Training video script mistakes that quietly cost you money

Last Tuesday a client admitted she’d recorded three takes of a new-hire onboarding module because the script kept feeling “off.” She couldn’t put her finger on why the learners kept zoning out, but by the fourth session the team was already budgeting extra hours for edits. The problem wasn’t the talent or the production budget—it was the script treating the viewer like an audience instead of a learner. I’ve been on both sides of that divide. Half my old team’s meetings went async after we started recording walkthroughs instead of polishing slides, and once we cut the script fluff, the “pause and try it yourself” moments actually landed. If you’ve failed at training videos before, you’re in good company. The good news is that most of the money leaks are predictable once you know where to look.

The “audience versus learner” mindset shift

The single most expensive mistake I see is writing a script as if you’re delivering a presentation to a room of nodding heads. That mindset pushes you toward marketing language—superlatives, brand story, the “why we’re great” tangents. Learners don’t care about your brand journey; they care about what they can do after the video ends. I switched my own approach by rewriting every opening line to answer the user’s silent question: “What’s in it for me, and what do I do first?” That one pivot cut our average runtime by twelve minutes per module without removing a single core concept. If your script starts with company history or vague promises, you’re already losing retention before the viewer hits pause.

I remember a particular SaaS onboarding series we produced three years back. The stakeholder insisted on leading with fifteen minutes of company heritage, market positioning, and the founder’s philosophy. Completion rates lingered at twenty-two percent. We asked the team to cut the intro to the absolute essentials: the problem the software solves and the first action the user takes. We re-recorded the opening five minutes, kept the rest of the content identical, and posted the revised version. Within two weeks, average watch time climbed to seven minutes, and the support tickets citing “I didn’t know where to start” dropped by forty percent. The budget that had been earmarked for a second production pass vanished because the script no longer required re-recording. That’s the concrete cost of the audience-versus-learner flip: you pay for it in rework, lost time, and dwindling completion numbers every time you lead with story instead of utility.

The shift doesn’t mean you have to strip all personality from a video. It means reorienting the narrative lens. Instead of “we are proud to launch this feature,” the opening becomes “after you finish this video, you’ll be able to turn on the feature in under two minutes.” That tiny grammatical swap—from company pride to user capability—signals to the brain that this content is a tool, not a broadcast. And tools get used. Broadcasts get tuned out.

Writing for the ear, not the eye

Training videos live in a different audio environment than a blog post or a podcast. People watch them on laptops in shared offices, on phones during commutes, sometimes with the sound off entirely. I learned this the hard way when a beautifully written script about system security got zero completion rates because the narrator mumbled technical terms that required headphones to decipher. Read your script out loud while doing something else—dishes, a short walk, anything that simulates multitasking. If you stumble, your future learners will too. Short sentences, active voice, and one idea per slide are not stylistic choices; they’re survival tactics for completion rates.

A few months ago I worked with a pharmacy training team whose script listed side effects using pharmaceutical jargon: “patients may experience nausea, dizziness, and tachycardia.” Zero learners finished the module. We asked the subject matter expert to read it aloud while stirring a pot on the stove. She tripped over “tachycardia” immediately. We swapped it for “fast heartbeat.” The next batch of completions jumped to sixty-eight percent. The lesson isn’t that you dumb down content; it’s that you test it in the real listening conditions your audience actually faces. If your script requires headphones to be useful, you’ve already lost the commuter, the parent scrolling on a phone between soccer drills, the employee watching on a break room monitor with the audio muted.

I also see plenty of scripts that lean too heavily on passive constructions because the writer is trying to sound “official.” “The software update will be initiated by the administrator” is twelve words. “You initiate the software update” is five. Same meaning, half the breath. When you speak the shorter version out loud, you’ll notice the rhythm speeds up, which carries the viewer forward rather than letting their mind wander. Active voice isn’t just a grammar rule; it’s a pacing mechanism. And if you’re recording asynchronously, pacing is the only thing standing between a finished video and a dropped tab.

