Eye contact is doing more work in your sales videos than your script
Video prospecting has become standard because it works — a face is harder to ignore than a paragraph. But the same thing that makes video effective makes it fragile: the prospect can see exactly how much attention you're paying them.
The case for video outreach
The reported numbers are strong, though they come from vendors and their customers rather than independent research, so treat them as directional rather than precise. Intercom reported cold email reply rates rising 19% after adding Loom to outreach. Qwilr reported conversion increases of up to 57% using video. Teams running personalised video commonly report reply rates several times higher than text-only sequences.
The mechanism isn't mysterious. Video proves a human spent time on this specific prospect, which is exactly the signal cold outreach usually fails to send.
Why gaze matters more here than anywhere else
The entire value of a personalised video is the implication that you're present and paying attention. Gaze is the fastest way to undermine that — a prospect watching your eyes track across a script is watching evidence that this is a template with their name pasted in.
In content marketing, a viewer who is already interested will forgive a lot. In cold outreach, you have a few seconds and no goodwill, so small credibility signals carry disproportionate weight.
The structural problem with personalised video
Sales videos are the worst case for eye contact by design. You're reading the prospect's name, their company, a detail from their LinkedIn — all from notes on a second screen — while recording thirty of them in a sitting.
The advice to 'just memorise it' doesn't survive volume. By video twenty, you're reading, and it shows.
Practical fixes, in order of effort
- Put the research notes directly under your webcam, not on a second monitor off to the side.
- Memorise only the first line — the opening is where attention is highest and drift is most damaging.
- Use a short fixed structure so you only need to look up for the personalised details.
- Record in batches with the same setup, so a single geometry fix improves every video.
- Correct gaze in post rather than re-recording — re-shooting is what makes personalised video stop scaling.
Fixing it at volume
The reason sales teams abandon video isn't that it doesn't work — it's that quality collapses as volume rises, and re-recording weak takes costs more time than the channel is worth. AI eye contact correction changes that trade-off: you record for speed and content, and fix the gaze afterwards.
GazeFix is built for exactly this shape of usage — pay per minute rather than per seat, so cost tracks the volume you actually send. It's launching soon; join the waitlist for early access and free launch credits.
What to actually measure
If you change something here, measure reply rate rather than open or view rate. Views tell you the subject line worked; replies tell you the video did. And test on a meaningful sample — reply rates on small batches are noisy enough to show improvements that aren't real.