affect, dwell time.
- [A Systematic Review of Instructor Presence in Instructional Videos: Effects on Learning and Affect](https://doi.org/10.1016/j.caeo.2021.100059) - Henderson & Schroeder, Computers and Education Open, 2021. The 12-study review, its either-way conclusion.
- [Instructors' presence in instructional videos: A systematic review](https://pmc.ncbi.nlm.nih.gov/articles/PMC9786415/) - Polat, Education and Information Technologies, 2022. The 41-study summary, the picture-in-picture trend.
- [Home Recording: Hardware Setup & Best Practices Guide for Faculty](https://answers.illinois.edu/161036) - University of Illinois, 2026. Camera height, framing, front lighting, test clip.