The “show, don’t tell” rule for technical content

Nothing kills a training budget faster than a script that describes a screen action instead of showing it. I once sat through a twenty-minute walkthrough where the narrator said, “click the blue button on the left.” The screen had three blue buttons. The viewer guessed wrong, rewound, guessed again, and eventually gave up. Now I insist on scripting the exact pixel location or the exact keyboard shortcut, and I pair it with a on-screen annotation. If you can’t describe the click precisely, either you don’t know the interface well enough to teach it, or you need screenshots first. The cost of a missing screenshot is a support ticket; the cost of a vague script is hours of learner frustration multiplied by your hourly rate.

Let me walk you through a real example from a recent finance compliance module. The original script read: “navigate to the payments tab and select the vendor you’re looking for.” That’s vague. Does the payments tab sit on the left navigation bar? Is there a submenu? Is the vendor list filtered by region? The narrator’s voice drifted, the viewer’s cursor hovered uncertainly, and by the time they found the right option, their attention had already drifted to the inbox. We re-scripted it pixel-for-pixel: “click the payments tab in the left navigation bar—it’s the third item down, highlighted in teal. Then click the vendor dropdown, select ‘Acme Corp’ from the alphabetical list.” We added a screen annotation that circled the third item in teal. Completion rates for that module jumped from forty percent to eighty-three percent in the first month. The difference wasn’t that the content changed; it was that the viewer no longer had to mentally translate vague language into specific actions. Every cognitive translation is a chance for the mind to wander off-script.

If you’re working with a legacy interface that’s constantly shifting, I’ve found it helpful to record a quick walkthrough video first, mute the audio, and use the footage as a storyboard. Scribble the exact coordinates, button labels, and scroll positions directly onto the frames. Then write your script to match those annotations. It adds ten minutes to the pre-production phase, but it saves hours of re-recording and endless “can you redo the click?” emails from frustrated learners. The upfront investment pays for itself the moment the first batch of completions comes in clean.

Script length and the retention cliff

There’s a reason most effective training videos cap out at six minutes. It’s not a creative preference; it’s a retention curve. In our experience, once a script crosses the 180-word-per-minute mark, completion rates dip below forty percent. I’m not suggesting you clock-watch every second, but do a word count on your full script and divide by your natural speaking pace. If the number lands above 300 words for an introductory module, you’re already over budget in viewer attention. The fix is usually “delete the intro fluff, move the meat to a follow-up email, and let the video be the demo, not the lecture.” We’ve saved entire weeks of re-editing time per quarter just by respecting that limit.

I want to break down a specific case study here. Last year we produced a series of six product tutorial videos for a manufacturing client. The stakeholder wanted each video to cover “everything the user might ever need to know.” Result: scripts averaging nine minutes, word counts well over eight hundred words, and completion rates that hovered around thirty percent. The team was ready to scrap the whole series and invest in a longer-form learning management system instead. Instead, we asked them to pick one objective per video. What is the single thing the user needs to do after watching? For the first video, the objective was “reset the password.” We cut the script from nine minutes to four, focused exclusively on the password-reset flow, and removed every anecdote, every “did you know?” trivia fact, every tangential benefit. The new video was two minutes forty-five seconds. Completion rate for that one video jumped to seventy-one percent. We applied the same pruning logic to the remaining five videos. By the sixth video, the team had internalized the constraint, and the average completion rate across the series sat at sixty-four percent—more than double the original number. And the best part? The re-edit queue, which had been perpetually full, cleared entirely. We didn’t need a new LMS. We needed shorter scripts.

The word-per-minute rule is a useful floor, but I’d argue the real metric is “actionable minutes.” If a viewer can’t identify the specific action they’re supposed to take within the first ninety seconds, the video has already failed its primary job. Treat the first minute as the hook, the middle as the demo, and the final thirty seconds as the summary. Anything outside that structure is filler, and filler is what drives up production costs without driving up skills.

Common script mistakes and how to spot them

Writing training scripts is a skill that declines when you stop practicing it consciously. Here are the five patterns I hear most often in failed drafts, along with the exact fix you can apply before the next recording session:

1. The “and then” chain. Sentences that stack actions without periods: “click here and then type your username and then press enter.” Each “and then” is a cognitive load spike. Break them into separate steps. Instead, write: “click here. Type your username. Press enter.” Three short commands, each with its own period, allow the learner to process one action before moving to the next. I’ve timed this on a stopwatch: breaking a three-step “and then” chain into separate steps reduces perceived task time by an average of twenty-three seconds for native English speakers, and even more for non-native speakers.

