By Niels Geominy, founder of RecruitmentAds.com
This is part 2 of a three-part series from RecruitmentAds.com, sponsor of EB EXP Fall 2026. Part 1 goes live Mon Nov 16, part 2 Wed Nov 18, part 3 Fri Nov 20.
Plenty of employers cannot film their people: cleared roles, regulated sites, night shifts, or nobody willing to be on camera. That is a production constraint, not a reason to skip video. It is a reason to change how you make it.
Here is the sequence, adapted from our eight-step guide, with one customer's published numbers at the end.
1. Start from the job posting
The most common way a recruitment video goes wrong is that it starts as a video. Start from an open role and its job posting. If the posting is thin, fix it first. A video cannot make an unclear job clear. It can only put the confusion in front of more people, faster.
Then write a one-screen brief with seven fields: the one job, who it is for, the one message, the single strongest fact, where it runs and in what cut, the action, and what the video will not say. The last field is what stops a stakeholder adding the benefits list in week two.
2. Write the cards first
In a feed, sound is off by default for most viewers, so the on-screen text is the script. Write six or seven cards, one idea each. The role goes on screen in the first half second and the strongest fact lands by two seconds. Usually that fact is the pay. Numbers beat adjectives: a stated pay band says more than "competitive salary" in fewer characters.
3. Reuse what you already have
For every asset, ask what you can get again next month without asking anyone for a favor. The job posting gives you the words. Brand files give you the logo, colors and type. Approved photos of your team can go into an on-brand layout.
If a real person appears, get the release signed first, write down where the video may be reused, and decide now what happens when that person leaves. A standard photo release does not cover synthetic generation. If a colleague's likeness will drive a generated presenter, the release has to say so explicitly.
4. Generate what is missing
When nobody can be filmed, three routes work: approved brand assets only, with no people in frame; existing approved team photography in an on-brand layout; or a generated stand-in, and you say so.
Generating from the job posting is what our product does, so read this as first-hand rather than neutral. You paste a job URL, the system pulls the title, requirements, perks, logo, fonts and brand colors into a locked template, you choose how the visuals are made, and you export the cuts. It is not a substitute for a day on site when what you need is footage of your own people. If the brief calls for a specific human moment, film it.
5. Label it
If a generated person looks real, say so on the video itself, from the first frame. For ads seen in the EU, Article 50(4) of the EU AI Act has required that disclosure since August 2, 2026, and photorealistic people are in scope even when they are invented. Our disclosure guide covers who the rule reaches, including US employers.
6. Cut it for the placement, then check it there
The placement decides the ratio: 9:16 for Reels, Stories and TikTok, 4:5 for the Facebook and Instagram feeds, 1:1 for LinkedIn and 16:9 for YouTube. The working rule is under fifteen seconds for feed placements. The eight product-made ads in our examples library measure 7.38 to 14.04 seconds.
Before you spend anything, drop the cut into the free ad mockup tool and look at it inside the platform chrome. Captions, buttons and progress bars sit over the edges of your frame.
Then measure the two numbers that decide whether it worked: qualified applications, and what each one cost.
Example: Dunhill
Dunhill Professional Search & Government Solutions, in Wilmington, North Carolina, recruits for security-cleared federal IT roles: Secret, Active Secret, TS/SCI. Its people hold clearances and stay off camera by design.
Dunhill's recruiters make the videos themselves, across about fifty open roles. They paste the URL of a live job posting, which already holds the role, clearance level, location and salary band, and pick one of several AI-generated professionals, each inside Dunhill's own branded template. They add a generated voice-over and music, then export a 1:1 square, fourteen seconds long.
Only the person is generated. Everything else comes from Dunhill: the job posting details and the branded template we built for them. Our customer story says plainly that the people on screen are AI-generated stand-ins, not Dunhill staff.
The videos go to a LinkedIn following of more than 150,000, and each one goes out three times: from the recruiter, from Dunhill and from VP Rob Stewart. No paid promotion.
In the first seven days, in February 2026, Dunhill's LinkedIn post impressions rose 158%. Individual video posts reached up to 89% engagement, where static posts got 3% or less. Those are first-week figures, not a sustained average or a benchmark to plan against. Dunhill was still making videos six months later.
"The engagement rate is crazy good," Rob Stewart said.
Sources
https://recruitmentads.com/recruitment-video/how-to
https://recruitmentads.com/customers/dunhill
https://recruitmentads.com/free-tools/ad-mockup
https://recruitmentads.com/resources/guides/ai-generated-rec...
https://recruitmentads.com/resources/employee-photo-release-...
This article is part of the RecruitmentAds.com series on the EB EXP blog. RecruitmentAds.com sponsored EB EXP Fall 2026 — watch their session, "Recruitment Videos at Scale, Without Picking Up a Camera," with Niels Geominy and Andrew Hall.