Send The Video Instead

I recorded a two minute video for a prospect last month. Personalized, their name in the first sentence, a walkthrough of the one thing I'd fix on their landing page.

They replied in eleven minutes. The text email version of the same message, sent to a similar prospect the week before, got nothing.

That's not a controlled study, and I'm not going to pretend it is. But anyone who's sent personal video in sales knows the pattern. Video gets opened, watched and answered at rates text doesn't come close to. The problem was never whether video works. The problem is that recording one takes fifteen minutes by the time you've set up the light, done three takes and fixed your hair, and you can't do that forty times a week.

This week's tool is HeyGen, and it's the closest thing I've found to solving that problem without making you look like a hostage reading a statement.

HeyGen turns a script into a video of you saying it. You record a short training clip once, it builds an avatar of you with your face and your voice, and from then on you type what you want to say and it produces the video.

It also does a few adjacent things that turn out to be just as useful. It translates videos into other languages while matching your lip movement to the new language. It can generate videos from templates for things like product walkthroughs and training. And it supports interactive avatars that can hold a conversation, though that part is less mature than the core product.

I use it through Galaxy.ai, which bundles it with a pile of other models under one login. If you're only going to use HeyGen, a direct account works fine. If you're already paying for three or four AI tools separately, it's worth doing the math on the bundle.

Here's what nobody tells you. The quality of every video you'll ever make depends almost entirely on the training clip you record on day one. People rush it, get a mediocre avatar, and conclude the tool doesn't work.

Spend thirty minutes on it. Here's what matters.

Light your face evenly. A window in front of you, not behind you. If you've got a cheap ring light or a softbox, use it. Shadows on half your face get baked into the avatar forever.

Frame it the way you'll use it. If most of your videos will be head and shoulders, record head and shoulders. The avatar inherits the framing.

Talk like you actually talk. Don't read in a presentation voice. Talk the way you'd talk to a client across a table. The avatar picks up your rhythm and your expressions from this clip, so give it your real ones.

Keep your hands mostly still. Big hand gestures can produce odd results. Small natural movement is fine.

Plain background. Busy backgrounds make the edges of the avatar less clean. You can always put a better background behind it later.

Record two or three versions if you can. It's a one time cost and it determines how every future video looks.

I've been testing this across a few different use cases. Here's where it's worth the money and where it isn't.

Sales follow up at volume. This is the headline use case and it's real. After a discovery call, a short video recap with the prospect's name, what they said their problem was, and the next step. You write the script in two minutes, and a lot of the drafting can be automated. If you record your calls with Fathom, you've already got the transcript. A model can draft a sixty second recap script from it, you edit it for thirty seconds, and HeyGen renders it.

Onboarding and training. Every business has the same ten explanations it gives every new client or every new hire. How to log in. How to submit a request. What happens in week one. Record each one as an avatar video, and when something changes, you edit the script and re render instead of reshooting. That last part is the real win. Traditional training videos go stale and nobody updates them because reshooting is a pain. Script based videos get updated because it takes five minutes.

Translation. If you have any customers who don't speak English as a first language, this is the feature that surprised me most. Take a video you already made, translate it, and the result is you speaking fluent Spanish or Portuguese or Japanese with your mouth matching the words. It's not perfect, and a native speaker will notice some things. But for an explainer or a welcome video it's dramatically better than subtitles, and it opens markets that were completely closed to a one person operation.

Content at scale. Short explainer videos for social, turning newsletter content into short clips, weekly updates. This works, but with a big caveat I'll come back to in a minute.

FROM THE AI NEWSROOM

The exact systems behind everything in this issue. The audit sheets, the routing logic, the templates and the review cadences, built out step by step so you can copy them straight into your own stack. One time, forty seven dollars.

Get the Blueprint for $47

.....

Here's the build for the sales follow up use case, because that's the one with the clearest return.