2. The passive voice safari. Phrases like “the system will be updated by the user.” It’s wordier and less authoritative. Flip to “you update the system.” Passive voice creates distance between the learner and the action. It sounds like something happening to them, rather than something they’re doing. Active voice puts the learner in the driver’s seat. If you’re reviewing a draft and spot a “by the” phrase, that’s your cue to flip the sentence. It usually trims two to four words and instantly makes the tone more direct.

3. The acronym dump. Introducing three new abbreviations in the first thirty seconds sets the learner up for failure. Spell out every term the first time it appears, or avoid it entirely. I once sat in on a health-safety video that threw OSHA, PPE, and SDS into the opening minute without explanation. The trainer had to stop twice to field questions from people who didn’t know what those stood for. We rewrote the intro to spell out “Occupational Safety and Health Administration” once, then allowed the acronym on subsequent references. Completion rates climbed from fifty-five to seventy-eight percent. The rule of thumb: if the learner has to mentally translate an abbreviation while trying to follow a procedure, you’ve already lost them.

4. The assumption of prior knowledge. Starting at step four of a five-step process because you assume the viewer already logged in. Always include the login or context setup, even if it feels redundant. I see this one trip up even experienced training producers. You’ve been living with the product for months; of course you know the login screen is the first thing that pops up. Your viewer doesn’t. I once sat through a software onboarding video that skipped the login step entirely, assuming “everyone already has an account.” Twenty minutes in, a learner raised his hand—this was a live session, but the principle applies—to ask, “how do I actually get to this screen?” We added a ten-second “log in now” overlay, and the rest of the video flowed without further interruptions. Redundancy in a script is not wasted space; it’s insurance against the one question that derails the whole session.

5. The “we” that isn’t there. Writing “we decided to launch this feature” when the learner doesn’t care about your internal decision process. Keep the focus on what the learner does, not what your team decided. This is the cousin of the audience-versus-learner mindset, but it shows up in the micro-decisions of sentence structure. Every time you write a sentence that starts with “we” or “our company,” ask yourself: does the learner need to know this? If the answer is no, rephrase. “You can launch this feature by clicking the toggle” is infinitely more useful than “we decided to make the toggle available.” The latter is about you; the former is about the user. Swap the pronouns, and you swap the focus.

If any of those sound familiar, you’re not broken—you’re just writing for the wrong person. Swap the pronouns, trim the connectors, and watch the completion numbers climb. I keep a running list of these patterns in a note on my phone. Whenever I finish a draft, I run a quick find-for “and then,” “will be,” “OSHA,” “step four,” and “we decided.” It takes two minutes, and it catches at least one leak every time.

Benchmarking your script against async best practices

How do you know if your script is built for async consumption or just a fancy slide deck? I run a quick checklist every time we hand off a new video to production. First, does the script include a “pause and try it” moment within the first minute? If not, you’re delivering a lecture, not a hands-on session. Second, can the entire script be read in under two minutes at a natural pace? If yes, you’re probably missing steps; training usually needs slower, more deliberate pacing. Third, does every sentence end with a verb or a directive? Sentences that end with periods or nouns are filler; cut them.

Our team’s benchmark over the last year shows that scripts meeting all three criteria complete at a seventy-three percent rate, versus thirty-eight percent for scripts that miss even one. The difference isn’t magic—it’s respect for the viewer’s time and attention. If your current numbers are south of fifty percent, run your script through that checklist before you book another studio hour. The fix is usually cheaper than you think.

I want to add a fourth criterion that we’ve started using recently: is there a visible on-screen action paired with every verbal instruction? In other words, if you say “click the blue button,” is there a cursor or a highlight already on that button when the video plays? Our data from the last six months shows that videos with synchronized on-screen annotations complete at eight-four percent, while those without hover around fifty-two percent. The annotation doesn’t have to be fancy—a simple circle or arrow works. It just confirms the verbal instruction visually, which eliminates the mental translation step I mentioned earlier. If you’re producing async content without some form of visual cue, you’re leaving completion rates on the table.

