A teleprompter for nonprofit fundraising — appeals that do not sound read

An appeal video runs on one currency: sincerity. Donors do not give to organisations, they give to a person who visibly means it — and eyes tracking lines of text read as evasion at the exact second you ask for money. A fundraising ask that looks read converts like a cold call. Your own comment sections already say it plainly: people give when they feel spoken to, and scroll on when they feel presented at.

Yet the wording cannot be improvised. The three-pounds-a-month claim was signed off; the beneficiary’s story is told within the consent that was actually given, and not a sentence further; the phrasing is what keeps you right with the fundraising regulator. Sign-off means saying it as written — approximately right is a compliance problem, not a style choice.

And your presenters are not presenters. The person who should front the appeal is the warehouse coordinator or the youth worker, because they are the ones who mean it — and they are also the ones who go rigid at a lens, apologise after every take, and quietly hope never to be asked again. Every retake spends the goodwill of somebody who volunteered a lunch hour.

Distribution adds its own demands. Most of the audience meets the appeal muted, mid-scroll, in a feed; supporters with hearing loss are part of the community you serve, not an edge case. Captions are not optional. And none of this has a budget line — every pound that goes to production is a pound that visibly did not go to the programme.

The flagship campaign gets an agency shoot once a year. It is good, and it costs roughly what a part-time coordinator does — which is exactly why it cannot be the monthly rhythm, and why everything between campaigns is a graphic with a donate link. The agency footage also ages badly: the warehouse has moved, the branding changed, and the monthly amount in the voiceover went up a year ago.

In-house attempts mean a printed script taped beside the phone, with the eye-line miss visible in every frame, or memorising — which quietly rewrites the approved wording one sentence at a time until the ask on film is not the ask legal saw. The other pattern is the founder fronting every video because nobody else will — which works until the founder is on leave and the channel goes dark for a month. A format only one confident person can operate is a single point of failure dressed up as a workflow.

Captions get typed by hand into a platform’s caption editor at the end of the month, or skipped “just this once”. Just-this-once becomes the norm by the third video.

Talk2Camera puts the approved script under the lens of the phone that is filming and scrolls it as the presenter speaks. Voice follow waits when they steady themselves — the pause before a hard sentence stays theirs — and catches up if they skip. The words on screen are the words that were signed off, spoken as written, with the presenter looking where the donor is. And if English is the presenter’s second language, the pace adapts to them rather than the other way round — voice follow has no set speed to fall behind.

For the volunteer who dreads seeing themselves, cover mode turns the prompter up until the script covers the camera entirely and they show through from behind the words. They are reading text on a screen, not watching their own face — and that one change settles most first-timers inside a take.

After the take, the eye-contact report replaces the feeling with a number: how much of the recording was spent looking at the lens, and the longest look away. When no face is found it reports zero rather than estimating. “Good enough” becomes a decision you can make on take two, instead of a doubt that produces take eight from a volunteer who wanted to help, once.

Captions can be burned into the picture or exported as a separate .srt file — the muted feed gets subtitles, and the accessibility workflow gets the file it actually wants. All of it happens on the phones the team already owns; on the free plan, exports carry a small Talk2Camera mark, which is an honest trade when the alternative was a production invoice. And because the .srt is plain text, the same file doubles as the transcript for the website’s accessibility page — typed once, used twice.

A food-bank charity records its monthly update on a Tuesday, beside the pallets. The warehouse coordinator presents; the manager approved the script on Monday, three-pound ask and all. Cover mode is on, because she hates watching herself — the shelves show through behind the scrolling words.

Take one, a forklift interrupts; she stops, the script waits, she carries on. Take two is clean. The eye-contact report says nearly the whole take was spent looking at the lens, longest look away under a second — good enough, decided in thirty seconds, and the second take is the one posted rather than the eighth. Nobody stayed late, and nobody has to be talked into doing it again next month — which is the real test of a monthly format.

The video goes out with the .srt uploaded beside it: watchable muted, accessible to the supporters who need it, and the approved wording intact from sign-off to screen. Total production cost: nothing that was not already in the room.