After every discovery call, the transcript lands in my notes automatically. A Make.com scenario picks it up and sends it to a model with this prompt:

The script lands in a review queue. I read it, fix anything that's off, and approve it. Then it goes to HeyGen for rendering, and the finished video link drops into a draft email ready for me to send.

Total time on my end per prospect: about two minutes. That includes reading the script and hitting send.

The important thing is that approval step. The script is never sent without me reading it. More on why in a second.

Here's the part of this review that matters most, and it's not a feature.

Sending someone a video of your avatar without telling them it's an avatar is a trust problem waiting to happen. People are getting better at noticing, and the moment a prospect realizes the "personal video" wasn't recorded by you, the personal part of the message backfires. It reads as a trick.

My rule: I say so. A short line in the email, something like "quick recap video below, made with my AI avatar so I can get these out the same day." Nobody has ever been bothered by that. Several people have asked how I set it up, which turned into its own conversations.

The content is still yours. You wrote the script, or at least edited and approved every word. The recap reflects a real conversation you had. The only thing automated is the camera. Being upfront about that keeps all the upside and none of the downside.

This matters more than it did a year ago. Meta is reportedly preparing to put its Muse Video model on its APIs, and there's a lot more synthetic video coming at everyone from every direction. The people who stay trusted in that environment will be the ones who are clear about what's real.

Being straight about the limits.

It's not great for emotional or high stakes messages. Delivering bad news, apologizing to a client, anything where your actual face and actual tone matter. The avatar is good, but it's a little too consistent. Real people get a bit uncomfortable when things are hard, and that discomfort is part of the message. Record those yourself.

Long videos get uncanny. At sixty to ninety seconds, most people won't notice anything off. At eight minutes, small repetitive patterns in expression start to show. Keep avatar videos short.

Social content at scale is a trap. Yes, you can produce thirty short videos a week. So can every one of your competitors. Your feed filling up with polished avatar clips doesn't make you stand out, it makes you blend in. Use it for the things that genuinely benefit from video and keep your face on camera for the rest.

Pronunciation needs checking. Company names, industry jargon, unusual names. Listen to the first render of any script with proper nouns in it. Most tools let you fix pronunciation by spelling words phonetically in the script.

Buy it if you do real volume in sales conversations, and follow up is where deals stall. The time savings on recap videos alone justifies the cost for most people doing more than five discovery calls a week.

Buy it if you have training or onboarding content that goes stale because nobody wants to reshoot it.

Buy it if you have, or want, customers who speak another language.

Skip it if your sales happen mostly in person or on the phone and video follow up doesn't fit how your customers buy. Skip it if you only need a video once a month. Record that one yourself.

And skip it if the only reason you're considering it is to pump out more social content. You don't have a volume problem there. Almost nobody does.

Don't take my word for it. Run the numbers on your own pipeline.

For thirty days, split your follow ups. Half get your normal text email. Half get a sixty second avatar recap video with an honest line about how it was made. Keep everything else the same.

Track three things: reply rate, meetings booked from the follow up, and deals that moved forward.

If video wins by a meaningful margin, and in my experience it usually does for follow ups, you've got your answer and the math on the subscription is easy. If it doesn't, you've saved yourself a monthly bill and learned something about your buyers that's worth knowing.

FROM THE AI NEWSROOM

For operators who would rather build it with someone than build it alone. The full stack, installed alongside you, with the decisions made in the room instead of guessed at. Ninety seven dollars.

Join the Accelerator for $97

.....

Before you sign up for anything, try the manual version.

Pick your next three follow ups after a sales call. Record a sixty second video on your phone for each one. Name, their problem in their words, one next step.

See what happens to the reply rate.

If it jumps, and the only reason you won't keep doing it is the time, that's exactly the problem HeyGen solves. If it doesn't jump, video isn't your bottleneck and you just saved yourself a subscription.

Jordan

The AI Newsroom | Practical AI for people with a business to run.