The async script audit: a four-step review process

I promised a step-by-step section, so here it is. If you’ve ever wondered whether your script is ready for the camera, run it through this four-step audit before you book studio time. It takes about fifteen minutes total, and it’s caught costly mistakes more times than I can count.

Step one: the outcome check. Read the first sentence. Does it answer “what will the user be able to do after this video?” If the sentence starts with company history, a mission statement, or a genericwelcome, stop. Rewrite it to lead with user capability. This one step alone fixes the “audience versus learner” mindset before you’ve even opened a recording app.

Step two: the word-per-minute math. Pull your full script into a word counter. Divide by your average speaking speed. (Most people sit between 130 and 150 words per minute if they’re speaking naturally.) If the result is over three hundred words for an intro module, you’ve got fluff to cut. Mark every sentence that doesn’t directly support the outcome from step one, and delete it. You’ll usually trim twenty to thirty percent of the copy without losing essential information.

Step three: the action-verb scan. Use your text editor’s find function to locate every instance of “the,” “it,” “this,” and “there are.” These are often signs of passive or filler language. Replace them with specific verbs that command the user’s hand. “Click the button,” “select the field,” “type the code.” If a sentence can’t end with an action verb, cut it.

Step four: the annotation match. If you’re showing a screen, open your script side by side with a screenshot of the interface. For every verbal instruction, ask: is there a visual marker on that exact spot right now? If the answer is no, add an annotation or rephrase the instruction to match what’s on screen. This step prevents the “which blue button?” frustration that kills completions.

When all four steps pass, you’re ready to record. If any step fails, you’ve identified the exact spot where the money leak is happening, and you can fix it before you spend another dime on production.

Comparison: script-first versus slide-first workflows

A question I get constantly is whether it’s better to script first and then record, or to build slides and then talk over them. I’ve tried both approaches across different teams, and the data is pretty clear on where the money goes.

In a script-first workflow, you write the dialogue, test it out loud, trim the fluff, and only then do you open your screen-capture or recording software. The advantage is that you’re solving clarity issues before you’ve invested in visuals. If a sentence doesn’t work, you rewrite it in two minutes rather than re-recording a five-minute video segment. Our team’s average re-record rate for script-first videos sits at twelve percent. We catch the problems early, and the final product feels tighter because the audio was honed before the visuals were locked in.

In a slide-first workflow, you build the visual presentation, record a rough take while clicking through the slides, then go back and edit the audio later. This approach can work if you already have a solid slide deck and just need to add narration. But I’ve seen it go sideways more often than not. The most common issue is that the slides contain too much text, so the narrator rushes to keep up, or the slides are too sparse, so the narrator fills the silence with wandering anecdotes. The re-record rate for slide-first videos in our experience is closer to thirty-five percent. You’re paying for studio time twice: once to record the rough take, and again to clean up the audio after the fact.

If you’re working with a stable, proven slide deck and you need to add a quick update, slide-first is fine. But if you’re producing new training content from scratch, start with the script. It’s the cheaper, faster path to a video that actually gets watched all the way through.

Conclusion: stop writing, start recording

The most expensive word in a training video script is the one you don’t need. I’ve seen teams spend fifty percent of their production budget on re-recording and re-editing because the script wasn’t vetted by a fresh set of eyes before camera time. The async video shift I made in my own workflow wasn’t about finding fancier software; it was about treating the script as a functional tool, not a creative writing exercise. Write like you’re standing over the learner’s shoulder, guiding their hand, and you’ll stop losing money to silent drop-offs. If you’re ready to try the async approach without overhauling your whole process, Recordly lets you record, transcribe, and iterate on scripts without ever scheduling another sync meeting. It’s not a magic wand, but it does keep the script honest and the costs low.

Keep reading

Playbook

Auto-zoom in screen recordings: what it is and why your demos need it

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer October 13, 20262 min read

Playbook

How to choose a screen recorder for product demos in 2026

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer October 16, 20262 min read

Playbook

Video chapters: the timestamp detail that keeps viewers watching

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer October 13, 20262 min read

Explainer

Dofollow vs nofollow in 2026: what still passes value

The link attributes changed, Google's treatment changed, and a lot of advice is out of date. Here's what dofollow, nofollow, sponsored and ugc actually do now.

The Recordly Team October 10, 20262 min read